EPA 747-R-96-007
May 1997
LEAD EXPOSURE ASSOCIATED WITH
RENOVATION AND REMODELING ACTIVITIES:
ENVIRONMENTAL FIELD SAMPLING STUDY
VOLUME I: TECHNICAL REPORT
Prepared By
Battelle
505 King Avenue
Columbus, OH 43201
for
Technical Programs Branch
Chemical Management Division
Office of Pollution Prevention and Toxics
Office of Prevention, Pesticides, and Toxic Substances
U.S. Environmental Protection Agency
Washington, DC 20460
-------
DISCLAIMER
Mention of trade names, products, or services does not
convey, and should not be interpreted as conveying official EPA
approval, endorsement, or recommendations.
This report is copied on recycled paper.
-------
Contributing Organizations
This study was funded and managed by the U.S. Environmental Protection Agency. The
study was conducted collaboratively by two organizations under contract to the Environmental
Protection Agency, Battelle, and Midwest Research Institute. Each organization's responsibilities
are listed below.
Battelle
Battelle was responsible for designing the study, recruiting participants, creating and
maintaining the study data, conducting the statistical analysis, and producing the final report.
Midwest Research Institute (MRI)
MRI was responsible for field data collection, chemical analysis, and reporting of chemical
analysis results.
U.S. Environmental Protection Agency
The Environmental Protection Agency was responsible for oversight in the developing the
study plan, managing and coordinating the overall study, and reviewing and editing this report.
EPA Project Managers included Dan Reinhart, Darlene Watford, Susan Dillman, and Betsy
Dutrow. Cindy Stroup was the Branch Chief of the Technical Programs Branch under whose
direction the study was conducted.
-------
TABLE OF CONTENTS
Page
EXECUTIVE SUMMARY xiii
1.0 INTRODUCTION TO THE OVERALL RENOVATION AND REMODELING
(R&R) STUDY 1-1
1.1 Objectives of the R&R Study 1-1
1.2 Scope of the R&R Study 1-2
1.2.1 Definition of Renovation and Remodeling 1-2
1.2.2 Components of the Scope 1-2
1.2.3 Specification of R&R Target Activities 1-3
1.3 Approach 1-3
1.3.1 Blood-Lead Measurements Versus Environmental Measurements 1-4
1.3.2 Data Collection 1-4
1.4 Peer Review 1-7
2.0 LITERATURE SEARCH AND INFORMATION GATHERING 2-1
2.1 Relevance of Information on Abatement 2-1
2.2 Characterization of Renovation and Remodeling 2-2
2.3 Review of the Available Literature 2-3
2.4 Review of Available Data Sources 2-5
2.5 Summary 2-7
3.0 QUALITY ASSURANCE 3-1
4.0 STUDY DESIGN FOR THE ENVIRONMENTAL FIELD SAMPLING STUDY (EFSS) 4-1
4.1 Objectives of the EFSS 4-1
4.2 Environmental Sampling 4-2
4.3 Evaluation of Sampling Methodology 4-3
4.4 Human Subjects Review 4-4
4.5 Data Gaps and Limitations of the R&R Study 4-4
4.6 Sampling and Analytical Methods 4-5
4.6.1 Sampling Methods 4-6
4.6.2 Analytical Methods 4-11
5.0 OVERVIEW OF RECRUITMENT IN THE EFSS 5-1
5.1 Details and Results of the Recruitment Process 5-3
5.1.1 Carpet Removal 5-4
5.1.2 Window Replacement 5-5
5.1.3 Large Projects 5-5
5.1.4 CED Phase 5-6
5.2 General Characteristics of the Workers and Units Recruited 5-7
5.2.1 Worker Characteristics 5-7
5.2.2 Unit Characteristics 5-9
5.2.2.1 Comparability with Units from Previous Studies 5-10
6.0 DATA ANALYSIS OVERVIEW 6-1
6.1 Data Analysis Objectives 6-1
6.2 Approach to Data Analysis 6-2
6.2.1 Common Approaches Taken Within Each Study Phase 6-3
6.2.2 Estimating Lead Disturbance in a 6' x 1' Gradient Region 6-5
6.2.3 Significant Digits 6-7
iv
-------
TABLE OF CONTENTS
Page
7.0 METHODOLOGY ISSUES AND RESULTS 7-1
7.1 Stainless Steel Dustfall Collectors vs. Post- minus Pre-activity Floor Dust
Method 7-1
7.2 Stainless Steel Dustfall Collectors: Collecting Samples 1 Hour vs. 2 Hours
after R&R Activity 7-6
7.3 Collecting Samples via Vacuum Techniques vs. Wipe Techniques 7-8
7.4 Comparing Results Between Side-by-Side Samples 7-11
8.0 STUDY RESULTS FOR INDIVIDUAL PHASES 8-1
8A-1.0 STUDY DESIGN IN THE CARPET REMOVAL PHASE 8-2
8A-1.1 Sampling Design for the Carpet Removal Phase 8-2
8A-1.2 Unit Characteristics 8-6
8A-2.0 Study Results for the Carpet Removal Phase 8-7
8A-2.1 Characterize Lead Disturbance and Potential Lead Exposure 8-8
8A-2.1.1 Personal Worker Exposures 8-8
8A-2.1.2 Potential Occupant Exposures to Airborne Lead 8-9
8A-2.1.3 Lead Disturbance and Potential Occupant
Exposure to Lead In Dust 8-11
8A-2.1.3.1 Vacuum Dust Samples Taken Pre- and
Post-Activity from Floor Surfaces 8-11
8A-2.1.3.2 Vacuum Dust Samples from Stainless Steel
Dustfall Collectors 8-12
8A-2.1.3.3 Vacuum Dust Samples from Window Sill
Surfaces Taken Pre- and Post-Activity 8-13
8A-2.2 Assess Factors or Measurements Related to Lead Disturbance 8-13
8A-2.2.1 Personal Worker Exposures 8-14
8A-2.2.2 Potential Occupant Exposures 8-14
8A-2.2.3 Predicting Lead Disturbance and Potential
Occupant Exposures to Lead in Dust 8-1 5
8A-2.2.3.1 Vacuum Dust Samples Taken Post-
Activity from Floor Surfaces 8-15
8A-2.2.3.2 Vacuum Dust Samples from Stainless
Steel Dustfall Collectors 8-16
8A-2.2.3.3 Vacuum Dust Samples Taken Post-Activity
from Window Sill Surfaces 8-16
8A-2.2.4 Estimating Variance Components 8-16
8A-2.3 Evaluate Different Media and Methods as Indicators of Exposure 8-18
8A-2.4 Summary of Results 8-20
8B-1.0 STUDY DESIGN IN THE WINDOW REPLACEMENT PHASE 8-22
8B-1.1 Sampling Design for the Window Replacement Phase 8-22
8B-1.2 Unit Characteristics 8-26
8B-2.0 Study Results for the Window Replacement Phase 8-27
8B-2.1 Characterize Lead Disturbance and Potential Lead Exposure 8-27
8B-2.1.1 Personal Worker Exposures 8-27
8B-2.1.2 Potential Occupant Exposures to Airborne Lead 8-29
-------
TABLE OF CONTENTS
Page
8B-2.1.3 Lead Disturbance and Potential Occupant Exposure to
Lead in Dust 8-29
8B-2.1.3.1 Vacuum Dust Samples Taken Pre- and
Post-Activity from Floor Surfaces 8-30
8B-2.1.3.2 Vacuum Dust Samples from Stainless
Steel Dustfall Collectors 8-30
8B-2.2 Assess Factors or Measurements Related to Lead Disturbance 8-33
8B-2.2.1 Vacuum Dust Samples from Window Well Surfaces Taken
Pre-Activity 8-34
8B-2.2.2 Lead Disturbance as a Function of Pre-Activity
Lead Loadings 8-34
8B-2.2.3 Lead Disturbance as a Function of Distance from
the Windows 8-35
8B-2.2.4 Estimating Variance Components 8-39
8B-2.3 Correlations Between Lead in Different Sample Media 8-41
8B-2.4 Summary of Results 8-42
8C-1.0 Study Design in the CED Phase 8-45
8C-1.1 Sampling Design for the CED Phase 8-45
8C-1.1.1 CED Activities 8-46
8C-1.1.2 Sample Types and Locations 8-48
8C-2.0 STUDY RESULTS 8-52
8C-2.1 Structure of CED Data 8-53
8C-2.2 Characterizing Lead Disturbance and Potential Lead Exposures 8-55
8C-2.2.1 Personal Worker Exposures 8-55
8C-2.2.2 Lead Disturbance and Potential Occupant Lead Exposure 8-60
8C-2.2.3 Adjusting Estimates of Lead Disturbance for
the Amount of R&R Activity 8-65
8C-2.3 Factors or Measurements Related to Lead Disturbance 8-67
8C-2.3.1 Pre-activity Measures of Lead Contamination 8-67
8C-2.3.1.1 Results of Paint Chip Samples 8-67
8C-2.3.1.2 Results of Dust Wipe Samples from
HVAC Systems 8-70
8C-2.3.1.3 Results of Dust Vacuum Samples (Pre-
Activity in Denver) 8-70
8C-2.3.2 The Relationship Between Airborne Lead, Lead in Settled
Dust, and Paint Lead Levels 8-71
8C-2.4 the Potential for Cross Contamination Within R&R Field Studies 8-74
8C-2.5 Discussion of the Results of the CED Phase 8-76
8D-1.0 INTRODUCTION AND OBJECTIVES OF THE CLEANUP INVESTIGATION 8-78
8D-2.0 STUDY DESIGN IN THE CLEANUP INVESTIGATION 8-78
8D-2.1 Sampling Design in the Cleanup Investigation 8-79
8D-3.0 STUDY RESULTS FOR THE CLEANUP INVESTIGATION 8-80
8D-3.1 Lead Levels in Paint to Be Disturbed 8-82
8D-3.2 Pre- Versus Post-Cleanup Results 8-83
vi
-------
TABLE OF CONTENTS
Page
8D-3.3 Post-cleanup Results 8-86
8D-3.4 Next-Day SSDC Lead Loading Results 8-87
8D-3.5 Conclusions 8-87
8E-1.0 DESCRIPTION OF RELEVANT DATA LOCATED FROM OTHER SOURCES 8-89
8E-2.0 SUMMARY OF PERSONAL WORKER EXPOSURE DATA FROM OTHER SOURCES . . . 8-94
8E-2.1 Comparison of Personal Worker Exposure Data from Other Sources with
EFSS Results 8-94
8E-3.0 RESULTS OF THE ANALYSIS OF SURFACE PREPARATION DATA FROM
OTHER SOURCES 8-99
8E-3.1 Analysis to Estimate Average Personal Worker Exposure Levels
Across Studies for Surface Preparation 8-101
8E-3.2 Settled Dust Samples in the Massachusetts Department of Health Study
Associated with Surface Preparation 8-104
9.0 OVERALL RESULTS 9-1
9.1 Characterize Lead Disturbance and Potential Lead Exposures 9-1
9.1.1 Personal Worker Exposures 9-2
9.1.2 Conversion of Task Length Averages (TLAs) into 8-Hour
Time Weighted Averages (TWAs) 9-5
9.1.3 Limitations of the Data and Additional Information Required
for anExposure Assessment for Workers 9-8
9.1.4 Summary Statistics for Lead Disturbance and Potential
Exposure to Occupants 9-9
9.1.4.1 Adjustment to a Standard Unit of Activity 9-10
9.1.5 Estimated Effect of Clean-up on the Total Lead Disturbed 9-14
9.1.6 Limitations of the Data on Lead Disturbance and Additional
Information Required for an Exposure Assessment for
Occupants 9-14
9.2 Assessing Factors or Measurements Related to Lead Disturbance 9-15
9.3 Evaluating Different Media as Indicators of Exposure 9-16
10.0 CONCLUSIONS 10-1
1 1.0 REFERENCES 11-1
LIST OF APPENDICES
APPENDIX A SUPPORT TABLES AND FIGURES OF EFSS DATA A-1
APPENDIX B DATA PROCESSING AND OUTLIER DETECTION B-1
APPENDIX C STATISTICAL METHODS AND MODELS IN THE EFSS C-1
APPENDIX D QUALITY CONTROL D-1
VII
-------
TABLE OF CONTENTS
LIST OF TABLES
Page
Table 2-1. Potential Sources of Existing Data Associating Lead Exposure With a Particular
Renovation and Remodeling Activity 2-7
Table 5-1. Summary of Recruitment Efforts and Results in the EFSS 5-8
Table 5-2. Descriptive Statistics for Pre-Activity Lead Loadings on Floors, Window Sills,
and Window Wells in the EFSS, with Comparison to Levels Observed in
Previous Abatement Studies 5-11
Table 7-1. Percentage of Sample Pairs (Regular and Side-by-Side QC Samples) Where Result
for the First Sample Collected Was Different From the Result for the Second Sample
Collected (Vacuum Samples Only) 7-13
Table 8A-1. Numbers and Types of Environmental Samples (Regular and QC) Proposed
in the EFSS Carpet Removal Phase Within Each Study Unit 8-2
Table 8A-2. Dwelling Units Included in the EFSS Carpet Removal Phase 8-6
Table 8A-3. Estimates of Total Variability and its Estimable Components in
Log-Transformed Lead Measurements by Sample Type for the Carpet
Removal Phase 8-18
Table 8A-4. Pearson Correlations of Geometric Mean Lead Loadings Between Pairs of
Sample Types and Approaches in the Carpet Removal Phase 8-19
Table 8B-1. Numbers and Types of Environmental Samples (Regular and QC) Proposed
in the EFSS Window Replacement Phase Within Each Study Unit 8-23
Table 8B-2. Dwelling Units Included in the EFSS Window Replacement Phase 8-26
Table 8B-3. Estimates of the Average Amount of Lead Disturbed by Window
Replacement Activity in a 6' by 1' Rectangular Region Perpendicular to
the Window 8-32
Table 8B-4. Results of Tests for Significant Covariates in the Model Fitting to Lead Loading Data
in the Window Replacement Phase 8-35
Table 8B-5. Estimates of Total Variability and its Estimable Components in
Log-Transformed Lead Measurements By Sample Type in the Window
Replacement Phase 8-40
Table 8B-6. Pearson Correlation Coefficients of the Geometric Mean Lead Loadings
Between Pairs of Sample Types and Approaches in the Window
Replacement Phase 8-43
Table 8C-1. Sampling Design for Each CED Activity Within Each Study Unit 8-49
Table 8C-2. Number of Field Samples Collected and Available for Analysis in the
CED Phase 8-55
Table 8C-3. Task Length for Each Combination of Target Activity and Substrate in
the CED Phase 8-56
Table 8C-4. Model Estimates of the Geometric Mean and Standard Deviation of
Variance Components of Worker Personal Exposure to Airborne Lead (Task
Length Average) for Each CED Activity Based on a Variance Components
Model of log(PEMijkl) 8-58
Table 8C-5. Estimates of the 50th, 75th, and 95th Percentiles for Worker
Personal Exposures, Along With 95% Confidence Intervals, by Activity
in the CED Phase 8-58
Table 8C-6. Estimates of the Increase in Personal Air Exposures that are Attributable
to a Substrate Effect (Wood versus Plaster) in the CED Phase 8-60
VIM
-------
TABLE OF CONTENTS
Page
Table 8C-7. Estimates of Lead Disturbance for Each CED Activity Based on
the "Population" Random Effects Modeling of Settled Dust
Lead-Loading Results 8-63
Table 8C-8. Estimates of the Log Increase in Intercept and Slope for the Relationship
Between log (Dustijk|) and Distanceijk| that are Attributable to a Substrate
Effect (Wood versus Plaster) 8-64
Table 8C-9. Estimated Differences in the 6' x 1' Lead Loading Gradient Attributable to
a Substrate Effect (Wood verses Plaster) 8-65
Table 8C-10. Estimates of the Potential Occupant Lead Exposure for Each CED Activity
Relative to a Standard Unit of Activity 8-66
Table 8C-11. Descriptive Summaries of Paint Chip Lead Loadings (mg/cm2) from
Samples Collected Within Each Activity in the CED Phase 8-68
Table 8C-12. Descriptive Summaries of Paint Chip Lead Loadings (mg/cm2) by Substrate
in the CED Phase 8-68
Table 8C-13. Descriptive Statistics for HVAC Wipe Sample Lead Loadings Across
and Within Units in the CED Phase 8-70
Table 8C-14. Pre-activity Settled Dust Results (Lead Loading and Sample Concentration)
from Window Sills and Shelves in the CED Phase 8-71
Table 8C-15. Estimated Substrate Effect on Log-Transformed Personal Exposure
Concentrations (log(PEM)), Before and After Adjusting for Paint Lead
Loading, in the CED Phase 8-72
Table 8C-16. Estimated Substrate Effect on Settled Log-Transformed Dust (log(Dust)),
Before and After Adjusting for Paint Lead Loading, in the CED Phase 8-73
Table 8C-17. Correlations of PEM Lead Concentrations, Settled Dust Lead Loadings, and
Paint Lead Loadings, Within an Experimental Unit, in the CED Phase 8-74
Table 8C-18. Statistical Estimates, with Standard Errors, of Worker Personal Exposure
to Airborne Lead, and 6' x 1' Gradient Dustfall Lead Amounts for Each
CED Activity 8-76
Table 8D-1. Numbers and Types of Environmental Samples (Regular and QC) Proposed
in the EFSS Cleanup Investigation 8-82
Table 8D-2. Geometric Mean of Pre-Cleanup Lead Loadings, and the Average Ratio of Post-to
Pre-Cleanup Lead Loadings on the Same Floor Surface 8-84
Table 8D-3. Estimated Lead Loadings (//g/ft2, with Standard Errors) for Pre- and Post-Cleanup,
and Percent Reduction from Pre- to Post-Cleanup Periods, As Estimated from
Statistical Modeling Procedures 8-86
Table 8E-1. Summary of the Presence of Lead-Based Paint, Task Lengths, Personal
Exposure Sample Sizes, and Percent of Personal Exposure Samples
Above Detection Limit by Study and Target Activity 8-93
Table 8E-2. Summary of Personal Worker Exposure Levels as Measured by a Task
Length Average (//g/m3) for the Other Sources of Data and for Data
From the EFSS 8-95
Table 8E-3. Summary Statistics for Data Used in the Statistical Analysis of the
Surface Preparation Data (Dry Methods Only) 8-102
Table 8E-4. Model Estimates of the Geometric Mean, 95% Confidence Interval and
Standard Deviation of Variance Components of Worker Personal Exposures (Task
Length Average in //g/m3) for Interior and Exterior Dry Surface Preparation . . . 8-103
Table 8E-5. Summary Statistics for the Different Types of Settled Dust Samples
Collected in the Massachusetts DOH Study for Interior Dry and Wet
Surface Preparation 8-106
IX
-------
TABLE OF CONTENTS
Page
Table 9-1. Estimated Percentiles and Confidence Intervals of Personal Worker Exposures
Measured by a Task Length Average for R&R Activities and Tasks 9-3
Table 9-2. The Percent of Workers Whose TLA Personal Air Lead Concentration
Would Exceed the OSHA PEL (50//g/m3), as Estimated from the Observed
Distributions of TLAs in this Study 9-5
Table 9-3. Estimated Lead Amounts (//g) Distributed Within a 6' x 1' Gradient Region,
with Standard Errors and 95% Confidence Intervals, in the EFSS 9-12
LIST OF FIGURES
Figure 1-1. Overall Design Structure of the Renovation and Remodeling Study 1-6
Figure 4-1. Cyclone Dust Collector 4-7
Figure 4-2. Personal Air Monitoring Sampler (Left) and Ambient Air Monitoring
Sampler (Right) 4-9
Figure 6-1. Example of a Boxplot 6-5
Figure 6-2. Portrayal of Model Predicting Lead Exposure as a Function of Distance
from R&R Activity 6-6
Figure 7-1. Lead Loadings (//g/ft2) for 1-Hour Stainless Steel Dustfall Collector Samples
Versus the Difference in Loadings Between Post-Activity and Pre-Activity
Floor Dust Samples in the Carpet Removal Phase 7-2
Figure 7-2. Lead Loadings (//g/ft2) for 1-Hour Stainless Steel Dustfall Collector
Samples Versus the Difference in Loadings Between Post-Activity and Pre-
Activity Floor Samples Taken Zero Feet from the Windows in the Window
Replacement Phase 7-3
Figure 7-3. Lead Loadings (//g/ft2) for 1-Hour Stainless Steel Dustfall Collector Samples
Versus the Difference in Loadings Between Post-Activity and Pre-Activity
Floor Samples Taken Six Feet from the Windows in the Window
Replacement Phase 7-3
Figure 7-4. Lead Loadings (//g/ft2) for 1-Hour Stainless Steel Dustfall Collector Samples,
and the Difference in Loadings Between Post- and Pre-Activity Floor Dust
Samples in the Carpet Removal Phase 7-4
Figure 7-5a. Lead Loadings (//g/ft2) for 1-Hour Stainless Steel Dustfall Collector Samples,
and the Difference in Loadings Between Post- and Pre-Activity Floor
Samples, Collected at Zero Feet from the Windows in the Window
Replacement Phase 7-4
Figure 7-5b. Lead Loadings (//g/ft2) for 1-Hour Stainless Steel Dustfall Collector Samples, and
the Difference in Loadings Between Post- and Pre-Activity Floor Samples,
Collected at Six Feet from the Windows in the Window Replacement Phase .... 7-5
Figure 7-6. Difference in Loadings Between Two-Hour and One-Hour SSDC Samples
(Two-Hour Minus One-Hour) Versus the Difference in Elapsed Wait Times
Between the Two Samples, for Carpet Removal (c) and Window
Replacement (w) Phases 7-7
Figure 7-7. Two-Hour SSDC Lead Loadings Versus One-Hour SSDC Loadings for the
Carpet Removal (c) and Window Replacement (w) Phases 7-7
Figure 7-8. Plot of Lead Loadings for Paired Vacuum/Wipe Samples at Each Location
and Study Unit in the EFSS 7-9
Figure 7-9. Plot of Wipe Sample Loading Versus Vacuum Sample Loading for Paired Samples
in the EFSS 7-10
-------
TABLE OF CONTENTS
Page
Figure 7-10. Lead Loadings for Adjoining Samples Collected via Vacuum Techniques
(Sample Collected First Versus Sample Collected Second) 7-12
Figure 8A-1. Three Settled Dust Sampling Locations at Adjacent Areas Within Each Unit,
and the Types of Regular and QC Samples Collected Within Each Location, in the
Carpet Removal Phase 8-5
Figure 8A-2. Scatterplot of Task-Length Average Personal Exposure Lead
Concentrations (//g/m3 air) for Each Worker Within Each Study Unit in the
Carpet Removal Phase 8-10
Figure 8A-3. Boxplots of Lead Concentrations (//g/m3 air) for Personal Exposure and
Ambient Air Samples in the Carpet Removal Phase 8-10
Figure 8A-4. Scatterplot of Ambient Air Lead Concentrations (//g/m3 air) Within Each
Study Unit in the Carpet Removal Phase 8-11
Figure 8B-1. Settled Dust Sampling Locations at a Selected Window in the Window
Replacement Phase 8-24
Figure 8B-2. Task-Length Average Personal Air Lead Concentrations, and Ambient Air
Sample Lead Concentrations, Within Each Study Unit in the Window
Replacement Phase 8-28
Figure 8B-3. Lead Loadings for Post-Activity Versus Pre-Activity Floor Samples at 0, 3,
and 6 Feet from the Windows 8-31
Figure 8B-4. Geometric Mean Lead Loadings from Floor Surfaces as a Function of
Distance from the Window 8-31
Figure 8B-5a. Lead Loadings (//g/ft2) for Pre-Activity Floor Samples Versus Distance from
the Windows at Which the Samples Were Taken 8-37
Figure 8B-5b. Lead Loadings (//g/ft2) for Post-Activity Floor Samples Versus Distance
from the Windows at Which the Samples Were Taken 8-37
Figure 8B-5c. Lead Loadings (//g/ft2) for Differences Between Adjoining Post-Activity and
Pre-Activity Floor Samples Versus Distance from the Windows at Which
the Samples Were Taken 8-38
Figure 8B-6. Lead Loadings (//g/ft2) for One-Hour Post-Activity SSDC Samples Versus
Distance from the Windows at Which the Samples Were Taken 8-38
Figure 8C-1. Scatterplot of Personal Exposure Loadings by Task Type and Unit ID 8-57
Figure 8C-2. 75th Percentile and Associated 95% Confidence Interval for Personal
Exposure to Airborne Lead (//g/m3) from Each Combination of CED Activity
and Substrate 8-59
Figure 8C-3. Estimated Lead Fallout Gradient in Settled Dust from 0 to 6 Feet Away
from the Edge of the Activity Space 8-63
Figure 8C-4. Estimated Lead Fallout Gradient in Settled Dust, From 0 to 6 feet Away From the
Edge of the Activity Space - Adjusted to the "Standard Unit of Activity" 8-67
Figure 8C-5. Scatterplot of Paint Chip Loadings by Task/Unit in the CED Phase 8-69
Figure 8C-6. Scatterplot of Paint Chip Loadings by Substrate in the CED Phase 8-69
Figure 8C-7. Lead Loadings Within Wipe Dust Samples Collected from Inside
HVAC Ductwork 8-71
Figure 8D-1. Settled Dust Sampling Locations Specified by the Sampling Design for
the Cleanup Investigation 8-81
Figure 8E-1. Comparison of Geometric Mean Task-Length Average Personal Worker Exposures
(//g/m3) for the EFSS and Other Data Sources During Window Replacement . . . 8-96
Figure 8E-2. Comparison of Geometric Mean Task-Length Average Personal Worker Exposures
(//g/m3) for the EFSS and Other Data Sources During Demolition 8-96
XI
-------
TABLE OF CONTENTS
Page
Figure 8E-3. Comparison of Geometric Mean Task-Length Average Personal Worker
Lead Levels (//g/m3) for the EFSS and Other Data Sources During
Component Removal 8-97
Figure 8E-4. Comparison of Geometric Mean Task-Length Average Personal Worker
Lead Levels (//g/m3) for the EFSS and Other Data Sources During
Cleanup Activities 8-98
Figure 8E-5. Geometric Mean Task-Length Average Personal Worker Exposures (//g/m3) During
Cleanup and Dry Interior Surface Preparation for Various Studies Compared to
the Combination Activities of the NIOSH-OU Study 8-98
Figure 8E-6. Geometric Mean Task-Length Average Personal Worker Exposures (//g/m3)
for Other Data Sources During Dry Interior Surface Preparation 8-99
Figure 8E-7. Geometric Mean Task-Length Average Personal Worker Exposures (//g/m3)
for Other Data Sources During Wet Interior Surface Preparation 8-100
Figure 8E-8. Geometric Mean Task-Length Average Personal Worker Exposures (//g/m3)
for Other Data Sources During Dry Exterior Surface Preparation 8-100
Figure 8E-9. Geometric Mean Task-Length Average Personal Worker Exposures (//g/m3)
for Other Data Sources During Wet Exterior Surface Preparation 8-101
Figure 8E-10. Comparison of Results from Statistical Analysis of Geometric Mean Task
Length Average Personal Worker Exposures (//g/m3) for Exterior Surface
Preparation: Overall Mean Estimate and Individual Study Mean Exposures
with 95% Confidence Bounds 8-103
Figure 8E-11. Comparison of Results from Statistical Analysis of Geometric Mean Task
Length Average Personal Worker Exposures (//g/m3) for Interior Surface
Preparation: Overall Mean Estimate and Individual Study Mean Exposures
with 95% Confidence Bounds 8-104
Figure 9-1. Ranking of R&R Activities and Tasks Based on the Estimated 75th Percentile
of Worker Exposures Measured by a Task Length Average 9-4
Figure 9-2. Hypothetical 8-Hour Time-Weighted Average Worker Exposures for Each
Activity/Task and for Various Activity Durations 9-7
Figure 9-3. Estimated Distribution of Dust Lead in a 6' x 1' Gradient for Various
Target Activities and Tasks, Based on Total Amount of Activity Performed in
the EFSS 9-11
Figure 9-4. Estimated Distribution of Dust Lead in a 6' x 1' Gradient for Various Target
Activities and Tasks, Based on Performing the Standard Unit of Activity 9-11
Figure 9-5. Estimated 6' x 1' Gradient Lead Amounts as a Function of Increases in
the Standard Unit of Activity for Each Generic R&R Task 9-13
XII
-------
Executive Summary
The Residential Lead-Based Paint Hazard Reduction Act (Title X) required the U.S.
Environmental Protection Agency (EPA) to conduct a study of lead exposure resulting from
renovation and remodeling (R&R) activities (the R&R study). The information obtained from the
study is to be used primarily to help determine which groups of people require training,
certification, or educational materials because of the potential lead exposure associated with R&R
activities they perform. This report presents the results of one of the principal data collection
efforts in the R&R study: the Environmental Field Sampling Study (EFSS). The EFSS, through
the collection of environmental measurements, assessed the amount of disturbance and potential
exposure to lead that resulted from selected R&R activities. The monitored activities included
both specific R&R activities, such as carpet removal and window replacement, as well as
miscellaneous generic activities such as drilling, sawing, or surface preparation (sanding, paint
scraping, etc.). Environmental samples collected in the EFSS included over 90 personal air
samples taken within the breathing zone of R&R workers as they performed specific R&R
activities, and over 500 samples of dust that settled on building surfaces within a specified period
following completion of an activity.
Worker exposure was assessed using the airborne lead levels from each worker's
breathing zone, as measured by a task-length average (TLA) exposure. A worker's TLA
represents average airborne exposure for the worker during conduct of the activity. The average
TLAs were high during the conduct of many of the R&R activities, exceeding the OSHA
permissible exposure limit of 50 |ig/m3 for four of the R&R activities. Average TLAs were
greater than 100 |ig/m3 for paint removal, interior demolition, and sawing, and greater than 49
|ig/m3 for interior surface preparation and central heating system maintenance/repair. Exposures
resulting from drilling, carpet removal, window replacement, and exterior surface preparation
were considerably lower (below 20 |ig/m3). The TLA exposure for each activity (as estimated in
the EFSS) can be combined with worker profile information (available from outside sources) to
characterize worker exposure.
Potential exposure to building occupants was assessed using the dust samples collected
by vacuum techniques from stainless steel dustfall collectors placed at specified distances from the
activity. Lead loadings from these samples were measured as indicators of the amount of lead
disturbed by an R&R activity and available for exposure to occupants. With the exception of
carpet removal and drilling into plaster, all activities monitored in the EFSS deposited significant
amounts of lead, ranging from 328 |ig/ft2 for sawing lead-painted plaster to 42,900 |ig/ft2 for paint
removal. Paint removal, demolition, sawing, and disturbing central heating system ductwork were
more likely to cause airborne lead to scatter and settle over a widespread area, while window
replacement and drilling confined the disturbed lead to a smaller area. While simple broom and
shop-vacuum cleanup substantially reduced the total amount of lead available to occupants,
cleanup efficiency declined as the distance from the activity increased. In addition, the average
XIII
-------
amount of lead following cleanup often remained above 100 |ig/ft2, the current EPA guidance
level for floors. The estimates of lead amounts within settled dust presented in this report can be
linked with information on types and durations of activities, types of work practices and cleanup
activities, and human health effects to provide a more complete characterization of occupant
exposure associated with R&R activities.
XIV
-------
1.0 INTRODUCTION TO THE OVERALL RENOVATION
AND REMODELING (R&R) STUDY
On October 29, 1992, the United States Congress enacted the Residential Lead-Based Paint
Hazard Reduction Act (Title X of HR 5334). This includes Title IV of the Toxic Substances
Control Act that requires the U.S. Environmental Protection Agency (EPA) Administrator to
conduct a study of lead exposure associated with renovation and remodeling activities. In
particular, paragraph (2) of Section 402 (c) states:
The Administrator shall conduct a study of the extent to which persons
engaged in various types of renovation and remodeling activities in target housing,
public buildings constructed before 1978, and commercial buildings are exposed to
lead in the conduct of such activities or disturb lead and create a lead-based paint
hazard on a regular or occasional basis.
EPA conducted the above study, hereafter referred to as the Renovation and Remodeling
(R&R) study, from 1993 through 1995. Results of the R&R study are documented in three
separate reports:
• "Lead Exposure Associated With Renovation and Remodeling Activities: Summary
Report"
• "Lead Exposure Associated With Renovation and Remodeling Activities:
Environmental Field Sampling Study," a technical report on environmental
measurements of lead associated with renovation and remodeling; this report also
includes the results of the literature review and a summary of data collected from
other extant sources; and
• "Lead Exposure Associated With Renovation and Remodeling Activities: Worker
Characterization and Blood-Lead Study," a technical report on blood lead levels and
work practices of renovation and remodeling workers.
Chapters 1 and 2 of this report include a discussion of the overall design of the R&R study
and the complementary roles of its two principal data collection efforts: the Environmental Field
Sampling Study (EFSS, or Environmental Study) and the Worker Characterization and Blood-
Lead Study (WCBS, or Blood-Lead Study). Subsequent chapters deal with the design,
implementation, and results of the EFSS.
1.1 OBJECTIVES OF THE R&R STUDY
The overall or programmatic objective of the R&R study was to determine which groups
of people doing R&R work require training, certification, or educational materials because of their
potential lead exposure. In direct response to the scope of work outlined in the legislation, the
study was designed to satisfy two primary technical objectives:
1-1
-------
1. Determine the extent to which persons engaged in various types of R&R activities in
target housing (i.e., housing constructed prior to 1978), public buildings constructed
before 1978, and commercial buildings are exposed to lead.
2. Determine the extent to which persons engaged in various types of R&R activities
disturb lead and create a lead-based paint hazard, on a regular or occasional basis, to
building occupants or other exposed individuals.
1.2 SCOPE OF THE R&R STUDY
The broad scope of the study mandated by Title X made it an arduous task to design and
implement the R&R study. The study required multiple field studies and decisions concerning
priorities, focus, and representativeness. This section presents the final decisions on key
definitions and delineation of scope. Chapter 2 provides insight into how these decisions were
made.
1.2.1 Definition of Renovation and Remodeling
In accordance with the United States Census Bureau's C-50 report, "Expenditures for
Residential Upkeep and Improvement" (U.S. Census Bureau, 1987), remodeling is defined as any
construction-related work on an existing property intended to either maintain or improve the
property. In addition, the work must be on items permanently attached or firmly affixed to some
part of the house or property.
The major components of remodeling include:
• Maintenance (painting, papering, floor sanding, furnace cleaning or adjustment, etc.)
• Repairs (plumbing, heating, electrical work, etc.)
• Additions and alterations (adding a wing, room, porch, deck, shed, basement, fence,
driveway, etc.)
• Major replacements (bathroom, kitchen, roof, water pipes, central heating system,
siding, etc.).
Renovation is defined as work on an existing property intended to make a major
improvement in the property.
1.2.2 Components of the Scope
The scope of the study mandated by the Title X legislation called for an assessment of the
lead exposure for different categories of:
• Individuals (specifically R&R workers, building occupants, and other exposed
individuals)
1-2
-------
• Environments (specifically private housing constructed before 1978, public buildings
constructed before 1978, and commercial buildings)
• R&R activities.
Private housing constructed before 1978, known as target housing., represented one category in
which lead exposure was to be assessed, while public or commercial buildings were placed into
one of two additional categories, according to whether or not children regularly inhabited the
buildings. The types of activities considered in the R&R activities segment are discussed in the
following section.
1.2.3 Specification of R&R Target Activities
The EPA assembled a list of R&R activities associated with lead exposure. This list was
developed as a result of more than 200 contacts with other government agencies, lead poisoning
prevention experts, industry representatives, labor unions, and other concerned groups. At a final
summary meeting on April 16, 1993, in Washington, D.C., with a number of these contacted
individuals, the EPA obtained data and subsequently defined eleven categories of R&R activity
with potential for lead exposure that could be addressed by this study. These categories,
subsequently called target activities, were:
1. Paint removal
2. Surface preparation
3. Removal of large structures
4. Window replacement
5. Enclosure of exterior painted surfaces (i.e., siding)
6. Carpet or other floor covering removal
7. Wallpaper removal
8. HVAC (central heating system) repair or replacement including duct work
9. Repairs or additions resulting in isolated small surface disruptions
10. Exterior soil disruption
11. Major renovation projects involving multiple target activities.
There were several reasons for choosing HVAC repair over plumbing or electrical work as a
target activity. First, both plumbing and electrical work were believed to often only disturb small
areas of painted surfaces. Second, disturbance of lead in pipes, joints, and soldered connections
was considered out of the scope of the study, which was focused on lead-based paint hazards.
Finally, there was concern that furnace ductwork could be a significant reservoir for large
amounts of lead dust.
1.3 APPROACH
The initial phase of the R&R study involved an extensive literature review and
information-gathering process. This process uncovered the currently available information on
lead exposure related to R&R activities. It also helped in decision making concerning the focus of
1-3
-------
the study. The results of this process are presented in Chapter 2. The major conclusion of the
literature review and information gathering was that, with the exception of paint removal,
insufficient information was available for an exposure assessment of different categories of R&R
activities. As a result, new data collection was required to address study objectives. This section
discusses the approach to new data collection.
1.3.1 Blood-Lead Measurements Versus Environmental Measurements
The existence and extent of lead exposure created by R&R activities may be assessed
either by blood-lead or environmental-lead measurements. Once the need for new data collection
was identified, the R&R design team considered the advantages and disadvantages of each
approach. A study assessing blood-lead concentrations provides for direct measurement of an
internal (absorbed) dose of lead. However, blood-lead concentrations can be explained not only
by recent exposure but also by historical exposure and by many secondary factors (such as age,
nutrition, and smoking). A study assessing measurements of lead in the environment (dust, air, or
soil) makes a direct link between R&R activities and measurements of lead disturbance.
However, these measurements serve only as estimates of the amount of lead available for potential
inhalation or ingestion, therefore representing only a potential internal dose to humans.
An optimal study design would involve measuring worker and occupant blood-lead
concentrations and environmental-lead levels before, during, and after R&R activity. However,
measuring blood-lead concentrations before and after an activity was not feasible for ethical and
legal reasons. Measuring environmental lead levels before and after an activity was complicated
by serious recruitment and liability problems because of the desire to target typical R&R jobs in an
unregulated environment.
The approach taken for this study circumvented these problems by defining two principal
data collection efforts: one characterizing environmental lead disturbance resulting from R&R
activities and the other focusing on the effect of R&R activity on worker blood-lead
concentrations through a retrospective study. A third effort, currently in the design stages, will be
a retrospective study to evaluate the impact of the conduct of R&R activity on elevated blood-
lead concentrations in children. This effort will be conducted jointly with the University of
Wisconsin and the State of Wisconsin Public Health Department.
1.3.2 Data Collection
Environmental Field Sampling Study
The first principal data collection effort of the R&R study was the Environmental Field
Sampling Study (EFSS). In it, environmental measurements were taken to assess the relative
disturbance of and exposure to lead associated with selected R&R activities. Activity categories
that showed the greatest measured amounts of lead exposure were assumed to be the primary
contributors to any potential health effect.
1-4
-------
The focus of the EFSS was on monitoring specific R&R activities. An alternative
approach was considered, namely to monitor specific R&R worker groups as they performed
their normal duties with whatever mix of activities was encountered. The decision to focus on
activities rather than specific worker groups was supported by all representatives who attended
the April 16, 1993, study design meeting in Washington, D.C., described in Chapter 2. The
primary reasons for focusing on activities were that:
1. A focus on activities provided the best understanding of exactly what was causing the
lead exposure.
2. A focus on activities was the most efficient way to assess a wide variety of R&R
worker groups. Exposure estimates based on worker groups would be applicable only
to the monitored groups. Exposure estimates based on specific activities, on the other
hand, could be combined with worker profile information for any given worker group
to assess that group's exposure. Worker profile information includes information on
the types of activities workers conduct, the type of work practices and worker
protection they use, and the percent of time they work in buildings with lead-based
paint.
3. Exposure estimates based on R&R activities provide information useful in the
development of subsequent guidelines for the conduct of R&R.
The data collection effort for the EFSS included six of the 11 target activities: removal of large
structures (demolition), window replacement, carpet removal, HVAC repair or replacement,
surface preparation, and repairs with minimal surface disruption. Paint removal was excluded
because exposure associated with paint removal could be assessed from the literature. Exterior
siding, wallpaper removal, and exterior soil disruption were excluded by consensus, because the
study design team and the individuals consulted in the information-gathering phase considered
these target activities to be of secondary importance. It is possible that inferences about exterior
siding and exterior soil abatement may be made from professional judgment and comparison with
other activities. Wallpaper removal is an activity conducted primarily by painters who were
assessed based on other activities they perform — most notably surface preparation and paint
removal.
The EFSS was supplemented by an extensive search for extant data that could be used
either to fulfill data requirements for a specific activity or to confirm results obtained in the EFSS.
Worker Characterization and Blood-Lead Study
The second principal data collection effort of the R&R study was the WCBS. The WCBS
involved collecting questionnaire information and blood-lead measurements from R&R workers
to 1) characterize blood-lead concentrations in specific worker groups, 2) determine if specific
worker groups or specific R&R activities are associated with increases in blood-lead
concentrations, and 3) collect information to be used to develop worker profiles. The WCBS was
intended to obtain information independent from the EFSS that would provide a direct measure of
1-5
-------
effects on worker health and to validate the results of the EFSS. Target R&R activities examined
in the WCBS included removal of large structures (demolition), window replacement, carpet
removal, HVAC repair or replacement, and paint removal. Post-activity cleanup was also
considered.
Figure 1-1 shows the design and coordination of the R&R study. This report presents the
technical results of the EFSS (right-hand side of Figure 1-1). As the figure indicates at the
bottom, the results of the EFSS and WCBS are integrated in the Summary Report for the R&R
study.
1.4 PEER REVIEW
RENOVATION AND REMODELING STUDY
I
Delineation of Scope; Literature Review and Information Gathering
I
Worker Characterization and Blood-
Lead Study (Blood-Lead Study)
Environmental Field Sampling Study
(Environmental Study)
Identify worker groups
and recruit workers
Large R&R
Projects with
Multiple Activities
Collect blood samples and
questionnaire information
Characterize workers.
Develop relationship between
blood lead levels and activities
WCBS
(Blood-Lead Study)
Technical Report
Decision on Assessment Method
• Professional judgment
• Literature
• New field studies
— Controlled designed field study
— Monitoring field study
Use of other extant data sources
Develop relationships between
activities and measurements of
lead in the environment
EFSS
(Environmental Study)
Technical Report
R&R Summary Report
Figure 1-1. Overall Design Structure of the Renovation and Remodeling Study
1-6
-------
This report on the Environmental Field Sampling Study (EFSS) was reviewed
independently by members of a peer review panel. Comments which are important for
interpreting the study results or which resulted in important modifications to the report are
discussed below. All peer reviewers recommended publishing the report with minor revisions.
A primary concern for a number of reviewers was the limited sample sizes in the EFSS
study. A separate section which discusses data gaps and data limitations is included in the report.
In addition, the sample size consideration has been highlighted through characterization of the
EFSS independent monitoring jobs as case studies and through judicious use of language such as
"potential for disturbing significant amounts of lead during R&R activities" that do not overstate
conclusions that can be drawn from this data. Sample sizes are discussed and documented
throughout all reports.
Related to the sample size issue was concern over the number of statistical analyses
conducted and the exploratory nature of some analyses. Additional cautionary language was
added to the reports to warn of exploratory analyses, the effect of multiple statistical comparisons,
and the limitation of small sample size on the ability to detect statistical significance.
Another concern expressed by reviewers was the relevance of abatement data to an
exposure assessment for R&R workers, and similarly the effect of using trained abatement
workers for some of the work conducted in the EFSS. Several sections of the EFSS report
address these concerns directly. Section 2.1, Relevance of Information on Abatement, and
Section 2.2, Characterization of Renovation and Remodeling, discuss the differences between
abatement work and renovation and remodeling. Section 5.0, Overview of Recruitment,
addresses the criteria applied in the EFSS for accepting jobs and workers as representative of
typical R&R work. Professional abatement workers in the EFSS did wear respirators and follow
abatement personal hygiene procedures to protect themselves. However, they did not follow
standard abatement procedures, such as use of wet methods, and were instructed to perform the
tasks as they are typically conducted in an unregulated environment. For example, demolition
was conducted dry with hammers and crow bars, and sawing was conducted dry with a circular
saw and no HEPA attachment. Since no dust minimization procedures were used, the work was
considered representative of typical renovation and remodeling work. On the other hand,
available data sources from professional abatement work that did involve dust minimization were
not included in any data summaries.
Several comments related to clarification of the terms "surface preparation" and "paint
removal." Although there is certainly overlap between the two activities, there was general
concurrence among all parties consulted during the design of the study that it was important to
distinguish between the two activities. These terms are defined throughout the report to make as
clear as possible the exact type and duration of activity that took place.
Concern was also expressed over the inability to collect both blood-lead and environmental
lead measurements from the same group of workers and/or occupants. Human subjects review,
for both ethical and legal reasons, would not allow measuring blood-lead concentrations for
occupants (young children) before and after conduct of an activity that was suspected of causing a
1-7
-------
hazard. For workers, the difficulty in this study was recruiting typical R&R workers operating in
an unregulated environment. For this group of workers, employers were very reluctant to
participate even as the study was conducted. Contractors were concerned over lawsuits by
workers in the event that the study revealed a worker's blood-lead increased as a result of a
specific job they were assigned to. We had very few contractors participating in either phase of
the study. Employees participated in the WCBS largely because of either their own interest or the
interest and encouragement of their national and local union. Gaining access to work sites for
environmental and biological sampling would have required participation of the contractors,
homeowners, and workers. If such sampling was conducted under forced cooperation, then the
results may have been biased. If the study had focused on lead abatement workers this may not
have been a problem, but with a focus on typical R&R workers who were not, at the time of this
study, using worker protection practices, there were many problems recruiting contractors to
participate. In short, the difficulty in recruiting contractors was in getting at the population of
interest: unregulated R&R workers not specializing in lead abatement.
One reviewer requested more information to show that the QC data are consistent with the
statistical analysis applications and results. As a result of this comment, more documentation was
added to the reports, including references to appendices and quality assurance project plans.
EPA has established a public record for the peer review under administrative record
AR152, "Lead Exposure Associated with Renovation and Remodeling Activities Peer Review."
The record is available in the TSCA Nonconfidential Information Center, which is open from
noon to 4 PM Monday through Friday, except legal holidays. The TSCA Nonconfidential
Information Center is located in Room NE-B607, Northeast Mall, 401 M Street SW, Washington,
D.C.
1-8
-------
2.0 LITERATURE SEARCH AND INFORMATION GATHERING
The initial phase of the R&R study involved an extensive literature review and
information-gathering process. This phase, conducted in concert with the development of EPA
guidelines for R&R ("Reducing Lead Hazard When Remodeling Your Home"), had three primary
objectives:
1. Define and characterize R&R and its component activities.
2. Collect the available information concerning human lead exposure and environmental
lead hazards resulting from R&R activities.
3. Identify any data sources that could be used in this study's assessment of lead
exposure and hazards related to R&R.
The goal of the information-gathering process was to identify and thoroughly examine the
sources of information regarding lead exposure during R&R activities. This chapter presents the
methodology, findings, and the conclusions of the literature search and information gathering
phase.
2.1 RELEVANCE OF INFORMATION ON ABATEMENT
Early in the process of gathering information about R&R activities, it was decided to
exclude information developed during lead-based paint abatement. This exclusion was a difficult
decision. A number of R&R activities are conducted during abatement (e.g., paint removal and
window replacement), and comparable worker populations perform both abatement and the R&R
activities. Moreover, considerable information is available in the scientific literature on the human
lead exposure and environmental lead contamination resulting from implementation of various
abatement strategies. However, there were some important differences between R&R and
abatement that impact the relevance of abatement results.
The work practices associated with abatement are likely to be considerably different from
those associated with R&R activities. It is recommended that abatement workers wear protective
equipment (e.g., respirators) during the activities that disturb lead, and that the disturbed lead be
contained by spreading and taping into place polyurethane sheeting. The procedures employed in
abatement were developed to either minimize the lead dust generated (e.g., misting) or restrict its
spread outside the immediate area (e.g., via vacuum attachments). Abatement often includes
extensive cleanup procedures, such as wet-mopping the area with a detergent solution of 5%
trisodium phosphate (TSP), then vacuuming with a high-efficiency particle accumulator (HEPA)
vacuum. In addition, abatement workers, by definition, operate more frequently and
knowledgably in a lead-contaminated environment than do R&R workers. Both their training and
awareness of the hazards of lead are usually more advanced than that of workers performing
strictly R&R activities. Therefore, their blood-lead concentrations and the environmental lead
generated by their activities are not representative of general R&R workers and their activities in a
non-abatement environment.
2-1
-------
2.2 CHARACTERIZATION OF RENOVATION AND REMODELING
The first objective of the information gathering phase was to characterize R&R, thereby
defining the scope of the study and the activities to be examined. The questions to be addressed
included:
• What is renovation and remodeling?
• Who performs it and what procedures do they employ?
• Is there a recognized lead hazard from such activities?
• If so, what activities have been identified as producing a hazard?
To answer these questions, the available literature was examined and prominent individuals in the
field of lead exposure (both residential and occupational) and within the R&R industry were
consulted. The primary goal was to obtain data on lead exposure to workers and occupants (as
measured by body-lead burden) resulting from R&R activities. If such data were unavailable,
information on environmental lead contamination was sought.
A comprehensive list of individuals or groups currently involved in lead research and
policy making from national committees, major trade industries, published authors, federal and
state agencies, academia, and medical institutions was compiled. People on the list were
contacted by telephone or personal interview. Target activities and worker classifications
associated with R&R were drafted by sorting through the literature and the information and
perspectives offered in the many phone calls and interviews. After more than 200 interviews were
completed, a final summary meeting was held on April 16, 1993, in Washington, D.C. with a
number of the contacted individuals. From this meeting, a general concurrence on the scope of
R&R and its component target activities and worker classifications was reached. The resulting
characterization and definition of R&R target activities to be considered was presented in Section
1.2.3.
In addition to defining R&R and its component target activities and classifying the
workers who perform these activities, the actual work practices involved in R&R needed to be
identified. Many of the contacts from national trade unions and government agencies stressed the
need to focus on "typical" R&R practices. They also stressed their belief that abatement practices
would not be representative of typical R&R practices. Although no information was identified
that directly described actual R&R work practices, some information concerning guidelines for
performing R&R work was uncovered. The National Association of Home Builders (NAHB) has
produced guidelines entitled, "What Remodelers Need to Know and Do About Lead — An
Interim Guide" (NAHB, December 1992). These guidelines parallel the HUD Guidelines for
Evaluation and Control of Lead-Based Paint Hazards in Housing (U.S. Department of Housing
and Urban Development, July 1995). Information concerning typical work practices was targeted
for collection in the WCBS component of the R&R study.
Finally, the parties attending the Washington meeting prioritized the identified R&R target
activities according to the need for data collection efforts to assess the lead exposure or
2-2
-------
environmental lead hazard generated. As a result, the following activities (listed in no particular
order) were chosen for new data collection as part of this study:
Carpet removal
Window replacement
Removal of large structures (demolition)
HVAC work
Isolated surface disruptions.
2.3 REVIEW OF THE AVAILABLE LITERATURE
The second objective of the information gathering phase was to thoroughly explore and
understand the existing body of information concerning the relationship between renovation and
remodeling and lead exposure, and to identify any data that could be used to assess this
relationship. Beginning with the individuals asked to define R&R, a search was performed to
identify published reports, papers, data, or other individuals or organizations that would provide
information on lead exposure or hazards associated with R&R activities. This method of
information gathering was pursued until no previously unidentified sources were uncovered.
In addition to soliciting opinions and information from interested parties, an extensive
literature search was undertaken. The search covered all information published in the last 15
years pertaining to the lead exposure of occupants, R&R workers, activities, R&R methods within
an activity, R&R in industry, and R&R in the military. Using the on-line library search system,
DIALOG, journals available through MEDLINE, NTIS, Engineering Information, Enviroline,
Pollution Abstracts, Occupational Safety and Health, and other sources were examined for
information relating to lead exposure and renovation and remodeling. More than 500 potentially
relevant articles were identified.
From this search, it was confirmed that literature directly discussing lead hazards or
exposure created by R&R activities was limited. Only 12 articles focused on renovation and
remodeling. Moreover, there appeared to be no definitive lead exposure study examining R&R
activities. In fact, all references to R&R activities were either secondary or anecdotal. For
example, four articles described case studies of individuals suffering from elevated blood-lead
levels during renovation of their homes. Overall, however, these twelve articles did provide
evidence of lead elevations in both body burden (blood) and environmental media (dust, air, and
soil) resulting from R&R activities.
On the other hand, there was a significant body of literature discussing R&R-related
activities, such as lead abatement. Approximately 20 such articles and reports were identified,
providing detailed information about the association between disturbances to an existing lead
reservoir (e.g., lead-contaminated soil or lead-based paint) and environmental or body-burden
lead levels. As was indicated in Section 2.1, the activity performed and the actual work practice
in an abatement are likely to be different from that of R&R. In fact, an argument could be made
that exposures measured during an abatement activity represent either a best- or worst-case
scenario when applied to the same renovation and remodeling activity. They could represent a
2-3
-------
best-case scenario because of the extensive precautions taken during abatement work. They
could represent a worst-case scenario because of the amount of lead disturbed during an
abatement. Therefore, only to the degree that an abatement activity is considered "similar" to a
typical R&R activity would the abatement literature prove useful in evaluating the lead hazard
associated with an R&R activity.
Despite the scarcity of directly relevant literature, there did appear to be a consensus that
R&R activities can produce both body-lead burden and environmental lead contamination. In a
longitudinal study of blood-lead levels in neonates, Rabinowitz and Needleman note that "infants
residing where lead paint is being resurfaced (as typically performed) may be at special risk of
increased lead exposure" (Rabinowitz et al, 1985). The Center for Disease Control and
Prevention's (CDC's) publication, "Preventing Lead Poisoning in Young Children," reports that
"many cases of childhood lead poisoning that result from renovation and remodeling of homes
have been reported" (Centers for Disease Control and Prevention, 1991). The Department of
Health and Human Services' (HHS') "Strategic Plan for the Elimination of Childhood Lead
Poisoning," cites as priorities for lead abatement "homes with lead-based paint that are being
renovated and remodeled for other reasons" (U.S. Dept. of HHS, February 1991). Dr. Julian
Chisolm, a respected doctor involved in childhood lead research, and other contacts, expressed
their personal opinion that R&R activities are responsible for elevated blood-lead levels in many
middle-class children. Renovation and remodeling activities, by their very nature, have the
potential to disturb existing reservoirs of lead (e.g., intact lead-based paint, lead dust-
contaminated carpet). Once the disruption occurs, the lead is available for contamination of
workers and residents. This fact, along with the acknowledged link between environmental lead
levels (especially dust-lead levels) and blood-lead concentrations, leads to a recognition that R&R
activities are a possible source of lead exposure to both children and adults.
A debate arises, however, when assessing which R&R activities produce a lead hazard.
Upon closer examination of the literature, the activity cited most often as responsible for lead
exposure and hazards is lead-based paint removal. Even industry sources (e.g., the National
Association of Home Builders) concede the hazard of paint removal. However, some literature
sources claim that other activities such as surface enclosure or component removal generate a
non-significant lead hazard. In fact, this dichotomy is fairly complete: lead-based paint removal is
widely recognized as generating a lead hazard, but other R&R activities are only sparsely
addressed. For example, the NAHB suggests that "while remodeling and lead abatement tasks
other than paint removal may generate small amounts of airborne lead dust and paint debris, with
proper cleanup they represent a minimal hazard to workers and occupants" (NAHB, December
1992). Within the literature, R&R activities have been effectively characterized into two groups:
paint removal and everything else.
Chemical stripping, surface sanding, and heat gun stripping are procedures usually used for
paint removal. All are widely acknowledged to generate a lead hazard in the presence of lead-
based paint. There are numerous studies of lead poisoning of construction workers employed at
removing lead-based paint. The HUD Abatement Demonstration Study found unacceptably
elevated levels of personal exposure to airborne lead for workers removing lead-based paint with
heat guns, despite temperature controls to prevent vaporization of the paint (U.S. Dept. of HUD,
2-4
-------
1991). Industry also concedes the problem as reported in the NAHB guidelines entitled, "What
Remodelers Need to Know About Lead" (NAHB, December 1992). The evidence from the
literature and the broad consensus suggests that additional studies of paint removal will yield little
new information on this activity's potential lead exposure to workers and occupants.
Few references address lead hazards or exposure associated with the other renovation and
remodeling activities, and the information is insufficient for assessing the hazards presented in
R&R activities other than paint removal. A National Institute of Safety and Occupational Safety
and Health (NIOSH) evaluation of the lead hazard produced during the HUD Abatement
Demonstration Study found that "less than 5% of the personal exposures to lead measured for ...
enclosure and replacement methods, and none of the exposures for encapsulation, ... exceeded the
OSHA PEL* of 50 |ig/m3" (NIOSH, February 1992). These figures were for abatement methods,
however, not for typical R&R methods. Conversely, an abatement study conducted by NIOSH
evaluating the hazards associated with cleaning up lead-based paint found that short-term personal
exposure levels met or exceeded the OSHA PEL limit in 16 of 36 workers. NIOSH reports that
"... the results of the study are of interest because many construction workers potentially perform
similar activities during renovation ..." (NIOSH, Ohio University, May 1993). Additionally, the
Comprehensive Abatement Performance (CAP) Pilot Study reported elevated dust-lead levels
resulting from extensive renovation (including large surface demolition and plumbing installation)
of a formerly abated housing unit (Battelle Report to USEPA, September 1994). A two-year
monitoring study of lead exposure in infants found "home refmishing" as a contributor to dust and
blood-lead levels (Rabinowitz et al, 1985). The specific activities that constituted home
refmishing were not thoroughly discussed, but activities other than exclusively paint removal were
considered. Therein lies much of the reason for the scarcity of information about other R&R
activities. Discussions of R&R activities in the literature focus on the source of the lead hazard,
not the activities that contribute to that hazard. Since many R&R activities occur within the
residence, lead-based paint is usually identified as that source. Any information about the
activities is anecdotal. The lead hazard or poisoning documented in case studies may be caused
by activities other than paint removal, but details are not available. It was concluded, as a result,
that the lead hazard or exposure from renovation and remodeling activities other than paint
removal cannot be assessed from the literature alone.
2.4 REVIEW OF AVAILABLE DATA SOURCES
Since the literature search yielded little information concerning the lead hazard associated
with R&R activities other than paint removal, a comprehensive search was undertaken to uncover
any existing sources of data that might relate a specific R&R activity to the lead exposure
presented by that activity. Starting with the contact list compiled earlier and likely resources
identified in the literature, a list of potential data sources was created. This list included
individuals from special interest groups in the construction industry, independent researchers, the
U. S. military, public housing authorities (PHAs), and other government agencies. To ensure a
comprehensive search, individuals who were contacted were asked to identify possible data
Permissible exposure limit.
2-5
-------
sources or individuals who might lead to a data source. If a potential new source was identified,
then the individual or organization was added to the list. This was continued until no new
individuals were identified.
Telephone calls were made to each individual on the list. During the course of the
conversation, the scope of the project was conveyed, emphasizing strongly the need to find data
that could directly relate lead exposure or contamination to a specific R&R activity. During the
phone interview, a determination was made whether the recommended data source was
representative of a renovation and remodeling activity. Generally, data collected when abatement
techniques were employed was not considered representative of the exposure that may occur
during "typical" R&R activity. [It is recognized that what constitutes "typical" R&R is currently
under revision, owing primarily to the issuance of OSHA's Interim Final Rule for Lead Exposure
in Construction (29 CFR 1926.62), which became effective during the term of this study.]
Exceptions were allowed for some abatement activities, such as exterior work and encapsulation,
since these activities are very similar to R&R activities. If there was any doubt as to the
applicability of the data, the person was asked to send the data with some documentation if
possible. A determination was made whether to include the data after the data collection methods
were more closely reviewed. This ensured that a potential data source was not missed because of
miscommunication or misunderstanding. Those individuals with data that fit well into the scope
of the study were asked to send the data in whatever format was convenient, with documentation
whenever possible. All data were compiled into a common data set so that summaries could be
made.
More than 40 individuals and organizations were contacted. However, very few renovation
and remodeling data sources were uncovered. A large body of information covers specific
activities and the lead exposure associated with these activities when abatement techniques are
employed, but very few sources collected information during activities that resemble or actually
are R&R settings. For instance, abatement data from several PHAs were available, but because
many of the abatement projects were sponsored by HUD grants, HUD's guidelines on abatement
(which include extensive worker and occupant protection measures and cleanup) were followed.
Some of the activities that very closely resemble renovation and remodeling activities, even when
abatement guidelines are followed, include interior and exterior encapsulation, component
replacement, and some surface preparation. Whenever possible, these activities were extracted
from the abatement studies and included in the data base.
Several potential data sources reported monitoring data that could not be related to specific
activities or to R&R activities that were conducted on surfaces coated with lead-based paint. As
a result, these data could not be placed in an appropriate context in this study and therefore were
not included in the data base. Other contacts have confirmed the lack of lead exposure data
related to renovation and remodeling. One military official, representing a data source included in
Table 2-1, stated that before their study in 1992, they could not find any data related specifically
to R&R.
2-6
-------
Table 2-1 lists several characteristics of the potential data sources identified. Of the 18
identified sources, the results from 10 studies were received. Of those 10, eight had information
that related exposure to a particular R&R activity. These eight data sources are discussed in
Section 8E of Chapter 8.
Table 2-1. Potential Sources of Existing Data Associating Lead Exposure With a Particular
Renovation and Remodeling Activity
Identified Data Sources
National Association of Home Builders (NAHB)
Private Environmental/
Abatement Contractors
Public Housing Authorities (PHAs)
National Institute of Occupational Safety and
Health (NIOSH)
California Department of Health;
New York State Department of Health;
Massachusetts Department of Environmental
Health
U. S. Military
Activities
• Demolition
• Window Replacement
• Component Removal
• Window Replacement
• Carpet Removal
• Demolition
• Surface Preparation
• Surface Preparation
• Kitchen Remodeling
• Exterior work
• Various R&R activities
• Cleanup
• Surface Preparation: Sanding and
Scraping
• Window Replacement
• Surface Preparation: Sanding, Scraping,
Power Sanding
• Cleanup
• Exterior Component Installation
• Interior Component Removal
• Painting
Number of
Sources Identified
2
4
5
2
3
2
In general, data provided by the PHAs were not representative of typical R&R or were not
available to the study. Much of the PHA data were collected in an abatement setting that was not
representative of typical R&R. Because much of the abatement work was contracted out, the
data did not reside with the PHAs. The need to obtain data from contractors greatly diminished
the potential for collecting PHA data. Only one PHA contractor provided data to this study.
The data identified and obtained from sources other than a PHA tended to be more closely
related to R&R activities. The individuals contacted were very much aware of the potential
problems of lead exposure to R&R workers and were sensitive to the issues involved in locating
data specific to R&R. Several good prospects were developed through these individuals, and
potentially informative data sets were obtained.
2.5 SUMMARY
Available information on the lead hazards associated with R&R activities other than paint
removal is very limited. Despite a comprehensive literature search and hundreds of phone calls to
individuals very familiar with lead issues and policies, little information was uncovered. Only
2-7
-------
eight sources were found to contain data on lead exposure related specifically to R&R activities.
The data collected from these other sources are summarized and discussed in Section 8E of
Chapter 8.
Two additional studies were identified that were in the planning stages at the time the EFSS
was conducted: 1) a study sponsored by a state government agency (with support of a HUD
grant) of surface preparation using dry scraping, heat guns, and torches, and 2) a study sponsored
by the NIOSH of the renovation and remodeling of post offices and General Accounting offices
throughout the U.S. These studies have the potential to provide additional useful data for
investigating the relationship between lead exposure and R&R.
2-8
-------
3.0 QUALITY ASSURANCE
In the EFSS, as in any environmental sampling program, a variety of sources of error
exists that could potentially affect the quality of the study results. A careful assessment of
these sources of error was made, and quality control measures were implemented to help
minimize their effects. The most important quality control measure was the Quality Assurance
Project Plan (QAPjP). The QAPjP is intended to document the entire process involved in
executing the field program, along with appropriate quality assurance measures. QAPjPs were
prepared for the carpet removal and window replacement phases, and an addendum was
prepared for the CED phase. The entire project team contributed to the QAPjPs, which were
reviewed and approved by the EPA/OPPT technical staff and Quality Assurance Officer.
Subjects covered in the QAPjP include:
• Project overview.
• Project organization and management structure, including organization and
personnel responsibilities, and personnel qualifications.
• Study objectives including chemical data quality objectives.
• Method selection and analytical method performance including objectives and
specifications for precision, accuracy, and verification and validation.
• Sampling plan including sampling frame construction, screening protocols, unit
selection criteria, field QC samples, and sample size determination.
• Analysis plan including data review and transfer procedures, data quality
assessment, and statistical analysis procedures.
• Sample collection procedures, including specification of the field sampling team,
determination of sampling locations, and sampling schedule, transfer and storage
procedures.
• Sample identification, tracking, and handling.
• Sampling equipment, including equipment performance requirements, preventive
maintenance, corrective actions, contamination avoidance, and equipment
calibration.
• Sample preparation procedures.
• Data processing procedures, including storage, transfer, tracking, and outlier
detection procedures.
• Specification of all data collection forms.
3-1
-------
• Health and safety procedures, including Human Subjects Review.
• Audit requirements, including system, performance evaluation, and data audits,
and audit reporting and corrective action.
• Reporting requirements.
The QAPjPs may be referenced for additional details on any of the above subjects.
Appendix B in Volume II of this report documents data processing procedures followed
in this study and presents results of the outlier detection analyses and decisions. Only one
outlier in the three phases of the EFSS study was deleted from the statistical analysis as
documented in Appendix B.
Appendix D in Volume II presents the results of analysis of quality control samples
collected in the study including field blanks and field side-by-side samples. In all three phases
of the EFSS, the results of analyzing initial calibration verification and continuing calibration
verification samples in each instrumental analysis batch were within the protocol criteria of
+.10%. This indicates that the analytical instrument was properly calibrated for all of the
sample analyses. Tables D-3a through D-3c in Appendix D report the status of meeting data
quality objectives within batches in the carpet removal, window replacement, and CED
phases, respectively.
3-2
-------
4.0 STUDY DESIGN FOR THE ENVIRONMENTAL
FIELD SAMPLING STUDY (EFSS)
The purpose of the EFSS was to assess the lead exposure and lead disturbance associated
with various types of R&R activities through environmental measurements of lead in air and dust.
The EFSS consisted of four phases of data collection that either addressed a specific type
of R&R activity or a group of activities. These phases were:
1. An information-gathering phase, to uncover the current body of information
concerning lead exposure associated with renovation and remodeling and to analyze
environmental lead exposure data collected in other studies that investigated
exposures associated with specific R&R target activities, or with large R&R projects
2. A carpet removal phase, to investigate lead exposure associated with carpet
removal activities
3. A window replacement phase, to investigate lead exposure associated with window
replacement activities
4. A controlled, experimentally designed (CED) phase, to investigate lead exposure
associated with several R&R activities (large structure removal, small surface
disruption, HVAC repair, generic R&R activities) that are difficult to isolate in an
actual R&R job or are components of larger R&R activities.
The first phase was described in Chapter 2. In the other three phases, the same sampling and
laboratory analysis protocols were applied to environmental samples from various media. This
chapter presents common elements of the study design for these three phases. Chapter 8 presents
details on the specific implementation of the general study design.
4.1 OBJECTIVES OF THE EFSS
Technical objectives for the R&R study were presented in Section 1.1 of Chapter 1. The
primary technical objectives of the EFSS were to characterize the:
1. Personal exposure to airborne lead for workers during the performance of different
R&R target activities and combinations of activities, and determine if worker
exposure to airborne lead during those activities exceeds 50 |ig/m3 [the OSFEA
permissible exposure limit (PEL)].
2. Airborne lead levels in areas adjacent to the activity area (the area — room, floor,
etc. — in which the R&R activity took place) for selected R&R activities.
3. Amount of lead disturbed that settles on building surfaces within a specified period
following completion of the R&R activity.
4-1
-------
4. Extent that lead disturbance and exposure are affected by various factors such as
distance from the activity and pre-activity measures of lead contamination in the
building.
Some secondary objectives of interest related to sampling methodology were incorporated into
selected phases of the EFSS to aid in refining the design of this study. The secondary objectives
are discussed in Section 4.3 and in greater detail in Chapter 7.
4.2 ENVIRONMENTAL SAMPLING
Although the activities, buildings, and, in some instances, types of samples differed for
each category of activities, the design and protocol for all environmental sampling were similar.
In general, for each monitored R&R activity, buildings containing lead-based paint suitable
for typical application of the activity were selected. Buildings were selected based on
predetermined screening criteria, as discussed in Chapter 5, that included taking X-ray
fluorescence (XRF) and/or inductively coupled plasma (ICP) samples on painted surfaces in
selected rooms of the unit.
Environmental measurements of lead were taken before, during, and after conducting the
target activity. The measurements taken in the EFSS included:
1. Personal Air Samples. Measures of airborne lead concentrations at a fixed flow
rate within each worker's personal breathing zone were collected by taking air
samples through a cassette filter mounted to the worker's lapel. An air sampler
pump was used to take the sample. The activity period was defined for each specific
R&R activity, but generally included immediate preparation for the activity, activity
conduct, and cleanup. For each worker, average exposure over the duration of the
activity period was calculated in |ig/m3.
2. Room (Ambient) Air Samples. Ambient air samples were collected for selected
activities in areas adjacent to the activity. Adjacent areas were used to address the
levels at which occupants might be exposed to airborne lead in other parts of the
building while conducting the activity.
3. Settled Dust Samples. Settled dust samples were taken either from stainless steel
dustfall collectors (see Section 4.3) or from selected areas such as floors, window
sills, window wells, and carpets. Samples were collected at varying distances from
the surfaces disturbed by the activity. Lead loadings from settled dust samples were
measured as indicators of the amount of lead disturbed by the activity and made
available as a potential exposure to occupants. Further discussion on the sampling
methodology for settled dust samples is provided in Section 4.6 below.
It was not possible to obtain a statistically based, representative sample of the exposed
populations or the defined jobs, because of recruitment difficulties in obtaining "real-world" R&R
4-2
-------
jobs as they are currently being conducted, cost constraints, and the wide variety of work
practices and building characteristics. Consequently, the EFSS focused on characterizing "case
studies" that were deemed "reasonably representative" (i.e., not atypical) of general R&R work as
it is currently conducted. Two different approaches were taken to characterize these case studies:
1. Monitoring "real-world" R&R jobs as they occurred in the field
2. Monitoring controlled, experimentally designed simulations of specified renovation
and remodeling activities in vacant buildings.
In the first approach, an attempt was made to select R&R jobs (case studies) that did not appear
atypical of general R&R work currently being conducted. These jobs needed to have high
potential for lead contamination because of the age of the building, presence of lead-based paint,
and types of activities being conducted.
In the second approach, the design team specified the type, quantity, and characteristics of
activities to be conducted in dwellings in the field. Experienced R&R workers, contracted to
perform the work, simulated target activities and generic R&R tasks (e.g., cutting and drilling) so
that measurements could be taken of lead disturbance associated with these tasks. The generic
tasks are often components of any number of R&R target activities and provide an indication of
potential hazards associated with many different R&R jobs.
Details of the sampling plan and design for each R&R activity or task monitored are
presented in Chapter 8 with the discussion of each phase of the EFSS.
4.3 EVALUATION OF SAMPLING METHODOLOGY
The EFSS was designed to measure lead disturbance and lead exposures associated with
specific short-term R&R activities. Sampling methodologies commonly used in occupational
studies, clearance testing, risk assessment, and previous abatement studies were employed
whenever possible in this study. In the case of settled dust, however, a different method of
characterizing the amount of lead disturbed by a specific activity was employed. Post-activity and
pre-activity measurements on building surfaces are commonly compared in lead contamination
studies in buildings. The alternative method involves placing one-square-foot stainless steel plates
on the floor immediately before the start of an activity at varying distances from the activity area.
These plates are referred to as stainless steel dustfall collectors (SSDCs). Settled dust samples
are collected from these plates at a specified period of time after completion of the activity.
Interest in evaluating the sampling methodology introduced additional objectives into the
early phases of the study. Those objectives were to:
• Compare results from the sampling methodology using SSDCs to results from the
sampling methodology employing pre- and post-activity floor samples
4-3
-------
• Compare results from settled dust samples collected one hour following completion
of an activity with results from settled dust samples collected two hours after
completion of an activity
• Compare results from settled dust samples collected using vacuum techniques with
results from samples collected using wipe techniques.
To assess the first objective, SSDCs were placed next to the post- and pre-activity floor
samples at all locations in the carpet removal phase, and at two out of three sampling locations in
each unit in the window replacement phase. The second objective was assessed by placing two
SSDCs side by side at all locations in the carpet removal phase, and at one out of three locations
in each unit in the window replacement phase. One side-by-side sample was collected
approximately one hour after completion of the activity and the other approximately two hours
after completion of the activity. The third objective was assessed by additional side-by-side
samples at two out of three sampling locations in each unit in the carpet removal phase and at a
total of 12 sampling locations during interior demolition work in the controlled, experimentally
designed (CED) phase. Results pertaining to sampling methodology evaluation are discussed in
Chapter 7.
4.4 HUMAN SUBJECTS REVIEW
All aspects of the field work involved in the EFSS were documented and submitted to
both the contractors' and EPA's Human Subjects Review Committees for review and approval.
All procedures complied with the requirements of the human subjects committees.
4.5 DATA GAPS AND LIMITATIONS OF THE R&R STUDY
Due to the broad scope of the EFSS and the R&R study as defined by Title X legislation,
combined with time and budget constraints, recruitment difficulties, and human subjects concerns,
insufficient information was available to make inferences on certain areas of interest to the study.
Data gaps that remain include:
1. Information on the relationship between R&R and occupants' (children's)
blood-lead concentrations. The combination of environmental measurements and
worker blood-lead concentrations was expected to provide sufficient exposure
information to address regulatory needs. However, the blood-lead concentrations of
R&R workers were very low, while the amount of lead released to the environment
was very high. These conflicting outcomes imply that measuring environmental lead
levels might be an inadequate surrogate for measuring human blood-lead levels.
This, combined with difficulties in determining a health-based standard for
acceptable environmental lead levels, places more importance on the relationship
between R&R and occupant exposure as measured by blood-lead concentrations.
Therefore a follow-on study will be conducted to evaluate the impact of the conduct
of R&R activities on elevated blood-lead concentrations in children. This study will
4-4
-------
be conducted jointly with the University of Wisconsin and the State of Wisconsin
Public Health Department.
2. Exposure of people other than occupants and workers was not assessed. Other
potentially exposed populations include workers' families and residents of
neighboring buildings.
It is important to note the limitations of the data collected in the EFSS:
1. The EFSS field monitoring work represents a series of case studies that were not
selected by a random sampling scheme. Generalizing the results of the EFSS to a
broader population must be based on a qualitative assessment of how representative
the case studies are of that population. However, no reason was uncovered to
believe that the case studies were atypical of general R&R work as it is conducted in
an unregulated environment containing high levels of lead-based paint.
2. Measurements of lead distributed into the occupants' environments were
collected before cleanup. The effect of different cleanup methods was measured
for two target activities in the EFSS, and data were collected in the WCBS on
typical cleanup methods employed by R&R workers. However, additional
information on the extent of typical cleanup would be useful.
3. The focus of the data collected in the EFSS was on target housing. Exposure
differences between building environments from a lead abatement perspective are
discussed in EPA's proposed rule 40 CFR Part 745, "Requirements for Lead-Based
Paint Activities," where an argument is made that public buildings and target
housing represent similar exposure environments. However, there are no data at
this time to assess whether environmental exposures monitored in target housing are
representative of environmental exposures encountered in public or commercial
buildings.
Further discussion of limitations on inferences drawn from the results of the EFSS are
presented in the discussion of overall results in Chapter 9.
4.6 SAMPLING AND ANALYTICAL METHODS
During the course of the EFSS, field sampling activities were conducted to monitor a
variety of renovation and remodeling activities. The following categories of field samples were
collected during the study:
Wipe sampling of settled dust
Vacuum sampling of settled dust
Personal exposure samples
Area airborne dust samples
Paint chip samples
4-5
-------
• Plaster samples.
The following two sections discuss the sampling and analytical methods applied to the
above categories of samples. Further details on sampling and analytical methods can be found in
the QAPjP.
4.6.1 Sampling Methods
Wipe Samples of Settled Dust
Wipe samples of settled dust were collected from the SSDCs, as well as directly from
other horizontal sampling surfaces (e.g., HVAC ducts). All wipe samples were collected using
commercially available, premoistened disposable wipes (Wash-a-bye Baby™). The protocol for
collecting a wipe sample is as follows:
• Don a clean pair of powderless vinyl gloves.
• Open the lid of the wipes container and remove several wipes. Discard the initial
wipes removed, and use the next wipe from the container to collect the sample.
• Place the wipe flat on the sample surface. (Note that in some cases the sample
surface was a 1-ft2 steel template; in other cases, the surface was a ventilation duct,
etc.)
• Using an open flat hand with the fingers together, wipe the marked surface in an
overlapping "S" pattern. The initial pattern should be from side to side and then
from front to back so that the entire sample collection area is covered.
• When sampling small areas (i.e., less than 1-ft2), the wipe should be folded in half
with the sample side folded in. Repeat the wiping procedure within the defined
surface area.
Vacuum Sampling of Settled Dust
Special equipment is employed for collecting vacuumed settled dust. The vacuum
sampling equipment included a modified HVS3 cyclone dust collector and a Red-Devil® portable
vacuum (Figure 4-1). The cyclone dust collector has an output nozzle on the top, which is
connected to the portable vacuum, and an inlet on the side, which is connected to a tygon
"collection" tube. The bottom segment of the cyclone dust collector is conical in shape, with
internal threads at the narrow end of the case to accommodate the sample collection bottle. A
Teflon sealing ring forms the joint between the cyclone top and bottom segments. A cyclone
4-6
-------
Figure 4-1. Cyclone Dust Collector
-------
dust collector has advantages over a dust collector that uses a filter. The collection bottle
eliminates the need for pre- and post-weighing and conditioning of the filter, and has a larger
capacity. The protocol for collecting a vacuumed settled dust sample is as follows:
• Place a clean grey PVC transition piece onto the tangential inlet of the cyclone head.
• Position the cyclone in a vertical position and securely screw the sampling collection
bottle into the lower threaded end of the cyclone.
• Insert the proper end of the collection tube into the grey transition piece. Attach the
vinyl tubing to the sampler case's 3/4-inch inlet.
• Run an extension cord from the nearest 110-V AC outlet (or generator) to the
designated sampling location and plug in the portable vacuum.
• Turn on the pump and vacuum the sample area in overlapping passes (i.e., at least
50% overlap), initially left to right, then front to back over the entire designated
area. Repeat vacuuming pattern for two minutes.
• When vacuuming is complete, turn off the vacuum and keep cyclone vertical. Raise
the humidity within the cyclone by slowly blowing three breaths into the grey
transition piece. Tap the cyclone case three times to dislodge any remaining debris.
• While keeping the cyclone vertical, unscrew the dust collection bottle from the
cyclone and recap the dust collection bottle.
Exposure Samples
Personal exposure monitoring to airborne lead was conducted using NIOSH method 7300.
The personal sampler, illustrated in Figure 4-2, pulls air across sample media at a known sample
flow rate. The flow rate of the sampling pumps is set between 1 and 4 liters per minute, with a
preference for 4 liters per minute. The exact flow rate is determined during calibration procedures
that are performed before and after each sampling period. The protocol for collecting personal
exposure samples is as follows:
• Samples are collected by using a battery operated pump (Gilian HFS 113 and 513)
attached to a tygon tubing with the sample media attached to the opposite end of the
tubing.
• Don clean vinyl gloves. Remove inlet and outlet plugs from an unused 0.8 |i, 37
mm mixed cellulose ester filter cassette. Attach the tygon tubing to the sample
media by inserting a nipple into the open end of the tubing (i.e., the end opposite of
that attached to the sampling pump) and inserting the nipple into the outlet opening
of the cassette.
4-8
-------
CO
Figure 4-2. Personal Air Monitoring Sampler (Left) and Ambient Air Monitoring Sampler (Right)
-------
• Mount the sampler at the waist using the belt clip attached to the sampler. Clip the
filter cassette to the worker's lapel (make sure that the cassette remains in the
worker's breathing zone), making sure to secure the tubing to the worker in a
manner such that is does not interfere (e.g., tape the tubing to the worker's back).
The inlet of the filter cassette should be positioned slightly downward to avoid dust
from passively falling into the cassette.
• Turn the sampling pump on and record time.
• When sampling is complete, turn the sampling pump off and record time. Cap the
inlet and outlet plugs of the cassette.
Area Airborne Dust Samples
Area exposure monitoring to airborne lead was conducted using Gilian AirCon High
Volume samplers. The ambient air sampler, illustrated in Figure 4-2, has a flow rate set at rates
up to 12.5 liters per minute. The exact flow rate is determined during calibration procedures that
are performed before and after each sampling period. The protocol for collecting area airborne
dust samples is as follows:
• Set up tripod at sample location and set pump on floor under tripod.
• Don clean vinyl gloves. Remove inlet and outlet plugs from an unused 0.8 \i, 37
mm mixed cellulose ester filter cassette. Mount the filter cassette to the tripod mast
at breathing-zone height (approximately five feet above the floor). Attach the tygon
tubing to the sample media by inserting a nipple into the open end of the tubing (i.e.,
the end opposite of that attached to the sampling pump) and inserting the nipple into
the outlet opening of the cassette.
• Turn the sampling pump on and record the time.
• When sampling is complete, turn the sampling pump off and record the time. Cap
the inlet and outlet plugs of the cassette.
Paint Chip Samples
The protocol for collecting a paint chip at a sampling location is as follows:
• Don a clean pair of disposable leather gloves.
• Using a cutting tool, score the perimeter of the area to be removed.
4-10
-------
• Affix a tray, paper funnel, or equivalent collection device directly below the
sampling location. Begin removing the paint from the substrate. If possible, peel
the paint off of the substrate by sliding the blade along the score and underneath the
paint. Remove all paint down to the bare substrate and transfer to the collection
container.
Plaster Samples
Samples are collected from painted surfaces to characterize lead levels in surfaces about to
be demolished or disturbed during renovation and remodeling activities. When collecting plaster
samples, the selection of the surface to be sampled is important. Deteriorated plaster can become
crumbly, increasing the potential for sample contamination. For example, when chipping into a
wall the plaster can easily become pulverized and fall to the floor. It is best to select a firm, rigid
surface and collect the plaster sample in accordance with the following steps:
• Don a clean pair of disposable vinyl gloves.
• Select the surface and area to be sampled (do not outline the sample area with any
type of marker, pen, pencil, etc.).
• Affix a tray, paper funnel, or equivalent collection device directly below the
sampling location. Note that because of the relative mass of the plaster compared to
the collection device, the person collecting the sample or an assistant may need to
support the device.
• Using a chisel and hammer, carefully break through the surface of the sample area.
• Begin removing the plaster by gently prying it away from the surface being sampled.
• Allow the plaster sample to drop into the collection device. As mentioned
previously, the collection device may need to be supported at this time.
• Carefully transfer the plaster sample from the collection device into a sample bottle.
Cap the sample bottle.
4.6.2 Analytical Methods
Analysis procedures used in the EFSS study were as follows:
• Settled dust, vacuum samples: Digested using a modified version of EPA SW-846
Method 3050. Analyzed using a modified version of EPA SW-846 Method 6010A,
Inductively Coupled Plasma (ICP).
4-11
-------
• Settled dust, wipe samples: Digested using a modified version of EPA SW-846
Method 3050. Analyzed using a modified version of EPA SW-846 Method 6010A,
Inductively Coupled Plasma (ICP).
• Air, personal exposure samples: Digested and analyzed using a modified version of
EPA SW-846 Method 3050.
• Air, area airborne samples: Digested and analyzed using a modified version of EPA
SW-846 Method 3050.
• Paint chip samples: Digested using a modified version of NIOSH Method 7082.
Analyzed using a modified version of EPA SW-846 Method 6010 A.
Primary instrumentation for the analysis of vacuumed dust and dust wipe samples was an
Inductively Coupled Plasma—Atomic Emission Spectrometer (ICP-AES). A Thermo-Jarrell Ash
simultaneous ICP was utilized for this analysis effort. Primary instrumentation for the analysis of
personal exposure and area airborne samples was a Varian SpectrAA 300Z graphite furnace
atomic absorption spectrometer (GFAAS).
More details on the analytical method, including protocols, are provided in the study's
QAPjP.
4-12
-------
5.0 OVERVIEW OF RECRUITMENT IN THE EFSS
The study objectives and design, discussed in the preceding chapters, defined three basic
criteria for the recruitment process:
1. The workers monitored and the R&R activities performed must be typical of general
R&R work as it is currently being conducted.
2. The buildings undergoing R&R must be representative of typical target housing or
public and commercial buildings and must contain interior lead-based paint.
3. Recruitment and subsequent study procedures must meet all requirements specified
by EPA, Battelle, and MRI's Human Subjects Committees.
Reasons for the first criterion, including reasons for the decision not to monitor abatement
work, were presented in Chapter 2. The decision to monitor R&R workers conducting activities
typical of general unregulated R&R work was reinforced during the recruitment process;
contractors repeatedly expressed the opinion that R&R work being conducted by government
contractors who take extensive precautions against lead exposure is not representative of much of
the R&R work being conducted by R&R contractors focused on the private sector. Our
definition of "typical unregulated R&R work" did allow inclusion of a worker who takes personal
precautions during the R&R activity, such as wearing a respirator or protective clothing.
However, we did not include dust reduction measures normally encountered in abatement work,
such as:
• Misting surfaces that will be disturbed
• Using plastic to enclose surfaces to be removed
• Using negative air or tools with HEP A vacuum attachments.
The second recruitment criterion was chosen to help focus the study on the most likely
environments for a lead exposure problem, while ensuring that buildings selected were not
atypical of the environments defined by the scope of the study. The screening of candidate
buildings included testing for lead-based paint using XRF methods and/or ICP analysis on paint
chip samples to measure lead concentrations (mg/cm2) on painted surfaces. The XRF testing was
intended for screening purposes only. Details on the minimum levels of lead-based paint required
for including a building in the study are presented in the QAPjP. In general, all sites included in
the study had extensive lead-based paint, and the XRF results were used along with other criteria
(such as the size and overall condition of the building) to choose between candidate sites. XRF
results were not used in any subsequent statistical analysis. Where measurement of the paint in
the substrate being disturbed was required, paint chip samples were collected at the time of the
R&R activity.
The third criterion, compliance with Human Subjects Committee guidance, affected
separate phases of the study differently. In the carpet removal and window replacement phases,
where R&R jobs already planned in the marketplace were being recruited, the primary Human
5-1
-------
Subjects requirements were that the study not alter any of the work that was to be monitored and
that workers and residents sign approved informed consent forms. In the CED phase, on the
other hand, Human Subjects guidance required that the work be conducted by contractors with
experience in worker protection practices, an OSHA-certified respirator program in place, and
who agreed to follow the OSHA lead standard for the construction industry (29 CFR 1926.62).
In addition, for the CED phase, buildings where the work was to be conducted were required to
be scheduled for subsequent demolition, gutting and complete restoration, or complete abatement
and clearance testing before reoccupancy.
Multiple approaches were taken in the attempt to recruit the "real-world" R&R
contractors and workers. These included newspaper ads, telephone solicitation, letters, flyers,
contact with unions, trade organizations, government agencies, and ongoing R&R programs, and
targeting schools and other institutions that were suspected to have reason to be interested in the
lead problem. A telephone script was composed that attempted to:
• allow R&R workers to understand the benefit of the study
• reduce their liability concerns
• stress that their participation would not disrupt the job or their normal work practice
in any way.
Attempts were made to recruit contractors directly as well as indirectly through the individuals
hiring the job or through union participation. Workers and contractors were offered
compensation for any inconvenience resulting from participation in the study.
Overall, recruitment of "real-world" non-abatement R&R jobs for participation in this
study was time consuming, cost intensive, and very problematic. The primary reasons for this
difficulty were:
1. Liability concerns. The fear of being held liable by workers, homeowners, the
public, EPA, OSHA, or NIOSH was the single biggest disincentive to participation
in the study. Very early in the recruitment process, one carpet removal contractor
told us "one lawsuit could put me out of business." This concern was repeated
often. Larger institutions often appeared reluctant to lose the protection that was
perceived to come with ignorance. The attempt to recruit the maintenance staffs of
several institutions was unsuccessful for this reason.
2. Negative reactions to EPA and OSHA. Many contractors expressed concerns that
their participation would result in either a fine from OSHA for working incorrectly,
or concerns about newly mandated requirements that they did not want or did not
feel were justified. The president of a window replacement firm with offices in
Wisconsin wrote,
5-2
-------
"A concern I have with the methodology of your study, is that new Federal
rules will be formulated based upon your study. Your study as presently
designed is not representative of the type of window replacements we do. In
fact, the vast majority of window replacements are done without disturbing
the original window frame. "
3. Contractors, agencies and individuals too busy to participate. Many individuals
we spoke with would not even take the time to fully listen to the description of the
study before stating, "I'm too busy and don't have time for this now. " Contractors
who seemed receptive but too busy to talk at the time of the first call were sent a
letter describing the study and then called back at a later date. In general, first
refusals were seldom converted by a letter or a follow-up phone call.
4. Difficulty in recruiting a " contracted job. " A unique challenge to this recruitment
effort was that it was not single individuals or organizations being recruited
ultimately but rather contracted R&R jobs. This required coordination and
recruitment of two parties and the ability to firmly determine schedules so sampling
could be arranged. The inconvenience involved was often perceived by those
recruited to outweigh any advantages to participation including the compensation
offered.
5. No match for the unit or work requirements. Many contractors did not work in
the older buildings (>50 years old) we required or did not meet phase-specific
requirements such as full replacement of wood windows.
6. Failure to fend lead-based paint after a building was recruited. In the recruitment
of large R&R projects and potential CED sites, a high ratio of buildings screened did
not contain appreciable amounts of lead-based paint. Six units that were eventually
used in the CED phase had sufficient interior lead-based paint, but seven other units
that were more than 50 years old did not contain sufficient lead-based paint
anywhere in the interior of the building.
The recruitment difficulties led to more emphasis on the CED phase of the study where
the recruitment effort became one of locating vacant buildings and cooperative owners rather than
willing contractors. Details of the recruitment process, results obtained, and lessons learned are
presented below.
5.1 DETAILS AND RESULTS OF THE RECRUITMENT PROCESS
The details and results of the recruitment process are presented below for each of four
study phases, in chronological order. Results of one phase were used to formulate the strategy
and effort for subsequent phases.
5-3
-------
5.1.1 Carpet Removal
The first four homes in the carpet removal phase were located in the Oakland, California,
area. They were selected through cooperation of the Alameda County, California, Planning
Department, Housing and Community Development Program. The four dwelling units were
already scheduled for lead-paint abatement with removal of the carpets planned as the first activity
in the abatement process. The Alameda County agency also provided floor plans and existing
XRF sampling results for these units. Representatives of the study field team visited each unit to
perform a visual survey and found them adequate for study participation.
The Alameda County agency also recruited the R&R contractor who performed all carpet
removal activities for the four units. The contractor was subject to regulations imposed by
Alameda County in such areas as waste disposal, worker safety, and contamination control of
residential belongings. There was no requrirement specifically to minimize dust generation (e.g.,
misting the carpet).
The attempt to select four additional homes in another city proved much more
problematic. Initially, contractors advertising in St. Louis and Kansas City, Missouri, were
contacted by phone. They all declined to participate in the study, primarily due to liability
concerns. Next, fliers were placed in home-improvement stores and government agencies and
trade union representatives were contacted. Finally, newspaper advertisements were placed in St.
Louis newspapers offering an incentive to homeowners replacing their own carpet, and providing
a toll-free number to call.
This last recruitment step provided additional candidate homes rather easily. The
homeowners did not express anything in the way of liability concerns, fear of EPA, NIOSH, or
OSHA, or conflicting business interests. Also, the homeowners considered the monetary
incentive as a positive motivating factor, while the contractors considered it insignificant.
Recruitment of resident-workers enabled the study to include data on carpet removal performed
by nonprofessionals. The candidate homes were classified according to the following criteria:
• Age of dwelling unit (priority given to homes built before 1940)
• Availability of dwelling unit to the study
• Length of time since carpet was installed (priority given to carpet in place for at
least 8 years)
• Lead-based paint in the dwelling unit (unit has not recently undergone lead
abatement).
Study recruiters narrowed the field to eight promising candidates, and the field team and an XRF
contractor performed XRF and visual surveys during scheduled screening visits. Based on the
findings of the screenings, study statisticians then selected four units for the study.
5-4
-------
5.1.2 Window Replacement
The initial approach to identify and recruit individuals performing window replacement
work or having this type of work done in their homes was to place advertisements in newspapers
in Baltimore, Maryland, and St. Louis, Missouri. The EPA-approved ads described the study,
solicited participants (homeowners or contractors), offered an incentive, and listed a toll-free
number to call. This method of recruiting was abandoned rather quickly because there was
virtually no response. In contrast to the carpet removal phase, no homeowners performing the
work themselves were uncovered.
The next step was telephone soliciation by trained recruiters using an EPA-approved
script. They solicited window installation companies and contractors in thirteen selected U.S.
cities. The cities were Baltimore, Hartford, Boston, Columbus, St. Louis, Youngstown, New
Haven, Washington D.C., Newark, Cincinnati, Chicago, Milwaukee, and Cleveland. Most
companies and contractors contacted were listed in municipal telephone directories. Additional
contacts included trade organizations, unions, large and small licensed window installation firms,
independent contractors, business owners, and home owners. Although this recruitment
technique yielded the largest number of contacts (more than 800), it was largely unsuccessful.
Only five individuals were willing to participate in the study: three contractors and two
building owners. All five were located in Columbus, OH; our assessment is that all five were
influenced to participate in the study because of Battelle's involvement in the study and their
reputation in the Columbus area. In fact, one of the window installation companies, which
located two of the four homes eventually enrolled, wanted to participate so that it could advertise
its participation in a study with Battelle as a way to enhance business.
The four homes eventually selected for the study were identified by three professional
contractors as candidate study units. Two contractors were general building contractors, and one
was a specialized window removal/installation contractor. Representatives of the field sampling
team visited each eligible unit to conduct a visual survey and to supervise an XRF screening of
painted window components. Information from this screening visit was used to determine
whether the unit would be selected for the study, and if so, which of three windows would be
considered for environmental sampling. The four study units were located in the Ohio cities of
Piketon, Columbus, Richwood, and Plain City.
5.1.3 Large Projects
From a design perspective, there was significant interest in obtaining a large R&R project
involving multiple activities conducted simultaneously. A sampling plan was developed for
monitoring such a project. However, this project could not be accomplished in the time frame
available to the study.
The recruitment effort for finding a large R&R project was extended to the following
potential candidate sources:
5-5
-------
1. Large public facilities such as hospitals, universities, and schools
2. Military bases and state and federal agencies controlling government buildings
3. Solicitation of help from national trade unions.
In general, individuals contacted for the large R&R project were more receptive than
those contacted in the window replacement recruitment effort. Most quickly understood the
study and were in agreement with its purpose. However, a major problem for almost all large
R&R projects was that final approval for participation did not rest with a single individual or
department. Typically, several levels of approval were required including that of the legal, health
and safety, and facilities departments.
For example, one month-long solicitation of a large university had resulted in approvals
from the Environmental Safety and Health Department, university lawyers, and Board of
Trustees, but a lack of final approval because of the concerns and narrower interests of the
Facilities Department. In large institutions, including universities, there appears to be concern
about disclosing a lead problem.
Trade unions cooperated in recruitment efforts to identify eligible jobs and contractors.
The United Brotherhood of Carpenters (UBC) alone sent over 2,000 letters to its local chapters
and district councils soliciting their help in locating jobs. The only solid prospect from this effort
was a seminary in Ohio. Although the seminary was more than 60 years old, no lead paint was
found in the interior.
Failure to locate a suitable large project led the project team to abandon the effort in favor
of the use of hired contractors performing simulated R&R work in the CED phase.
5.1.4 CED Phase
Recruitment for the CED phase was more successful than recruitment of "real-world"
jobs. The fact that contractors would be hired to do the work meant that basically only vacant
buildings needed to be found. Public housing authorities, government and private agencies were
willing to offer candidate buildings. Although many buildings were not suitable because they
were too deteriorated, sufficient numbers of vacant, but habitable buildings were located for the
CED phase. This phase of the study was conducted in Baltimore, Maryland, and Denver,
Colorado.
In Baltimore, a number of row houses scheduled for "gut rehab" were recruited from a
private developer. During initial site visits, some of the dwellings were determined to be in such
poor condition that they could be excluded from the study. Three buildings were in habitable
condition, but one was found to contain an insufficient amount of lead-based paint during the
XRF screening. The other two were enrolled in the study. The contractor employed in Baltimore
met all study requirements and had many years of general construction and R&R experience
before concentrating on lead abatement work.
5-6
-------
Buildings in Denver were recruited through the cooperation of Denver Housing Authority
(DHA). This agency identified eligible vacant buildings scheduled for subsequent abatement or
"gut rehab." CED activities were conducted in four different buildings. All activities were
conducted by DHA maintenance staff with experience in general R&R tasks as well as in lead
abatement.
A summary of the recruitment effort and results for each phase of the EFSS is presented in
Table 5-1.
5.2 GENERAL CHARACTERISTICS OF THE WORKERS AND UNITS RECRUITED
This section presents a general description of the workers and buildings (units) included in
the three phases of the EFSS. Further details on specific unit characteristics for a given study
phase are included in Chapter 8.
5.2.1 Worker Characteristics
The workers performing R&R activity in the EFSS differed in many aspects, including
experience in performing the activity, awareness of potential lead hazards associated with the
activity, and the level of protection from exposure. Following the R&R activity at each study unit
in the carpet removal and window replacement phases, field team representatives interviewed the
R&R workers to characterize the type of worker participating in this phase. The worker survey
collected information on such items as experience level, familiarity with the hazards associated
with lead exposure, and opinion on how typical the activity has been at a site compared to similar
jobs. Results of this survey were used only to characterize the workers participating in this study.
They cannot be used to make inferences on the general population of R&R workers.
Carpet Removal
In the carpet removal phase, the four Alameda County, CA, units had professional carpet
removal workers who were employees of an R&R contractor with experience in dealing with
environmental hazards such as lead contamination. Approximately 25% of the R&R jobs
performed by this contractor involved carpet removal, and approximately 30% of the jobs
consisted of lead-based paint abatement. In this study, the contractor's employees wore Tyvek
suits and respirators when performing carpet removal in each unit. The workers were aware of
various precautionary measures to reduce dust generation, although the only such measure taken
in this study was using plastic sheeting to cover occupant belongings and seal off nonactivity
areas.
The owner of the contracting company in Alameda County, who had more than 35 years
of R&R experience, participated in carpet removal in one unit and was monitored for personal
exposure. The four other workers for this contractor had from 3 months to 15 years of
experience in R&R and in carpet removal specifically.
5-7
-------
Table 5-1. Summary of Recruitment Efforts and Results in the EFSS
Carpet Removal Phase
Target for
Recruitment
Government agencies
and contractors
Contractors
Contractors
Homeowners
Method
Personal and inter-organizational contacts;
telephone calls
Telephone calls with predetermined script
Fliers posted in retail outlets frequented by
R&R contractors
Newspaper advertisements with cash
incentive
No. of
Contacts
>10
>100
N/A
Ads placed in
two cities
Results
4 units
No success
No success
4 units
Window Replacement Phase
Government agencies
Contractors or
homeowners
Contractors
Personal and inter-organizational contacts;
telephone calls
Newspaper advertisements with cash
incentive
Telephone calls with guided script and
follow-up letters
>10
Ads placed in
two cities
>800
No success
for non-
abatement
work
No success
4 units
CED Phase
Government agencies
and special interest
groups
Universities, hospitals,
government agencies
Personal and inter-organizational contacts;
telephone calls
Telephone calls with guided script
>10
>50
All required
units
obtained
Several leads
Large R&R Projects'11
Government agencies
and trade unions
Universities, hospitals,
contractors
Workers/contractors
Personal and inter-organizational contacts
Telephone calls with guided script and
follow-up letters
Letters sent to locals of United Brotherhood
of Carpenters
>20
>400
>2000
letters
No success
for non-
abatement
work
No success
No success
1 This study phase could not be accomplished in the time frame of this study.
5-8
-------
In contrast, the carpet removal workers in the four St. Louis, MO, area units were all
nonprofessional residents or helpers. The residents usually had no previous awareness of
potential lead exposure in performing carpet removal, and they only had informal previous
experience in performing R&R activities. Only two resident-workers in one unit identified
themselves as ex-R&R professionals or self-employed professionals, with less than 15% of their
experience in carpet removal and no experience in lead decontamination areas. Therefore, the St.
Louis workers were less experienced in performing carpet removal activity and in handling
potential lead contamination resulting from the activity.
Window Replacement
Workers participating in the window replacement phase were professional contractors
experienced in window replacement. The workers in three of the four study units claimed that
window replacement was the only R&R activity they performed, while the workers in the fourth
unit claimed that window replacement activities constituted 50% of their R&R work. The R&R
experience of the workers ranged from 1.5 to 27 years. Window replacement experience ranged
from 1.5 to 16 years.
None of the workers in the window replacement phase had any experience in abatement
work. Only two workers at one unit claimed to be aware of any potential risk for lead exposures
during window replacement; none of the workers took any precautions to reduce lead exposure.
CEP Phase
Unlike the carpet removal and window replacement phases, contractors in the CED phase
were hired by the EFSS study team to perform the specified activities at a study unit. All workers
in the CED phase were professional lead abatement workers. Full worker precautions were
taken, but no effort was made to reduce dust generated by the activity. The contractors were
instructed to do the assigned jobs in a manner typical of general R&R work in an unregulated
environment.
5.2.2 Unit Characteristics
All eight study units in the carpet removal phase were occupied, single-family residences.
Carpet removal was performed in a portion of a given unit. Two of the four units in the window
replacement phase were occupied single-family residences. The other two units were occupied,
but the floors on which the window replacement activity took place had been vacant and
neglected for some time. The CED phase included two vacant row houses in Baltimore,
Maryland, and four single-family homes in Denver, Colorado. All units in the CED phase were
scheduled for demolition, abatement, or gutting/restoration in the near future.
The study units had varying degrees of lead paint contamination and differed in their
overall observed levels of cleanliness. In addition, the amount and type of dust generated during a
specific R&R activity often differed from one unit to another. For example, workers in one unit
of the carpet removal phase reported excessive dust generation from disintegrated carpet padding,
5-9
-------
while carpet in another unit contained considerable surface debris. The discussion of results for
each phase in Chapter 8 presents additional details and demographic information on the study
units.
5.2.2.1 Comparability with Units from Previous Studies
This section compares the pre-activity lead levels of the recruited study units in the carpet
removal and window replacement phases to the pre-activity lead levels in other studies. Table 5-2
contains summaries of dust-lead loadings across EFSS units as well as for units from other
studies. Lead loadings are from dust samples collected from floors, window sills, and window
wells. Besides the carpet removal and window replacement phases, the following studies are
represented in Table 5-2:
• The HUD Abatement Demonstration Study (U.S. Dept. of HUD; 1990)
• The HUD National Survey of Lead-Based Paint in Housing (U.S. EPA, 1995)
• Two Kennedy-Krieger studies on abatement/maintenance methods (Farfel and
Chisolm, 1990; Farfel and Chisolm, 1991).
The HUD Abatement Demonstration ("HUD Demo") Study assessed the costs and short-
term efficacy of alternative methods of lead-based paint abatement. A total of 172 dwelling units
in seven metropolitan areas were abated in the FHA portion of the HUD Demo study. The HUD
Demo statistics in Table 5-2 represent loadings from wipe samples taken in these units as part of
post-abatement clearance procedures.
The HUD National Survey estimated the incidence of lead-based paint in U.S. public and
privately-owned housing. Table 5-2 includes statistics on lead loadings from 182 privately-owned
units from the National Survey, partitioned into two groups according to lead levels from XRF
testing.
The Kennedy-Krieger Institute studies examined the effect of different types of
abatement/maintenance methods on dust and blood lead levels. In the first study, 53 units were
earmarked for traditional abatement practices and 18 for modified abatement practices. These
units had multiple interior surfaces with lead-based paint and at least one child with elevated
blood lead levels. The second study considered experimental abatement practices in six older row
homes with multiple lead-based paint hazards. In Table 5-2, the geometric means of pre-
abatement lead loadings for the Kennedy-Krieger studies are summarized for three groups of units
in the following order: traditional practice units, modified practice units, and experimental
practice units.
5-10
-------
Table 5-2. Descriptive Statistics for Pre-Activity Lead Loadings on Floors, Window Sills,
and Window Wells in the EFSS, with Comparison to Levels Observed in
Previous Abatement Studies
Study111
# Samples
Log Std.
Dev.
25th Percentile
(jug /ft2)
Geometric Mean
(jug /ft2)
75th Percentile
(jug /ft2)
Floor Dust Lead Loadings
EFSS Carpet Removal Phase
EFSS Window Replacement Phase
HUD Demonstration
National Survey121
High XRF
Low XRF
Kennedy-Krieger studies131
32
39
1026
234
304
280
82
70
1.91
3.26
1.53
1.82
1.61
2.61
28.4
23.6
0.70
0.22
14.7
637
66.0
2.40
0.64
251
288
520
49.0
9,120
185
8.23
1.91
Window Sill Lead Loadings
EFSS Carpet Removal Phase
HUD Demonstration
National Survey121
High XRF
Low XRF
Kennedy-Krieger studies131
16
783
123
126
249
95
34
2.32
1.79
2.64
2.13
88.8
26.7
1.42
0.37
418.0
89.10
8.40
1.57
1,340
1,800
4,610
2740
297
49.7
6.59
Window Well Lead Loadings
EFSS Window Replacement Phase
HUD Demonstration
National Survey121
High XRF
Low XRF
Kennedy-Krieger studies'31
11
756
56
38
150
37
28
0.94
1.93
2.28
2.46
54,700
138
47.4
3.27
135,000
506
220
17.2
15,500
18,300
29,400
296,000
1,860
1,020
90.4
111 Data for the EFSS phases reflect all regular and QC side-by-side samples collected prior to R&R activity.
HUD Demonstration data represent post-activity clearance results for units in the FHA portion of the study.
121 "High XRF" entries reflect readings for units with at least one interior and one exterior XRF reading
exceeding 10 mg/cm2. "Low XRF" entries reflect readings for units where all XRF readings were below 1.0
g/cm2.
131 The first two entries in each cell reflect pre-abatement readings from homes where traditional and modified
abatement procedures, respectively, were to be employed (Farfel and Chisolm, 1990). The third entry in
each cell reflects pre-abatement readings in units from Farfel and Chisolm, 1991.
5-11
-------
Compared to units in the HUD Demo study, the units in the carpet removal phase had
generally lower lead loadings in floor dust, but higher lead loadings in window sills. The units in
the window replacement phase had higher lead loadings in floor dust, and much higher lead
loadings in window wells. It should be noted, however, that all floor samples collected in the
window replacement phase were within 6 feet of a window, possibly accounting for the high lead
loadings from floors. Also, the data for the HUD Demo study primarily represented post-activity
sampling after cleanup procedures.
The two EFSS study phases had considerably higher lead loadings than National Survey
units. However, the "blue nozzle" vacuum sampler used to collect dust samples in the National
Survey was shown to be approximately five times less efficient than the wipe and vacuum
sampling techniques used in the HUD Demo and the EFSS. Consequently, reported results may
be lower than what would have been observed using more efficient sampling techniques.
The Kennedy-Krieger results are higher than all other study results except those from the
window replacement phase.
In general, the variability observed was similar across all studies. From Table 5-2 one can
conclude that the observed lead levels in carpet removal units were not atypical of inhabited units
with lead contamination. Units included in the window replacement phase, on the other hand, had
relatively high levels of pre-activity lead contamination, especially in the window wells. These
levels were high in both the vacant and occupied activity areas.
5-12
-------
6.0 DATA ANALYSIS OVERVIEW
As introduced in Chapter 1, the technical objectives of the R&R study were to characterize
lead exposure to workers conducting R&R activities in contaminated housing, investigate how
these workers disturb lead and create a lead-based paint hazard to occupants or other exposed
individuals, and provide technical input and recommendations for developing guidelines and
regulations on lead exposures resulting from R&R activities. This chapter presents an overview
of those data analysis objectives and approaches that were consistent across the EFSS phases and
that addressed these technical objectives.
6.1 DATA ANALYSIS OBJECTIVES
Data summaries and statistical analyses were centered around data analysis objectives
established for the EFSS. The data analysis objectives address the overall objectives of the EFSS
(presented in Chapter 1). The study phases shared many of the same data analysis objectives.
These objectives are categorized below according to the goal which they satisfy.
A. Characterize Lead Disturbance and Potential Lead Exposure
• Characterize average personal exposure levels of airborne lead for workers over the
duration of a given R&R activity and determine if these levels exceed 50 |ig/m3 (the
OSHA Permissible Exposure Limit (PEL))
• Characterize lead loadings and concentrations within samples of dust that settles on
flat surfaces (e.g., floors, window sills) within a specified period during and following
completion of R&R activities.
Carpet removal and window replacement phases only:
• Characterize airborne dust lead levels in rooms adjacent to "activity rooms" (rooms
in which the R&R activity occurs) while the activity is in progress.
In a separate evaluation:
• Determine the effect that typical cleanup procedures have on lead loadings in settled
dust that remains following the activity as a lead exposure to occupants.
B. Assess Factors or Measurements Related to Lead Disturbance
• Characterize the extent to which lead disturbance and exposure are affected by
indicators of lead contamination in the dwelling unit, extent of exposure to the
activity, and building characteristics. (Depending on the study phase, indicators
include lead loadings in settled dust existing prior to the start of activity, available
lead loadings on painted surfaces disrupted by the activity, distance of sample
location from the activity, type of substrate disturbed, and elapsed time in which the
activity took place.)
6-1
-------
• Quantify the components of total variability in lead measurements taken within
samples of various media. (Variance components could constitute variability
attributable to sampling across different dwelling units, sampling from different
locations or workers within the same unit, or sampling from different areas within the
same location.)
The following data analysis objectives, grouped as a third category, were devised to provide
information to aid in the design and development of future lead exposure studies:
C. Evaluate Different Media and Methods as Indicators of Lead Exposure
Carpet removal and window replacement phases only:
• Compare settled dust lead loadings on stainless steel dustfall collectors (SSDCs)
introduced prior to the start of activity with baseline-adjusted settled dust lead
loadings directly from the floor surface
• Compare lead loadings from settled dust samples collected one hour following
completion of activity with loadings from samples collected at two hours following
completion of activity
• Compare lead loadings from settled dust samples collected using vacuum techniques
with loadings for samples collected using wipe techniques
• Examine correlations between lead measurements among different collection media.
CED phase only:
• Investigate the potential for cross-contamination between activities.
Data analyses addressing the first three objectives within Category C provide information to
evaluate competing sampling methodologies as presented and analyzed in Chapter 7. The
remaining objectives are addressed by the analyses presented in Chapter 8 for each R&R activity.
6.2 APPROACH TO DATA ANALYSIS
Most of the data analysis objectives and the approaches taken to analyzing the data were
consistent across the study phases. Data analysis in the EFSS was centered on characterizing the
distributions of lead levels in various media affected by each of the target R&R activities.
Variability in lead measurements was characterized as a function of sampling from different units,
from different locations/workers with a unit, and from different areas within the same location.
The extent of correlation present between lead loadings in settled dust and personal/ambient air
lead levels, and between lead exposure and building and activity characteristics, was investigated.
All statistical analyses and data summaries were performed using Versions 6.03 and 6.08 of the
SAS® System.
6-2
-------
Lead levels as determined from analytical methods performed on the environmental field
samples were expressed in terms of loadings and/or concentrations. For personal air and ambient
air cassette samples, lead levels were expressed as concentration of lead per cubic meter of air
sampled (|ig/m3). For each settled dust sample, lead loadings expressed the amount of lead (jig)
per unit area (square foot) sampled. For the purposes of this study, dust lead loading provides a
measure of the total amount of lead disturbed in a specified area by the studied R&R activity. For
vacuum dust samples, lead concentration was also reported, representing lead amounts (jig) per
unit weight (g) of dust. However, while lead concentration in dust provides information to
characterize the nature of the exposure and to provide insight into appropriate avoidance or
cleanup practices, concentration data provide no indication of the total amount of lead disturbed
by the activity. Because the primary objective of the study was to characterize the amount of lead
disturbed and the potential human exposure associated with an activity, loadings were emphasized
in the statistical evaluation of lead disturbance in each phase of the EFSS. For settled dust
samples, statistical comparisons and modeling were performed on loadings, while only data
summaries were performed on concentrations and physical sample weights. For personal and
ambient air samples, statistical analysis was performed on the reported lead concentrations.
Personal exposure monitoring during R&R activity provides data on the amount of lead per
volume of air sampled within the breathing zone of a monitored worker. These data represent
average exposures over the duration of an activity period and are referred to in this document as
task-length averages (TLA) (in jig lead per cubic meter of air). Exposure data are calculated from
the amount of lead collected within the sample cassette(s), the duration of monitoring, and the air
pump flow rate. The TLA for a worker performing a given R&R activity at a study unit is
calculated as follows:
Tl . _ total lead across all cassettes at a unit for a worker
(duration of task)-(flow rate)
This approach to calculating the TLA for personal exposure monitoring was followed in each
study phase of the EFSS.
6.2.1 Common Approaches Taken Within Each Study Phase
Log Transformations
Throughout the EFSS, analysis of lead loading or concentration data was performed after
taking a natural-logarithm (or "log") transformation of the data. In many standard statistical
inference procedures, the assumption of normally distributed data is necessary to make accurate
inferences on the central tendency and variability within the data distribution. In this setting, lead
loadings and concentrations appeared to originate from skewed distributions. The arithmetic
mean is heavily influenced by large data values in the upper tail of the distribution, and therefore
may not be an appropriate representation of the central tendency of the data. However,
distributions of the log-transformed data more closely resembled a normal distribution (as
determined through normal probability plots and results of the Shapiro-Wilk test for normality).
6-3
-------
In such a situation, if the arithmetic mean of the log-transformed data is calculated and then
exponentiated, the result (called the "geometric mean") is an estimate of the median of the
distribution of untransformed data. Thus, due to lognormality, the geometric mean is an estimate
of the central tendency of the distribution and is more robust to outliers than the arithmetic mean.
As a result, summaries of lead loadings and concentrations within each phase are based on
geometric means and on variability associated with log-transformed data. These summaries,
which also include minimum and maximum observed data values, are presented in tabular form
within Appendix A.
Pearson Correlation Coefficients
The Pearson correlation coefficient was calculated to quantify the extent to which linear
relationships were present between a pair of measurements. Values of the correlation coefficient
close to one indicates a positive correlation, while values near -1 represent negative correlation.
Statistical tests for nonzero correlation were executed to determine whether the coefficient differs
significantly from zero at the 0.05 level.
Statistical Modeling
Statistical models were fitted to lead levels in various media to evaluate the effects of
external factors (called covariates) on these data. For example, models within the carpet removal
phase included the effects of pre-activity lead loadings in the carpet and on the floors. F-tests
were performed to determine whether specific covariates had statistically significant effects on the
modeled data at the 0.05 level. Statistical models were also fit that included exclusively random
effects, such as a dwelling unit effect, to determine the contribution to overall variability in the
study data which could be explained by these effects. Further detail on the approach to statistical
modeling is presented in Appendix C. In some instances, small numbers of data points restricted
the extent to which model effects could be characterized within statistical models.
Fitting linear statistical models to log-transformed data affects how the modeling results are
interpreted relative to the untransformed data. In this study, the modeled data (i.e., log-
transformed data) are expressed as a linear combination of fixed and random factors. Therefore,
the models are considered "additive" models relative to the log-transformed data. However, by
exponentiating both sides of the model equation, one sees that the models express the
untransformed data as a multiplicative function of the model effects. That is, the effect of a single
factor is measured by its multiplicative influence on the untransformed data, and comparisons
among factor effects are expressed in terms of ratios of geometric means rather than as
differences in arithmetic means.
Boxplots
In each study phase, results of data analysis were summarized in tabular and graphical form.
Graphical displays were used to ease interpretation and comparison among sample types. Two-
dimensional scatterplots illustrate correlations present among pairs of measurements and portray
6-4
-------
data values across the study by study unit and location within unit. Boxplots display percentiles
and other statistics, as well as extreme data points, in the form presented in Figure 6-1.
1000-
100-
10-
Figure 6-1. Example of a Boxplot
The lower and upper limits of the "box" portion of the boxplot represent the 25th and 75th
percentiles, respectively, of the observed data distribution. The length of the box represents the
data's interquartile range (IQR), or the difference between the 25th and 75th percentiles, which is
an indicator of data variability. The horizontal line within the box is the 50th percentile, or
median. The diamond symbol is the arithmetic mean. Vertical lines extend from the top (or
bottom) of the box to the value of the most extreme data point which falls within 1.5 IQRs from
the box. Each data point extending from 1.5 to 3.0 IQRs from the box is plotted with a "+"
symbol, while each data point extending beyond 3.0 IQRs from the box is plotted with a "*"
symbol.
6.2.2 Estimating Lead Disturbance in a 6' x 1' Gradient Region
In this study, a statistic was developed to characterize lead disturbance (and potential
exposure to occupants) associated with a specific R&R activity. Taking into account the
relationship between the amount of lead in the environment and the distance from the activity, this
statistic estimated the amount of lead exposed (or disturbed) along a 6-foot line (lead gradient)
emanating from the activity source. For each activity, a statistical model (Lead = aeB'dlstance) was
fitted to express lead levels in settled dust as a function of distance from activity. Details on the
model fitting process are provided in Section C.5 of Appendix C. An example of the fitted model
is given in Figure 6-2.
6-5
-------
From this model, lead exposure for an activity is quantified by calculating the area under the
fitted curve from 0 to 6 feet from the activity. This measure represents an estimate of the amount
of lead present in a 6-foot by 1-foot column extending from the activity (Figure 6-2). Throughout
the report, this measure will be referred to as the amount of lead in a 6' x 1' gradient region. The
6-foot distance was selected based on distance considered in the sampling design. For each
activity, uncertainty in the amount of lead within a 6-foot by 1-foot column was expressed by
calculating an approximate 95% confidence interval on the estimated amount.
Distance (ft)
Task
Figure 6-2. Portrayal of Model Predicting Lead Exposure as a Function of Distance from
R&R Activity
Two alternative calculations to characterizing lead disturbance and potential exposure to
occupants were considered but not implemented in this study: 1) a weighted combination of
estimated lead amounts at one and six feet from the activity, and 2) total volume of lead exposed
within a hypothetical (6 by 12 foot) area adjacent to the activity. The development of a
weighting scheme for the 1- and 6-foot estimates involved assumptions about potential cleanup
activities and occupant behavior. For example, is a greater exposure potential posed by lead six
feet away from the activity than lead one foot away? Given the limited existing information upon
which to base such assumptions, this calculation was rejected. The total volume measure would
have provided an interesting measure of lead exposure but required assumptions about the
character of the dust dispersion resulting from the activity. For example, would dust generated by
sawing through plaster disperse symmetrically from the cutting area? What would be the shape of
the response surface that would characterize dispersion over a two-dimensional area? Since the
6-6
-------
EFSS was not designed to assess such questions, these assumptions would have been based on
field experiences and mathematical convenience, and therefore this estimation scheme was
rejected.
6.2.3 Significant Digits
Laboratory results from the chemical analyses of samples collected in this study were
reported to five significant digits. However, the precision of most analytical methods employed in
the analyses would suggest that the analytical data have less than five significant digits.
Therefore, all results in this report are reported to three significant digits. Descriptive statistics
and results reported in tables in the Appendices present all digits as calculated.
6-7
-------
7.0 METHODOLOGY ISSUES AND RESULTS
As discussed in Section 4.3, the method used in the EFSS for obtaining settled dust samples
from floors differs from the standard "pre- versus post-activity sample" method taken in previous
studies. In the EFSS, stainless steel dustfall collectors (SSDCs) were placed on the floor
immediately before the start of an activity, and a single dust sample was collected from each
SSDC at specified time points following conclusion of the activity. An investigation was
conducted to evaluate the different approaches to collecting dust samples, relative to their ability
to recover and estimate lead loadings in settled dust. This chapter presents the findings of this
evaluation.
Section 7.1 compares lead loadings from SSDC dust samples versus those from paired dust
samples taken via the standard method of sampling directly from the floor surface (post- minus
pre-activity). Section 7.2 evaluates the amount of time necessary to allow lead-dust to settle
following conclusion of the activity before sample collection. Lead loadings of SSDC samples
collected via vacuum techniques are compared to those collected via wipe techniques in Section
7.3. Finally, Section 7.4 addresses differences in lead loadings between adjacent ("side-by-side")
samples, and how these differences may indicate whether taking the first sample unduly influences
the outcome associated with the second sample.
7.1 STAINLESS STEEL DUSTFALL COLLECTORS VS. POST- MINUS PRE-ACTIVITY FLOOR
DUST METHOD
In the carpet removal and window replacement phases of the EFSS, two competing
approaches to measuring the lead disturbed by the R&R activity were compared. Both
approaches yielded estimates of lead levels in dust that settled on floors during the activity and up
to one hour following the activity's completion. The protocols for the sampling approaches, both
employing vacuum dust sampling techniques, were as follows:
• Post- minus pre-activity floor: Collect two samples directly from the floor surface,
where one sample was taken immediately prior to start of the activity (baseline) and
the other was taken at one hour following the activity's conclusion from an area
adjoining the first sample area. Dust lead levels are estimated by subtracting the pre-
activity result from the post-activity result.
• Stainless steel dustfall collectors (SSDC): Place a clean, uncontaminated SSDC on
the floor surface immediately prior to start of the activity, and collect one dust
sample from the SSDC surface at one hour following the activity.
The first sampling approach (floor dust) has been taken in a number of previous programs to
measure settled dust levels resulting from some activity. The second approach (the SSDC)
addresses the same objective, but does not require an adjustment for a baseline level, thereby
using only one dust sample rather than two to estimate lead levels. For this reason, it was
expected that variability attributable to spatial trends and biases, sampling, and analysis would be
reduced by use of the SSDCs.
7-1
-------
In both the carpet removal and window replacement phases, each sampling location
containing a SSDC for post-activity dustfall sampling also contained two one-square-foot areas
for pre- and post-activity floor dust sampling. Plots of the SSDC lead loadings versus the post-
minus pre-activity floor dust loadings at the same location indicate that loadings from the post-
minus pre-activity floor dust method cover a much larger range than the SSDC lead loadings.
The plot for carpet removal is shown in Figure 7-1. Figures 7-2 and 7-3 show the plot for
window replacement samples collected at zero feet and six feet from the window, respectively.
Figures 7-1 and 7-3 show the large discrepancy in the variability between SSDC and floor
samples. The range of post- minus pre-activity floor loadings required the horizontal axes of
these plots to be broken to more clearly illustrate all the data.
-200 0 200 400 600 800 3000 5000 7000
Post-Pre Floor Loading Difference G^g/ft2)
SSDC = Stainless Steel Dustfall Collector
Figure 7-1. Lead Loadings (//g/ft2) for 1-Hour Stainless Steel Dustfall Collector Samples
Versus the Difference in Loadings Between Post-Activity and Pre-Activity Floor
Dust Samples in the Carpet Removal Phase
Additional plots of the observed SSDC sample loadings and post- minus pre-activity floor
loadings are found in Figure 7-4 for carpet removal and Figures 7-5a and 7-5b for window
replacement. These plots illustrate the difference in magnitude of variability between the two
sampling methods on a unit-by-unit basis. It should be noted that the variability observed in the
EFSS for side-by-side floor samples is similar to what has been observed in previous studies.
EPA's Comprehensive Abatement Performance Study (CAPS) (Battelle, 1994) estimated the log
standard deviation of for floor side-by-side sample loadings to be 0.93, similar to that observed in
the EFSS. NIOSH's Ohio University study reported coefficients of variation over 50 percent for
side-by-side wipe samples, again indicating high expected variability (NIOSH, 1993). Moreover,
from a statistical standpoint, it is expected that variability would be reduced when using the SSDC
methodology, since the SSDC method involves the variability of a single measurement
7-2
-------
400000
-300000
1
g
=5 200000
100000
Sfope=1
-500000 -100000 0 20000 40000 60000 80000 100000
Floor Loading Difference
Figure 7-2. Lead Loadings (//g/ft2) for 1-Hour Stainless Steel Dustfall Collector Samples
Versus the Difference in Loadings Between Post-Activity and Pre-Activity Floor
Samples Taken Zero Feet from the Windows in the Window Replacement Phase
5000
4000
3000
2000
1000
Slope=1
-500
500 1000 1500 5000 25000 45000
Floor Loading Difference
Figure 7-3. Lead Loadings (//g/ft2) for 1-Hour Stainless Steel Dustfall Collector Samples
Versus the Difference in Loadings Between Post-Activity and Pre-Activity Floor
Samples Taken Six Feet from the Windows in the Window Replacement Phase
7-3
-------
E-
%
a
10000.0
1000.0
100.0
10.0
1.0
0.1
: : : # : : : :
! ! ! o ! » ! ! !
o* ; ; * ; #* ; o ;*•;»»;
»o ; ; ; o° ; ; ° ; ;
» ; * ; ; ; * ; ; ;
o :° !oo! !°o!* ;o°; °
; *o ; ; ; ; ; ; oo
; ; * ! ! ; o o ; ;
; 4 ; ; :^: ^»^»»
1-01 1-02 1-03 1-04 2-01 2-02 2-03 2-05
Unit ID
o o o 1-Hour Vacuum Samples
from SSDC
Post-Pie Difference in
Floor Vacuum Samples
Note: Five results plotted at a value of 0.1 are actually negative.
SSDC = Stainless Steel Dustfall Collectors
Figure 7-4. Lead Loadings (//g/ft2) for 1-Hour Stainless Steel Dustfall Collector Samples, and
the Difference in Loadings Between Post- and Pre-Activity Floor Dust Samples in
the Carpet Removal Phase
•s
3-
f
1000000.0
100000.0
10000.0
1000.0
100.0
10.0
1.0
0.1
: : O : :
••;**; ; a ;
; ° ° ; ; ;
o ° ; ; ; ;
: O '- O O
1 Post-Pre Floor Vacuum (d=C
Unit
' • 1-Hour Vacuum (SSDC) (d=
Note: Results plotted at a value of 0.1 are actually negative.
SSDC = Stainless Steal Dustfall Collectors
Figure 7-5a. Lead Loadings (//g/ft2) for 1-Hour Stainless Steel Dustfall Collector Samples,
and the Difference in Loadings Between Post- and Pre-Activity Floor Samples,
Collected at Zero Feet from the Windows in the Window Replacement Phase
7-4
-------
1
%s
g>
I
100000.0
10000.0
1000.0
100.0
10.0
1.0
0.1
: : : o :
; ; o ; ° ;
; ; 8 ;
: : • : :
° ! 8 ! . ! !
8 ° . I I • I I
: O O '- O :- :-
' Post-Pre Floor Vacuum (d=6)
Unit
• 1-Hour Vacuum (SSDC) (d=
Note: Results plotted at a value of 0.1 are actually negative.
SSDC = Stainless Steel Dustfall Collectors
Figure 7-5b. Lead Loadings (//g/ft2) for 1-Hour Stainless Steel Dustfall Collector Samples,
and the Difference in Loadings Between Post- and Pre-Activity Floor Samples,
Collected at Six Feet from the Windows in the Window Replacement Phase
rather than the additive variability of a linear combination of two measurements. The SSDC also
eliminates the possibility of negative values (post minus pre) that can occur with side-by-side
samples but are not reasonable estimates of the amount of lead disturbed by an activity.
Preference for using the SSDC approach over the post- minus pre-activity floor approach is
further supported by the high linear correlation between the lead loadings from SSDC dust
samples and the lead concentrations from worker personal air samples. A correlation of 0.98 was
observed between these two measurements in the carpet removal phase, and a correlation of 0.94
was observed in the window replacement phase. Both correlations were higher than the
corresponding correlations between personal exposure concentrations and post-activity floor
loadings. These findings, presented in more detail in Chapter 8, support the hypothesis that the
SSDC approach yields a better measure of the lead disturbed by the R&R activity than the post-
minus pre-activity floor approach.
In summary, the SSDC methodology was judged to be superior to the post- minus pre-
activity floor sample approach for the following reasons:
1. The SSDC method exhibited less variability, consistent with expectations from
previous experience and statistical theory.
2. The SSDC method does not produce negative estimates of the amount of lead
disturbed by an activity.
7-5
-------
For these reasons, a decision was made to base estimates of the amount of lead disturbed by a
specific R&R activity on the results of samples taken from SSDCs. As a result of this decision,
only the SSDC method for estimating the lead generated by an R&R activity was used in the third
phase of the EFSS, the Controlled Experimentally-Designed (CED) phase.
7.2 STAINLESS STEEL DUSTFALL COLLECTORS: COLLECTING SAMPLES 1 HOUR VS. 2
HOURS AFTER R&R ACTIVITY
When defining the post-activity settled dust sampling approach, the study team had limited
information available in assigning a wait time to allow sufficient fallout of lead dust prior to
sample collection. Preliminary calculations based on particle physics indicated that roughly 90%
of lead particles greater than five jim that are disturbed as a result of R&R activity should settle
within one hour, even in turbulent air situations. As a result, the one-hour post-activity settled
dust sample from SSDCs was chosen as a reasonable indicator of lead disturbance in the R&R
study. However, it was of interest to test whether a one-hour wait was adequate, or whether
allowing a longer period of time for dust to settle would result in significantly higher lead
loadings. Therefore, at selected floor locations in the carpet removal and window replacement
phases, two SSDCs were placed side-by-side. One SSDC was sampled at one-hour post-activity,
and the other at two-hours post-activity. The lead loadings between the side-by-side samples
were compared to determine the magnitude of additional dustfall that occurs over a period of two
hours compared to one hour.
Figure 7-6 presents a scatterplot of the difference in loadings between the paired two-hour
and one-hour samples versus the difference in elapsed wait time between the two samples. Due to
the pace of the field activity and the number of samples collected, samples could not be taken
exactly at the time they were scheduled following the activity. At a given location, the difference
in elapsed wait times between the one-hour and two-hour samples ranged from 20 to 72 minutes;
most time differences ranged from 40 to 60 minutes. Figure 7-6 indicates no apparent increasing
trend in the differences in lead loadings as the difference in elapsed wait time increases (fitted
regression lines were not significantly different from the horizontal line at difference = 0). In fact,
the differences in sample results for three of the four sample pairs with the largest elapsed wait
times are close to zero. However, the plot does show that in 24 of the 36 paired samples (67%)
the two-hour sample loading is larger than the one-hour sample loading. In only one sample pair
does the one-hour loading exceed the two-hour loading by greater than 50 |ig/ft2, while the two-
hour loading exceeds the one-hour loading by greater than 50 |ig/ft2 for ten pairs. Since an effect
of the exact sampling time cannot be detected, a simple categorical classification of results into
one-hour and two-hour samples was maintained.
Figure 7-7 is a log-log scatterplot showing the relationship between the paired two-hour
and one-hour lead loadings among the study units in the carpet removal and window replacement
phases. The solid line within the plot corresponds to equality between the two responses. This
plot further illustrates that two-thirds of the sample pairs had higher lead loadings for the two-
hour sample compared with the adjoining one-hour sample. This outcome supports the
hypothesis that loadings tend to increase during the additional hour wait. Based on the analysis of
pre/post floor dust samples, one might anticipate that much of the difference between the
adjoining one-hour and two-hour sample results is due to spatial variability. However, a strong
7-6
-------
16500
16000
400
300
200
100
0
-100
-200
~c e ,-c
15202530354045505560657075
Difference (min) in Elapsed Wait Times
Figure 7-6. Difference in Loadings Between Two-Hour and One-Hour SSDC Samples (Two-
Hour Minus One-Hour) Versus the Difference in Elapsed Wait Times Between
the Two Samples, for Carpet Removal (c) and Window Replacement (w) Phases
100000
10000
1000
100
10
Sbpe=1
10 100 1000
1-Hour SSDC Loading
SSDC = Stainless Steel Dustfall Collector
10000
100000
Figure 7-7. Two-Hour SSDC Lead Loadings Versus One-Hour SSDC Loadings for the Carpet
Removal (c) and Window Replacement (w) Phases
7-7
-------
linear relationship, though not centered around the line of equality, was observed between the two
results in Figure 7-7. The Pearson correlation was 0.85 between the one- and two-hour lead
loading results (after taking logarithms). As a result, the extent of differences between adjoining
one- and two-hour lead loadings appears to be consistent across the range of observed data, with
higher readings associated with the two-hour results.
To further characterize the statistical relationship between the one-hour and two-hour lead
loadings from adjoining sample areas, a paired t-test was performed to determine whether the
difference in the log-transformed loadings at a location was significantly different from zero (or
equivalently, that the ratio of untransformed loadings at a location was significantly different from
one). Data were pooled across the carpet removal and window replacement phases for this test.
The p-value resulting from the paired t-test was 0.008, indicating that the difference in the log
lead loadings between one- and two-hour lead loadings was highly significant statistically. Similar
results occurred when the Wilcoxon signed-rank nonparametric test was applied. These results
are equivalent to concluding that the geometric mean of the ratio of 2-hour to 1-hour lead
loadings is significantly different from one. The estimated geometric mean is 1.57, which is an
estimate of the median ratio. Therefore, there is statistical evidence that waiting two hours after
an activity leads to increased lead loadings. Since statistical comparisons between activities within
the EFSS were based only on one-hour dustfall measurements (Chapters 8 and 9), effects due to
elapsed wait times do not enter into the comparisons.
7.3 COLLECTING SAMPLES VIA VACUUM TECHNIQUES VS. WIPE TECHNIQUES
For a given R&R activity, lead loadings from dustfall samples collected via vacuum
techniques were used to statistically characterize lead disturbance and potential lead exposure to
occupants. Using pre-moistened wipes to collect settled dust samples is a competing sampling
collection procedure that was employed in previous lead exposure characterization studies. The
vacuum and wipe sampling techniques from SSDCs were compared within the EFSS by analyzing
data from paired dust samples, where one sample in each pair was collected from an SSDC using
wipe techniques and the other from an adjoining SSDC using vacuum techniques. Paired
vacuum/wipe sampling was performed for carpet removal and for the demolition of plaster walls
within the CED phase. Differences between paired vacuum and wipe results were statistically
analyzed across the two activities.
In the carpet removal phase, two pairs of vacuum/wipe samples were collected within each
of the eight study units, providing 16 sample pairs. All 32 samples were collected from locations
immediately adjacent to rooms where carpet removal activity occurred. In the CED phase, 12
pairs of vacuum/wipe samples were collected following wall demolition within the two Baltimore
units. These pairs were collected at distances ranging from 6 to 10 feet from the demolished wall
one hour following conclusion of the R&R activity.
7-E
-------
The scatterplot in Figure 7-8 includes lead loadings for dust samples within the
wipe/vacuum sample pairs in the carpet removal and CED phases. Results from the same pair are
plotted immediately above and below each other. The extent of a linear relationship between the
(log-transformed) wipe and vacuum sample results within a pair is illustrated in Figure 7-9, where
wipe loadings are plotted versus vacuum loadings.
10000
1000
100
10
* *°
o
O : * : O
: : *
; o ;
1-01 1-02 1-03 1-04 2-01 2-02 2-03 2-05 CBJ-1 CED-2
Unit ID
ooo vacuum Samples * * * Wipe Samples
Figure 7-8. Plot of Lead Loadings for Paired Vacuum/Wipe Samples at Each Location and
Study Unit in the EFSS
The correlation coefficient between (log-transformed) vacuum and wipe sample results in
Figure 7-9 is 0.83, which is highly significantly different from zero (p=0.0001). Thus, as
expected, a strong positive relationship exists between log-transformed vacuum and wipe loadings
at a given location.
For the carpet removal phase, both wipe and vacuum loadings appear to have similar
variability and cover approximately the same ranges (the stars in Figure 7-9). However, in the
CED phase, variability among vacuum sample loadings was substantially higher than for wipe
sample loadings. This is evident in Figure 7-9, where the CED data (the circles) cover a wider
range horizontally than vertically. This result may indicate that wipe sampling is less sensitive
than vacuum sampling in environments with considerable amounts of dust, especially in situations
where the wipes may become saturated with dust.
7-9
-------
10000
1000
Slope=1
100
* *,
10
1
1 10 100 1000 10000
Vacuum Loading (/u.g/Ft )
*** Carpet Removal Phase ooo CED Phase
Figure 7-9. Plot of Wipe Sample Loading Versus Vacuum Sample Loading for Paired
Samples in the EFSS
The solid line in Figure 7-9 represents the line of perfect agreement between wipe and
vacuum sample results. In the carpet removal phase, 11 of the 16 pairs (69%) had higher loadings
for the wipe sample than for the vacuum sample. For 10 of these 11 pairs, the vacuum sample
loadings were less than 30 |ig/ft2. In the CED phase, 6 of 12 pairs had higher results for the wipe
sample, as vacuum samples were more highly variable in both directions from the median of the
wipe loading data. Generally, vacuum and wipe sample loadings were higher in the CED phase
than the carpet removal phase, as seen in Figure 7-8. This figure also suggests that wipe methods
may provide higher recoveries at low loadings, whereas vacuum methods may provide higher
recoveries at high loadings.
To statistically characterize the relationship in dust lead loadings between adjoining wipe
and vacuum samples collected one hour post-activity, a statistical model was fitted to the data in
Figure 7-8 (data from 28 vacuum/wipe pairs). The statistical model assumes that this relationship
is log-linear:
Indoading
) = ln(a) + pindoading ) + error
where a and P are unknown parameters. Statistical methods took into account that loadings for
both sample types are subject to error. If a and P both equal one, then the log-loading for the
vacuum sample differs from the log-loading for the adjoining wipe sample by random error. If a
is not equal to one, then the vacuum log-loading is consistently different from the wipe log-
loading by an amount equal to ln(a). If P is not equal to one, the extent of the difference between
the two results changes with the magnitude of the measurements.
7-10
-------
The estimate of P in the above model was 1.04, with a standard error of 0.14. Thus, P was
declared not significantly different from one (p=0.79), concluding that any difference between the
wipe and vacuum loadings was assumed constant across the observed measurements. The
estimate of a was 0.88, with an approximate 95% confidence interval of (0.57, 1.35). This
estimate is less than one, reflecting the decreased vacuum sample loadings relative to the wipe
sample loadings, as seen primarily in the carpet removal data. However, its 95% confidence
interval contains 1.0, implying that the estimate is not significantly different from one at the 0.05
level. Therefore, there is no statistical evidence (given the observed data) that the results for
paired wipe and vacuum samples differ significantly.
7.4 COMPARING RESULTS BETWEEN SIDE-BY-SIDE SAMPLES
In each of the three R&R phases (carpet removal, window replacement, and CED phases),
additional lead dust samples were collected from areas adjoining "regular" sample areas at
selected sampling locations. These samples, simulating field duplicates, were labeled "side-by-
side" samples, and were taken from floors and SSDC surfaces. Because the sample areas between
a regular and side-by-side sample were adjoining, and the same sampling technique was used to
collect both samples, any spatial variability between the two sample results was assumed to be
minimized. Also, both samples were collected within a few minutes of each other at specified
times during the activity process. As a result, one expects any differences between the two results
to reflect only variability in the sample collection and measurement processes. However, in the
summary and analysis of loading data obtained from vacuum dust sampling, the first sample
collected within a pair had consistently higher results than the second sample collected.
Figure 7-10 presents a log-log scatterplot of loadings between the first sample collected
versus the second sample collected for each regular/side-by-side pair collected in the study, when
vacuum sampling techniques were used. The majority of points in this plot fall above the 45° line,
indicating that the first sample collected in a pair was consistently higher than that of the second
sample collected.
For settled dust loadings, concentrations, and physical sample weights, Tables D-2a
through D-2c in Appendix D (one table for each study phase) list the results of regular and side-
by-side sample pairs as well as the differences in results between two samples in a pair. Results
are listed according to whether the sample was collected from a floor or SSDC surface, and
whether the sample was collected pre- or post-activity. In each of these breakdowns, loadings for
the first sample collected were larger than for the second sample collected in a majority of the
sample pairs.
To test the hypothesis that the results for the first sample taken from two adjoining sample
areas were significantly different from results for the second sample taken, two statistical tests
were performed. First, a binomial test was performed to test whether the percentage of sample
pairs having larger results in the first sample collected was significantly different from 50%. Next,
a Wilcoxon signed-rank test was performed to determine whether the first result minus the second
result (i.e., the differences in Tables D-2a through D-2c) was significantly different from
7-11
-------
1000000
100000
Sbpe=1
1 10 100 1000 10000 100000 1000000
Loading for Sample Taken Second fag/ft2)
*** Pre- Act. Floor °°°1-hr Post-Act. Floor
• • • Post-Act. SSDC (vac.) - Equal Results
Figure 7-10. Lead Loadings for Adjoining Samples Collected via Vacuum Techniques
(Sample Collected First Versus Sample Collected Second)
zero. The results of these tests are included in Table 7-1, where symbols indicate whether
statistical significance was observed at the 0.05 level. Significance was seen for sample loadings
in pooled sample pairs for the carpet removal and window replacement phases, as well as across
all phases. Significance was not as apparent for sample concentrations and weights. A lack of
significance in some cases may be the result of low statistical power to detect differences, due to
the small sample size.
As a result of the investigation of side-by-side data in the EFSS, it is hypothesized that
taking a vacuum dust sample may bias lead loadings that exist in adjoining areas. This conclusion
affects the interpretation of side-by-side vacuum sampling designs that are applied to simulate
field duplicates or to observe pre- and post-activity lead levels. However, a more detailed study
that solely addresses this hypothesis is necessary to make such a conclusion with sufficient
confidence.
7-12
-------
Table 7-1. Percentage of Sample Pairs (Regular and Side-by-Side QC Samples) Where
Result for the First Sample Collected Was Different From the Result for the
Second Sample Collected (Vacuum Samples Only)
Phase
Carpet Removal
Window
Replacement
CED
All Phases
Floor Samples (Pre-
Activity)
Floor Samples (Post-
Activity)
SSDC Samples111 1
(Post-Activity) | All Samples
Sample Loadings
62.5% (5/8)
66.7% (2/3)
63.6% (7/11)
75.0% (6/8)
100% (3/3)
___
81.8%T (9/11)
68.8%T (11/16)
100% (3/3)
55.6% (5/9)
67.9% T (19/28)
68.8%*T (22/32)
88.9%*T (8/9)
55.6% (5/9)
70.0%*T (35/50)
Sample Concentrations
Carpet Removal
Window
Replacement
All Phases
62.5% (5/8)
100% (3/3)
72.7%T (8/11)
50.0% (4/8)
33.3% (1/3)
45.5% (5/11)
56.3% (9/16)
33.3% (1/3)
52.6% (10/19)
56.3% (18/32)
55.6% (5/9)
56.1% (23/41)
Sample Weights
Carpet Removal
Window
Replacement
All Phases
50.0% (4/8)
0.0% (0/3)
36.4% (4/11)
75.0% (6/8)
100% (3/3)
81.8% (9/11)
68.8% (11/16)
66.7% (2/3)
68.4% (13/19)
65.6% (21/32)
55.6% (5/9)
63.4% (26/41)
* Percentage is significantly different from 50% at the 0.05 level by the binomial test.
T Results of the first sample taken are significantly different from the results of the second sample taken, at
the 0.05 level according to the Wilcoxon signed rank test.
111 Stainless steel dustfall collectors
7-13
-------
8.0 STUDY RESULTS FOR INDIVIDUAL PHASES
This chapter presents the results of data summary and statistical analysis of environmental
exposure data collected in the EFSS. Each phase in the EFSS was a substudy in itself, with
specific sampling designs and data organization. Consequently, the study designs and results for
each phase are presented as stand-alone sections, as follows:
Section 8A — Carpet removal phase
Section 8B — Window replacement phase
Section 8C — Controlled, Experimentally-Designed (CED) phase.
Data summaries and statistical analyses in these three sections were centered around data
analysis objectives (see Section 6.1) established for the EFSS. Analysis to address objectives on
sampling methodology evaluation was presented in Chapter 7.
Two additional sections in this chapter document other types of data collected and/or
summarized to address issues concerning study objectives:
Section 8D — Cleanup Investigation. Determine the effects associated with two types
of post-activity cleanup procedures on reducing settled dust lead levels
following R&R activity.
Section 8E — Data from Other Sources. Summarize data obtained from other extant
sources that relate R&R activities to lead exposures.
The appendices to this report (Volume II) provide supporting information to the main
results presented in this chapter. Supporting information includes additional data summaries and
detailed discussion of statistical methods.
8-1
-------
8A-1.0 STUDY DESIGN IN THE CARPET REMOVAL PHASE
Chapter 4 presented aspects of the EFSS study design that were common across all
phases. This section presents specific details of the study design as it was applied in the carpet
removal phase. This section also presents characteristics of the eight dwelling units monitored in
this phase.
8A-1.1 SAMPLING DESIGN FOR THE CARPET REMOVAL PHASE
The sampling design for the carpet removal phase was developed to address the objectives
presented in Section 6.1. Table 8A-1 presents the proposed types and numbers of environmental
field samples (both regular and QC samples) collected within each dwelling unit. These samples
included the following:
• Personal exposure samples taken during the activity
• Ambient air samples taken during the activity
• Settled dust samples taken before the activity from the carpet to be removed
• Settled dust samples taken before and after the activity from floor and window sill
surfaces
• Settled dust samples taken following completion of the activity from stainless steel
dustfall collectors (SSDCs) placed on the floor immediately prior to the activity.
Three vacuum dust samples were collected prior to the start of the activity from the carpet
to be removed. The Battelle field team leader selected the sample locations from the entire area
of carpet to be removed, in an attempt to represent average carpet debris across the activity area
in the unit. Areas with unusually high amounts or types of debris were not considered. A given
dust sample was collected from a total area defined by a 1-ft2 steel template, and lead content of
all these samples was used as a covariate in the statistical analysis to predict lead exposure which
can result from carpet removal.
Dust samples from floors and SSDCs were collected at three distinct locations in each
unit. The locations (labeled LI, L2, L3) were in "adjacent areas" or areas outside of, but at the
entrance to, an activity room. The sampling plan called for the three locations to be randomly
distributed among adjacent areas throughout the unit; however, the locations were also dictated
by logistical constraints, such as the ability of adjacent areas to accommodate the total area
required for dust sampling (from 5 to 8 square feet). Sampling locations were partitioned into
five, six, or eight subareas, each one square-foot in area, as illustrated in Figure 8A-1. A single
settled dust sample was collected from each subarea at a specific time during the activity.
8-2
-------
Table 8A-1. Numbers and Types of Environmental Samples (Regular and QC) Proposed in
the EFSS Carpet Removal Phase Within Each Study Unit
Regular Samples
When Sample
Was Taken
Pre-activity
During activity
1-hr. Post-
activity
2-hr. Post-
activity
Sample Collection Method
Vacuum
Ambient air pump
Personal air pump
Ambient air pump
Vacuum
Wipe
Vacuum
Sample Type/Location111
Settled dust (Floor)
Settled dust (Window sill)
Settled dust (Carpet surface)
Ambient air
Personal air (R&R workers)
Ambient air (adjacent room)
Settled dust (Floor)
Settled dust (SSDC)
Settled dust (Window sill)
Settled dust (SSDC)
Settled dust (SSDC)
Total
# Samples
Proposed
3
2
3
1
2
2
3
3
2
2
3
26
Field PC Samples
When Sample
Was Taken
Pre-activity
1-hr. Post-
activity
2-hr. Post-
activity
Sample Collection Method
Vacuum
Wipe
Personal air pump
Ambient air pump
Vacuum
Wipe
Vacuum
Sample Type/Location111
SBS settled dust (Floor)
Field blank
Field blank
Field blank
Field blank
SBS settled dust (Floor)
SBS settled dust (SSDC)
SBS settled dust (SSDC)
SBS settled dust (SSDC)
Total
# Samples
Proposed
1
1
1
1
1
1
1
1
1
9
111 SSDC = stainless steel dustfall collector
SBS = sample taken side-by-side with a regular sample
Immediately prior to the start of carpet removal, the field team collected one vacuum dust
sample within a one-square-foot area of the floor surface in each of the three sampling locations
designated in Figure 8A-1. The field team also collected one vacuum dust sample from one-half
of each of two window sills located in activity rooms. Lead content of these floor and window sill
dust samples represented baseline levels which were compared to lead levels in samples taken
upon completion of carpet removal from areas adjoining the pre-activity sample areas.
8-3
-------
Location L1:
Cell #1
Floor vacuum
(Pre-activity)
Cell #4
Floor vacuum
(1 hr. post-
activity)
Cell #2
Stainless steel
surface -- vacuum
(1 hr. post-activity)
Cell #5
Stainless steel
surface - vacuum
(2 hr. post-activity)
Cell #3
Stainless steel
surface -
wipe (1 hr. post-
activity)
Location L2:
Cell #1
Floor vacuum
(Pre-activity)
Cell #4
Floor vacuum
(1 hr. post-
activity)
Cell #2
Stainless steel
surface — vacuum
(1 hr. post-activity)
Cell #5
Stainless steel
surface — vacuum
(2 hr. post-activity)
Cell #3
Stainless steel
surface -
wipe (1 hr. post-
activity)
Cell #6
Stainless steel
surface - wipe (1
hr. post-activity)
(side-by-side)
Location L3:
Cell #1
Floor vacuum
(Pre-activity)
(side-by-side)
Cell #5
Floor vacuum
(1 hr. post-
activity)
(side-by-side)
Cell #2
Floor vacuum
(Pre-activity)
Cell #6
Floor vacuum
(1 hr. post-activity)
Cell #3
Stainless steel
surface - vacuum
(1 hr. post-activity)
Cell #7
Stainless steel
surface - vacuum
(2 hr. post-activity)
Cell #4
Stainless steel
surface - vacuum (1
hr. post-activity)
(side-by-side)
Cell #8
Stainless steel
surface - vacuum (2
hr. post-activity)
(side-by-side)
Note: Due to space considerations, no wipe samples were taken at location L3. In shaded subareas,
dust samples were taken directly from the uncarpeted floor surface.
Figure 8A-1. Three Settled Dust Sampling Locations at Adjacent Areas Within Each Unit,
and the Types of Regular and QC Samples Collected Within Each Location, in
the Carpet Removal Phase
8-4
-------
One pre-activity ambient air sample was taken to obtain an estimate of baseline airborne lead.
Because the ambient air samples took two hours to collect, the pre-activity ambient air sample
was sometimes taken on the day prior to carpet removal.
Personal and ambient air samples were collected during carpet removal activities. Personal
exposure sampling was conducted on up to two R&R workers per dwelling unit to measure lead
levels in air within the workers' breathing zones while they were performing carpet removal.
Ambient air samples were collected in two rooms adjacent to the activity to measure lead levels in
air dust that could be inhaled by others present.
When possible, doors to rooms where ambient air samples were being collected were
closed while carpet removal was being performed.
Post-activity sampling included settled dust samples from floors, window sills, and SSDCs
positioned immediately prior to the start of the activity. At one hour following completion of all
carpet removal activities, vacuum dust samples were collected from the remaining halves of the
two window sills where pre-activity dust samples were collected. Vacuum dust samples were
collected at three floor sampling locations one hour following the activity. Samples were taken
from the floor surface and from a SSDC adjacent to the pre-activity vacuum floor dust sample
area (as indicated in Figure 8A-1).
To compare results between wipe and vacuum sampling protocols, dust samples from
SSDCs were collected at one hour following conclusion of carpet removal. These samples were
taken from locations LI and L2 (see Figure 8A-1) using wipe sampling techniques. To evaluate
whether there was an increase in lead "fallout" in settled dust after waiting two hours rather than
one hour, dust samples were collected from SSDCs in each sample location (Figure 8A-1) using
vacuum techniques. All three samples were taken at two hours following the activity from areas
adjoining the one-hour post-activity vacuum sample area. The results of these additional samples
are presented in Chapter 7.
Table 8A-1 specifies 35 environmental samples in each dwelling unit, nine of which were
field quality control (QC) samples. One field blank was taken prior to the start of carpet removal
for every sampling medium (vacuum dust, wipe dust, personal air, ambient air) to determine the
extent of contamination present during sample collection. At one settled dust sampling location
(L3), side-by-side vacuum dust samples were taken from the floor surface and from SSDCs to
allow sampling and measurement error to be characterized.
The number of samples made it was impossible to collect all settled dust samples at
precisely the time at which they were to be collected (i.e., at one hour or two hours following
completion of carpet removal). The time period for settled dust sample collection ranged from 15
minutes before to 40 minutes after the scheduled time.
8-5
-------
Air samples were taken at relatively high flow rates (20 L/min for ambient air samples and 5
L/min for personal air samples) in the carpet removal phase. To make these rates more
compatible with NIOSH specifications, the rates were lowered to 12.5 L/min and 4 L/min,
respectively, in later phases of the EFSS. In addition, ambient and personal air samples taken in
the carpet removal phase were collected open-faced (with the entire inlet section of the cassette
assembly removed), in contrast to the NIOSH standard method (used in later phases of the EFSS)
of removing only the inlet plug. The fact that different results could be obtained from the two
methods should be considered when interpreting the results.
More details on the sampling design, including the protocols used to collect the samples,
are given in the Quality Assurance Project Plan.
8A-1.2 UNIT CHARACTERISTICS
Field sampling in the carpet removal phase took place in eight occupied dwelling units.
Table 8A-2 provides information on these units, including the days on which carpet removal and
field sampling were performed, the rooms from which carpet was removed, and the approximate
square footage of carpet removed.
Details on the recruitment of the units are presented in Chapter 5. Table 5-2 in Chapter 5
contains summaries of pre-activity dust lead loadings from floors and window sills in these units.
Chapter 5 also contains discussions of worker characteristics. It should be noted that carpet
removal was conducted by professional abatement workers in the four Alameda County units and
by homeowners in the four St. Louis units. While the work practice of these two types of
workers may differ, they reflect the wide variety of workers conducting R&R activities.
Table 8A-2. Dwelling Units Included in the EFSS Carpet Removal Phase
Study
Unit ID
1-01
1-02
1-03
Date of
Activity
06/22/93
06/23/93
06/24/93
Unit Description
Two-story house;
Oakland, CA;
100 years old;
peeling ext. paint;
rehabbed 15 yrs ago;
Basement apartment;
Oakland, CA;
100 year-old building;
Int. walls 10 yrs old;
Frequent flooding
Across from highway;
Single-story house;
Oakland, CA;
50 years old;
peelinq ext. paint;
Type of R&R
Workers
Prof.
Contractor
Prof.
Contractor
Prof.
Contractor
Activity Areas (approx. square
feet of carpet removed is in
parentheses)
Living Room (260)
Dining Room (143)
Hall (floor 1) (90)
Stairway (60)
Hall (floor 2) (72)
Living Room (162)
Bedroom #1 (127)
Bedroom #2 (149)
Living Room (212)
Hallway (34)
8-6
-------
Table 8A-2. Dwelling Units Included in the EFSS Carpet Removal Phase (Continued)
Study
Unit ID
1-04
2-01
2-02
2-03
2-05
Date of
Activity
06/25/93
07/13/93
07/15/93
07/14/93
07/16/93
Unit Description
Two-story duplex;
Alameda, CA;
75 years old;
Little soil covering;
Three-story house;
St. Louis, MO;
90 years old;
R&R work in progress (2nd floor
family room);
Two-story house;
Webster Groves, MO;
50 years old;
Two-story house;
St. Louis, MO;
96 years old;
Brick exterior;
Two-story house;
St. Louis, MO;
60 years old;
Brick exterior;
Type of R&R
Workers
Prof.
Contractor
Nonprof.
Resident
Nonprof.
Resident
Nonprof.
Resident
Semiprof.
Resident
Activity Areas (approx. square
feet of carpet removed is in
parentheses)
Living Room (256)
Dining Room (206)
Stairway (170)
Hall (level 2)(100)
Family Room (224)
Bedroom #1 (224)
Living Room (168)
Dining Room (144)
Stairway (120)
Living Room (210)
Dining Room (180)
Office (120)
Bedroom (120)
Hall (floor 1) (21)
8A-2.0 STUDY RESULTS FOR THE CARPET REMOVAL PHASE
This chapter presents the results of the statistical analysis of environmental sample data
from the carpet removal phase of the EFSS study.
Analytical data were available for statistical analysis from all environmental field samples
that were collected in the study. Table 8A-1 specifies 35 samples to be collected in each of the
eight study units, for a total of 280 samples. Of these proposed samples, 278 were successfully
collected. The two uncollected samples were personal air samples from units 2-03 and 2-05,
where both units had only a single worker conducting carpet removal activity. In addition, two
cassette samples were necessary to monitor personal air over the entire carpet removal activity for
both workers in unit 1-01 and for the single worker in unit 2-03. This resulted in three additional
samples collected making analytical data available for 281 regular and QC field samples in the
carpet removal phase. A summary of the samples collected and data received is displayed in
Table CR-1 of Appendix A.
This section presents results and conclusions of the statistical analysis of data from the
carpet removal phase. It is organized according to data analysis objectives (see Section 6.1).
Supporting discussion and materials are included as Appendices A through D.
8-7
-------
8A-2.1 CHARACTERIZE LEAD DISTURBANCE AND POTENTIAL LEAD EXPOSURE
Tables CR-2a through CR-6b in Appendix A contain summaries of lead concentrations in
personal and ambient air samples, and physical sample weights, lead loadings, and sample lead
concentrations in dust samples. The summaries, presented for each study unit as well as for the
entire carpet removal phase, include the arithmetic and geometric means, the standard deviation of
the log-transformed data, and the minimum and maximum observed values. As discussed in
Chapter 6, statistical analysis of dust lead levels focused primarily on log-transformed lead
loadings, or the amount of lead per unit surface area sampled.
The lead disturbance and exposure data observed in the carpet removal phase are
characterized below, according to the sample type or component sampled.
8A-2.1.1 Personal Worker Exposures
The duration of carpet removal activity (and of personal air monitoring) across the workers
in this phase ranged from 46 to 173 minutes, with an arithmetic mean of 92 minutes. Total
activity monitored included preparation, carpet removal, disposal, and cleanup (if any).
Monitoring occurred for two workers in six of the eight units, and for one worker in each of the
remaining two units, resulting in 14 estimates of personal air lead concentration. The two
workers who worked alone in a unit had the longest durations of personal air monitoring in the
study (155 and 173 minutes).
The amount of lead collected within the sample cassette(s), the air pump flow rate, and the
duration of monitoring were used to calculate a given worker's task-length average (TLA)
personal air lead concentration in |ig/m3. The concentrations, each representing instrument data
above the detection limit, are plotted in Figure 8A-2 for each study unit. A worker's TLA
exposure represents average exposure over the duration of performing the activity.
A boxplot of TLA lead concentrations is included as the rightmost boxplot in Figure 8A-3.
Descriptive statistics on personal air lead concentrations are displayed in Tables CR-2a and CR-2b
in Appendix A.
Four of the 14 monitored workers (29%) had TLA lead exposures above the OSHA PEL
of 50 |ig/m3. These four results occurred for the two workers in unit 1-01 (results of 51.3 and
59.9 |ig/m3) and for the two workers in unit 1-04 (results of 128 and 221 |ig/m3). Activity in both
of these units took more than 90 minutes per worker, and the carpets were visually observed as
excessively dirty. In contrast, the TLA lead exposure among the other six study units ranged
from 0.86 to 8.44 |ig/m3, well below the OSHA PEL.
The arithmetic mean of the 14 worker TLA lead exposures was 35.9 |ig/m3. However,
because the data are more closely characterized by a lognormal distribution than a normal
distribution, the geometric mean of 8.44 |ig/m3 is considered a better representation of the data,
as it estimates the median (or central tendency) of the data distribution.
8-8
-------
Figure 8A-2 illustrates that the range of TLA lead concentration data across the different
units is greater than the range of data within a unit. This result is further investigated in Section
8A-2.2.4.
8A-2.1.2 Potential Occupant Exposures to Airborne Lead
Within each unit, one pre-activity ambient air sample and two during-activity ambient air
samples were collected in the carpet removal phase. Each sample was collected for a 120-minute
duration (±4 minutes). The amount of lead collected within the sample cassette, the air pump
flow rate, and the duration of monitoring were used to calculate lead concentration in |ig/m3 air
for each sample. The lead concentrations are plotted in Figure 8A-4 for each study unit.
Boxplots of pre-activity and during-activity ambient air lead concentrations are included in Figure
8A-3. Descriptive statistics on ambient air lead concentrations are displayed with those for the
personal worker exposure concentrations in Tables CR-2a and CR-2b of Appendix A. All
instrument data on ambient air samples were reported above the detection limit.
The leftmost boxplot in Figure 8A-3 illustrates the range of lead concentrations in ambient
air observed during pre-activity periods within each unit. The concentrations were approximately
symmetric about the arithmetic mean of 0.10 |ig/m3, ranging from 0.05 to 0.17 |ig/m3. The
relatively low unit-to-unit variability in these concentrations indicates that the eight study units
were relatively homogeneous in their baseline airborne lead levels. Thus, no baseline adjustment
was made to the during-activity ambient air concentration prior to statistical analysis.
The lead concentrations for ambient air samples collected during carpet removal activity,
summarized by the center boxplot within Figure 8A-3, ranged from 0.06 to 13.4 |ig/m3 with a
geometric mean of 0.33 |ig/m3. The maximum reading of 13.4 |ig/m3, flagged by statistical outlier
tests (Appendix B), was more than twice that of the next largest reading (5.36 |ig/m3). The
ambient air samples associated with the two largest concentrations were collected within the same
two units where the largest worker personal exposure concentrations were observed. However,
the other two ambient air concentrations observed in these units were comparable to
concentrations in the other units.
As illustrated in the center boxplot of Figure 8A-3, the arithmetic mean for during-activity
ambient air lead concentrations (1.48 |ig/m3) was heavily influenced by the two largest
concentrations. The geometric mean of 0.33 |ig/m3, more representative of the data than the
arithmetic mean, was approximately three times higher than the geometric mean for the baseline
(pre-activity) concentrations. However, such low levels as observed here imply that the difference
is likely of little practical significance.
For one study unit, the geometric mean ambient air lead concentration during carpet
removal was less than the unit's baseline concentration. In contrast, three study units had
geometric means exceeding ten times their respective baseline concentrations.
8-9
-------
1000.0
I
100.0
10.0
1.0
0.1
1-01 1-02 1-03 1-04 2-01 2-02 2-03 2-05
Unit ID
The dotted line at 50 yug/m3 represents the OSHA PEL.
Figure 8A-2. Scatterplot of Task-Length Average Personal Exposure Lead Concentrations
(//g/m3 air) for Each Worker Within Each Study Unit in the Carpet Removal
Phase
1000.00
100.00
10.00
1.00
0.10
0.01
Prt
ar
*
M
*
i '
(
3 -Activity During -Activity Personal Exposure
nbient air ambient air samples
(n=8) (n=16) (n=14)
Sample Type
The dotted line at 50
represents the OSHA PEL.
Figure 8A-3. Boxplots of Lead Concentrations (//g/m3 air) for Personal Exposure and
Ambient Air Samples in the Carpet Removal Phase
8-10
-------
100.00 i
10.00
1.00
0.10
0.01 \
1-01 1-02 1-03 1-04 2-01 2-02 2-03 2-05
Unit ID
000 Pre-Activity Ambient Air Result • • • Ambient Air Result During Activily
The dotted line at 50
represents the OSHA PEL.
Figure 8A-4. Scatterplot of Ambient Air Lead Concentrations (//g/m3 air) Within Each
Study Unit in the Carpet Removal Phase
8A-2.1.3 Lead Disturbance and Potential Occupant Exposure to Lead In Dust
In the carpet removal phase, settled dust was collected from three types of flat surfaces to
characterize lead disturbance and potential occupant exposures to lead in dust. The three surface
types, and when they were sampled, were as follows:
• floor surfaces in adjacent rooms, sampled pre- and post-activity
• SSDCs in adjacent rooms, placed immediately prior to activity and sampled post-
activity
• window sills in activity rooms, sampled pre- and post-activity.
The results of analysis on settled dust samples taken from the three surface types are summarized
below.
8A-2.1.3.1 Vacuum Dust Samples Taken Pre- and Post-Activity from Floor Surfaces
Vacuum dust samples were collected from one-square-foot areas on floor surfaces prior to
the start of the activity and one hour following completion of the activity. One pre-activity and
one post-activity sample were taken from adjoining areas in three distinct locations in each unit
(Figure 8A-1). Each location was immediately adjacent to an activity room. The difference in
8-11
-------
lead levels between the adjoining (or paired) pre- and post-activity samples within each location
was an estimate of lead exposure in settled dust generated by carpet removal activity.
Results of vacuum floor dust sampling (physical sample weight, lead loading, and lead
concentration) are summarized in Tables CR-3a through CR-3d in Appendix A. Appendix A also
includes boxplots of lead loadings and concentrations (Figures CR-la and CR-lb for pre-activity
vacuum floor samples; Figures CR-2a and CR-2b for post-activity vacuum floor samples). The
geometric mean sample weight increased fourfold from pre- to post-activity (from 0.03 to 0.12 g),
while the geometric mean lead concentration more than doubled (from 475 to 1080 |ig/g). Larger
increases in geometric mean were observed for lead loadings, which increased nine times (from
14.4 to 130 |ig/ft2) from pre- to post-activity. These increases are equivalent to the geometric
mean of the ratio of post-activity to pre-activity results at a given sample location.
Lead loading was the measurement of most interest to compare between adjoining pre- and
post-activity samples. Lead loadings at a given sample location were expected to increase as a
result of the activity. However, at five of the 24 sample locations in the study, lower lead
loadings were reported for the post-activity sample than for the adjoining pre-activity sample.
This result indicates that spatial variability and/or measurement (analytical) error were
considerable for lead loadings on floors. A plot of the difference in loadings between post- and
pre-activity samples for each study unit was presented in Figure 7-4 of Chapter 7.
Section 7.1 of Chapter 7 gives an evaluation of the relative merit of using paired pre-/post-
activity floor samples versus single samples from SSDCs to measure the amount of lead disturbed
during carpet removal. This evaluation concluded that results for dustfall samples from SSDCs
were less variable and more highly correlated with results from other sampling media, and were
thereby preferred over the pre-/post-activity floor dust results to characterize lead levels. SSDC
results are given in the next subsection.
8A-2.1.3.2 Vacuum Dust Samples from Stainless Steel Dustfall Collectors.
In addition to the floor dust samples described above, vacuum dust samples were taken
from SSDCs as an alternative measure of lead disturbance. At each floor sample location (Figure
8A-1), a single settled dust sample was collected one hour following completion of carpet
removal activities from a SSDC placed adjacent to the post-activity floor dust sample area.
For this phase and the window replacement phase, other SSDCs were placed at each floor sample
location to address sampling methodology issues. The results of samples from SSDCs taken two
hours following completion of carpet removal were presented in Section 7.2 of Chapter 7, where
they were compared with the one-hour results. In Section 7.3 of Chapter 7, the one-hour post-
activity vacuum results were compared with the wipe dust sampling results from adjoining sample
areas. These two sections also provide additional summary and display of the results presented
herein.
Tables CR-4a through CR-4d in Appendix A present a summary of the results of vacuum
dust sampling from SSDCs one hour following completion of carpet removal activities. Boxplots
of lead loadings and concentrations within one-hour post-activity dustfall collectors samples are
8-12
-------
also included as Figures CR-2a and CR-2b, respectively, in Appendix A. These lead loadings
ranged from 2.61 (the detection limit) to 621 |ig/ft2, with a geometric mean of 24.3 |ig/ft2. A plot
of the loadings on a unit-by-unit basis is presented in Figure 7-4 in Section 7.1.
8A-2.1.3.3 Vacuum Dust Samples from Window Sill Surfaces Taken Pre- and Post-Activity.
Similar to the floor dust sample approach presented in Section 8A-2.1.3.1, pre- and post-
activity dust sampling was performed on two window sill surfaces per unit. The window sill was
located in a room where carpet removal activity took place. One-half of the window sill area was
sampled for settled dust prior to the start of carpet removal activities, and the other half was
sampled one hour following completion of these activities. These sample results provided
information on lead exposures in settled dust within an activity room.
Tables CR-5a through CR-5d of Appendix A present data summaries for the pre- and post-
activity window sill dust samples. Also in Appendix A are boxplots of lead loadings and
concentrations for pre-activity window sill dust samples (Figures CR-la and CR-lb) and for post-
activity window sill dust samples (Figures CR-2a and CR-2b). All window sill dust samples taken
in this study contained detectable amounts of lead.
Across all samples, the geometric mean loadings on window sills increased from pre- to
post-activity by approximately 58% (from 418 to 661 |ig/ft2), a smaller percentage increase than
the 800% increase observed within floor dust. However, for each of these measures, the
geometric means at pre- and post-activity were higher for window sills than for floors. The larger
data values for window sills (pre-activity) may be one reason for the smaller percentage increases
associated with window sills over floor surfaces.
Three of the eight study units had geometric mean values of less than one for the ratio of
post- to pre-activity lead loadings on window sills. This indicates that, similar to the finding for
floor dust loadings, spatial variability and/or analytical error associated with window sill lead
loadings were considerable.
8A-2.2 ASSESS FACTORS OR MEASUREMENTS RELATED TO LEAD DISTURBANCE
In order to characterize the statistical relationship between lead exposure in various media
and potential predictors of this exposure, as well as to quantify the various components of total
variability in lead exposures, statistical models were developed and fitted to lead measurements
obtained in the carpet removal phases. The models were fit to data across all study units using
analysis of variance (ANOVA) techniques.
Sections 8A-2.2.1 through 8A-2.2.3 present the results of fitting log-linear models to
predict lead measurements from a series of selected covariates. These covariates included pre-
activity lead loadings in the carpet removed, pre-activity lead loadings on floors and window sills,
and the duration of activity. The models test whether a change in the value of a specific covariate
significantly affects (in a multiplicative fashion) the value of the lead measurement being
predicted. The forms of these statistical models are found in Section C.I of Appendix C.
8-13
-------
The approach to characterizing variance components associated with a specific type of lead
measurement involves fitting the same models described in the previous paragraph, with only the
random effects associated with each variance component represented in the model (i.e., covariates
are excluded). These models are documented in Section C.2 of Appendix C. When estimable
based on the data, variance components in the model include variability associated with sampling
from different study units ("unit-to-unit" variability), variability associated with sampling multiple
locations or workers within a given unit ("within-unit" or "location-to-location" variability), and
variability associated with taking samples from multiple areas within the same location ("within-
location" or "replicate-to-replicate" variability). Results of estimating variance components are
found in Section 8A-2.2.4.
It should be noted that the model fittings in this section were performed on small numbers
of data points having relatively high variability. Therefore, tests for significant covariate effects
tended to have low power.
8A-2.2.1 Personal Worker Exposures
Model (CPT-1) in Section C.I of Appendix C was fitted to 14 log-transformed lead
concentrations (|ig/m3) from samples collected through personal air monitoring of R&R workers
during carpet removal activities. These data were plotted in Figure 8A-2 and represent average
exposure over the duration of performing an activity. Covariates in the model included:
• geometric mean lead loading in the removed carpet for the unit
• geometric mean pre-activity lead loading on floor surfaces in adjacent rooms of the
unit
• geometric mean pre-activity lead loading on window sill surfaces in activity rooms of
the unit
• amount of time that carpet removal took place in the unit.
Each of the coefficients associated with the above four (log-transformed) covariates was not
significantly different from 0 at the 0.05 level, indicating no evidence exists that these covariates
have a significant effect on personal worker exposure. This result could be due to high variability
observed in the data, relative to the small sample size.
8A-2.2.2 Potential Occupant Exposures to Airborne Lead
Model (CPT-2) in Section C.I of Appendix C was fitted to log-transformed lead
concentrations (|ig/m3) in air within adjacent rooms, obtained from ambient air monitoring during
carpet removal activities. The sixteen data points obtained from this phase are plotted in Figure
8A-4. Covariates in the ambient air model included:
8-14
-------
• geometric mean lead loading in the removed carpet for the unit
• geometric mean pre-activity lead loading on floor surfaces in adjacent rooms of the
unit
• geometric mean pre-activity lead loading on window sill surfaces in activity rooms of
the unit
• amount of time that carpet removal took place in the unit
• pre-activity ambient air lead concentration in the unit.
Each of the coefficients associated with the above five (log-transformed) covariates was not
significantly different from 0 at the 0.05 level. Thus, no significant effect on log-lead
concentration in ambient air was observed for these covariates, relative to the small sample size
and the level of variability in the data.
8A-2.2.3 Predicting Lead Disturbance and Potential Occupant Exposures to Lead in Dust
Modeling results are presented in the following subsections for models fitted to lead
loadings from floor, SSDC, and window sill dust-fall samples. The form of each model, and
estimates of model parameters that result from mixed-model analysis of variance, are found in
Section C.I of Appendix C.
8A-2.2.3.1 Vacuum Dust Samples Taken Post-Activity from Floor Surfaces
Model (CPT-3) in Section C.I of Appendix C was fitted to one-hour post-activity lead
loading data (|ig/ft2) from floor surfaces in adjacent rooms. The covariates used to predict these
lead loadings included:
• geometric mean lead loading in the removed carpet for the unit
• pre-activity lead loading on the floor surface adjoining the post-activity sampling area
• geometric mean pre-activity lead loading on window sill surfaces in activity rooms of
the unit
• amount of time that carpet removal took place in the unit.
None of the above covariates were found to be significantly associated with post-activity
floor lead loading at the 0.05 level when the model was fitted to the study data.
8-15
-------
8A-2.2.3.2 Vacuum Dust Samples from Stainless Steel Dustfall Collectors
As with lead loadings from floor surfaces, Model (CPT-3) in Section C.I of Appendix C
was fitted to one-hour post-activity lead loading data (|ig/ft2) from SSDCs in adjacent rooms. At
the 0.05 significance level, similar results were observed with SSDC samples as with the floor
dust samples: none of the covariates were significantly associated with SSDC lead loadings.
8A-2.2.3.3 Vacuum Dust Samples Taken Post-Activity from Window Sill Surfaces
Similar to Model (CPT-3), Model (CPT-4) in Section C.I of Appendix C was fitted to one-
hour post-activity lead loading data (|ig/ft2) from window sill surfaces in activity rooms. The
covariates used to predict these lead loadings included:
• geometric mean lead loading in the removed carpet for the unit
• pre-activity lead loading on the window sill surface adjoining the post-activity
sampling area
• geometric mean pre-activity lead loading on floor surfaces in adjacent rooms of the
unit
• amount of time that carpet removal took place in the unit.
The coefficient associated with lead loading in the adjoining pre-activity window sill sample
was significantly different from 0 at the 0.05 level (p=0.044). The estimate of this coefficient was
positive, indicating that high lead loadings in the pre-activity sample were associated with high
lead loadings in the post-activity sample from the adjoining sample location. None of the other
covariates were found to be significantly associated with post-activity window sill lead loading at
the 0.05 level.
8A-2.2.4 Estimating Variance Components
The sampling design for the carpet removal phase directed that samples be collected over
multiple study units, multiple locations (or workers) within a study unit, and multiple areas within
selected locations (i.e., side-by-side QC samples). In this way, key sampling components of total
data variability could be isolated and characterized. Therefore, when estimable, the magnitudes of
the following variance components were estimated using random-effects analysis of variance:
• "unit-to-unit" variability
• "location-to-location" (or worker-to-worker) variability within a unit
• "replicate-to-replicate" variability within a location.
Variance components were estimated for the following lead measurements:
• pre-activity floor dust lead loadings
8-16
-------
pre-activity window sill dust lead loadings
1-hour post-activity floor dust lead loadings
1-hour post-activity window sill dust lead loadings
1-hour post-activity SSDC vacuum lead loadings
1-hour post-activity SSDC wipe lead loadings
2-hour post-activity SSDC vacuum lead loadings
during-activity ambient air lead concentrations
TLA personal exposure lead concentrations.
Replicate-to-replicate variability could only be estimated for lead loadings on floors and SSDCs as
these were the only sample types where side-by-side QC samples were taken.
It is assumed that total variability is the sum of the variabilities associated with the above
three components. Any additional variance components are confounded within one or more of
these components.
Table 8A-3 contains the estimated total variability in log-transformed lead measurements,
plus estimates of the variance components, for each sample type. The results represent variability
in the log-domain, so they are expressed in log units.
The estimated variance components for the settled dust sample types in Table 8A-3 (i.e.,
the first seven rows of the table) were obtained from fitting a random effects analysis of variance
model (Model (C-2) of Appendix C) to data from regular and side-by-side QC samples. Results
for these sample types indicate that the estimated total variability was approximately one-third to
one-half of the estimated mean of the log-transformed measurements. For pre-activity floor and
window sill dust samples and post-activity window sill dust samples, the majority of total
variability in the data was attributed to results from different units. Variable results in side-by-side
2-hour dustfall collector dust samples led to high replicate-to-replicate variability for this sample
type. For the post-activity floor dust and 1-hour post-activity dustfall collector dust samples
(both vacuum and wipe), more variability was attributed to differing sampling locations within a
unit than to any other variance component.
Nearly all of the variability in log-transformed personal air lead concentrations was
attributable to variability across the different study units. This indicates that workers involved in
carpet removal at a given site are exposed to similar lead levels within their breathing zones. In
contrast, almost two-thirds of the total variability in ambient air concentrations was attributable to
differences between the two samples within a study unit.
3-17
-------
Table 8A-3. Estimates of Total Variability and its Estimable Components in Log-
Transformed Lead Measurements by Sample Type for the Carpet Removal
Phase
Sample Type111
Pre-Activity Floor Dust
Pre-Activity Window Sill
Dust
1-hr Post-Activity Floor
Dust
1-hr Post-Activity Window
Sill Dust
1-hr Post-Activity SSDC
Dust (vacuum)
1-hr Post-Activity SSDC
Dust (wipe)
2-hr Post-Activity SSDC
Dust (vacuum)
Personal air
Ambient air
Model-Estimate
of Mean Log
Measurement121
2.69
6.03
4.92
6.49
3.08
3.33
3.43
2.02
-1.12
Square Root of
Estimated
Total
Variability131:
Otot
1.99
2.39
1.71
2.28
1.49
1.47
1.53
1.77
1.60
Square Roots of Estimated Variance Components'4'
Unit-to-unit
(Ou)
1.76 (79%)
2.21 (85%)
0.82 (23%)
1 .90 (70%)
0.74 (25%)
0.68 (22%)
0.94 (38%)
1.73 (95%)
0.97 (37%)
Location-to-
location within a
unit (oj
0.00 (0.0%)
0.92 (15%)
1.30 (57%)
1 .25 (30%)
1.02 (47%)
1.14 (60%)
0.00 (0.0%)
0.39 (5%)
1 .26 (63%)
Replicate-to-
replicate within a
location (OR)
0.92 (21%)
—
0.76 (20%)
—
0.79 (28%)
0.63 (18%)
1 .20 (62%)
___
—
111 SSDC = Stainless steel dustfall collectors.
121 Estimates of the intercept term from the random effects model.
131 Total variability = O2ot = a2, + a2 + a2, , where these parameters were estimated from the random effects model using
restricted maximum likelihood.
141 Number in parentheses is the percent of total variability (o2ot) represented by the given variance component (the square
of the tabled value).
8A-2.3 EVALUATE DIFFERENT MEDIA AND METHODS AS INDICATORS OF EXPOSURE
To quantify the extent of linear relationship between lead exposure estimates from
different sample types and sample collection approaches, Table 8A-4 presents Pearson correlation
coefficients among pairs often variables that measure lead loadings in various media through the
study. These variables include:
• pre-activity dust lead loadings on floors, window sills, and carpet
• 1-hour post-activity dust lead loadings on floors and window sills (collected via
vacuum), and SSDCs (collected via vacuum and wipe)
• 2-hour post-activity dust lead loadings on SSDCs (collected via vacuum)
• lead concentrations in personal air and ambient air samples collected during carpet
removal activity.
8-18
-------
For each variable, the geometric mean was calculated for each of the eight study units. Pearson
correlation coefficients were calculated on these geometric means for each pair of variables. Note
that low correlation between two variables does not necessarily mean that no relationship exists
between them, but rather that the relationship is not linear in nature.
Correlations of greater than 0.9, indicating a high positive linear relationship, were
observed in lead loadings between the following pairs of dust collection methods:
• SSDCs sampled at one-hour post-activity using wipe techniques versus the same
sampling approach using vacuum techniques
• SSDCs sampled at one-hour post-activity using wipe techniques versus SSDCs
sampled at two-hours post-activity using vacuum techniques
• SSDCs sampled using vacuum techniques at one-hour post-activity versus that taken
at two-hours post-activity.
Table 8A-4. Pearson Correlations'11 of Geometric Mean Lead Loadings Between Pairs of
Sample Types and Approaches in the Carpet Removal Phase
Sample Type121
Pre-Act.
Floor
Pre-Act.
Window Sill
Pre-Act
Carpet
Post-Act.
Floor
Post-Act.
Window Sill
1-hr. Post-
Activity
SSDC (wipe)
1-hr. Post-
Activity
SSDC (vac.)
2-hr. Post-
Activity
SSDC (vac.)
Ambient
Air
Pre-Activity
Floor
1.00
Window
Sill
-0.06
1.00
Carpet
to be
Removed
0.72
-0.33
1.00
Post-Activity
Floor
0.42
-0.21
0.34
1.00
Window
Sill
0.20
0.77
-0.35
-0.14
1.00
S. Steel Dustfall Collector
1 hr.
Wipe
-0.30
-0.14
-0.30
0.70
-0.14
1.00
1 hr.
Vacuum
-0.26
-0.14
-0.23
0.70
-0.09
0.98
1.00
2 hr.
Vacuum
-0.27
-0.03
-0.29
0.73
-0.06
0.99
0.97
1.00
Air
Ambient
Air
-0.41
0.32
-0.21
0.41
0.08
0.72
0.77
0.78
1.00
Personal
Exposure
-0.33
-0.23
-0.18
0.67
-0.25
0.97
0.98
0.95
0.78
111 Correlations calculated on geometric means calculated for each study unit (n=8). All correlations greater than 0.71 are
significantly different from zero at the a =0.05 level.
121 SSDC = Stainless steel dustfall collectors.
8-19
-------
In addition, correlations of greater than 0.9 were observed between personal worker exposure
lead concentrations and dustfall lead loadings obtained from the following sampling approaches:
• SSDCs sampled at one-hour post-activity using wipe techniques
• SSDCs sampled at one-hour post-activity using vacuum techniques
• SSDCs sampled at two-hours post-activity using vacuum techniques.
One expected to see high correlations between one-hour and two-hour post-activity dust sample
results and between vacuum and wipe dust sample results. However, the high correlation
between personal worker exposure and loadings in dust on SSDCs indicated that samples
collected from the dustfall collectors may contain more information on lead exposed by carpet
removal than do pre- versus post-activity floor dust samples.
Correlations between post-activity floor dust lead loadings and other loadings from dust
collection methods were in the range of 0.7. While these correlations were lower than that
documented in the previous paragraph, they were still significantly different from 0 at the 0.05
level. Significant correlation was also observed between personal worker and ambient air
concentrations (0.78).
Correlations between pre-activity dust lead loadings from carpet to be removed and the
other variables were quite low and generally not significantly different from zero. Pre-activity
lead loadings versus post-activity lead loadings from adjoining sampling locations had moderately
positive correlation: 0.42 for floors and 0.77 for window sills.
8A-2.4 SUMMARY OF RESULTS
Following are the major results of the R&R carpet removal phase:
• Lead concentrations in personal air samples exceeded the OSHA PEL (50 |ig/m3) in
four of the 14 workers sampled and in two of the eight study units. Other than these
four results, personal air lead concentrations were less than 10 |ig/m3. The dominant
variance component of these data appeared to be the result of sampling in different
study units.
• Lead concentrations in ambient air samples collected during carpet removal activity
ranged from 0.06 to 3.38 |ig/m3, with a geometric mean of 0.33 |ig/m3.
• The difference in lead loadings from vacuum floor dust samples taken pre- and post-
activity had high variability in this study. In contrast, lead loadings from SSDC
samples had substantially lower variability. In addition, SSDC lead loadings had a
very high correlation with personal exposure lead concentrations. This indicates that
the dustfall collectors may be a more accurate and precise measure of the lead
disturbed by carpet removal than the difference in pre- and post-activity floor dust
samples.
8-20
-------
Within the vacuum floor dust samples, the geometric mean lead loadings increased
nine times (from 14.4 to 130 |ig/ft2) from pre- to post-activity. However, a wide
range of values, including negative results, was observed in post- minus pre-activity
floor dust lead loadings at a given location (-195 to 6130 |ig/ft2).
Lead loadings from SSDC samples ranged from 2.61 to 621 |ig/ft2, with a geometric
mean of 24.3 |ig/ft2. High unit-to-unit variability was observed with these data.
Generally, the covariates considered in the statistical modeling to predict log-
transformed lead loadings in various sample types were not significant at the 0.05
level, partially due to the small sample sizes considered. Pre-activity lead loadings on
window sills had a significant effect (at the 0.05 level) on post-activity window sill
lead loadings.
Significantly positive linear correlation was observed between personal worker
exposure concentrations and dust sample loadings from SSDCs, whether wipe or
vacuum techniques were used. This implies that, for carpet removal, dustfall samples
may provide some information on potential lead concentrations in personal exposure
samples from workers, and vice versa.
8-21
-------
8B-1.0 STUDY DESIGN IN THE WINDOW REPLACEMENT PHASE
Chapter 4 presented aspects of the EFSS study design that were common across all
phases. This section presents specific details of the study design as it was applied in the window
replacement phase.
8B-1.1 SAMPLING DESIGN FOR THE WINDOW REPLACEMENT PHASE
The sampling design for the window replacement phase was developed to address each of
the objectives presented in Section 6.1. Table 8B-1 presents the proposed types and numbers of
environmental field samples (both regular and QC samples) collected within each of four dwelling
units. The types of samples included the following:
• Personal exposure samples taken during the activity
• Ambient air samples taken during the activity
• Settled dust samples taken before the activity from the wells of three windows to be
removed
• Paint chip samples taken after the activity from interior and exterior components of
three removed windows
• Settled dust samples taken before and after the activity from floor surfaces at
varying distances from three windows to be removed
• Settled dust samples taken following completion of the activity from stainless steel
dustfall collectors (SSDCs) placed on the floor immediately prior to the activity
adjacent to and at 6 feet from each of three windows to be removed.
In addition to these sample types, plastic tarpaulins were to be placed on the ground
directly beneath the three selected windows on the exterior of the unit, in order to collect dust
and other settled matter which settles on the exterior ground as a result of window replacement.
However, exterior dust samples collected from these tarpaulins were archived for future
analysis and, therefore, were not included in the analysis in this phase.
In each unit, three windows slated to be removed were considered as the basis for each of
the settled dust and paint chip sample types. The windows were selected to represent all
approaches to window replacement, all types of windows being replaced, and the various lead
levels which could be encountered in the activity at the given unit. In addition, windows were
selected based on logistic issues, such as the availability of floor surface at specified distances
from the window for settled dust sampling. Figure 8B-1 displays the layout of settled dust
sampling locations at each selected window.
8-22
-------
Table 8B-1. Numbers and Types of Environmental Samples (Regular and QC) Proposed in
the EFSS Window Replacement Phase Within Each Study Unit
Regular Samples
When Sample
Was Taken
Pre-Activity
During
Activity
1-hr. Post-
Activity
2-hr. Post-
Activity
30-min. Post-
Activity
Post-Activity
Sample Collection
Method
Vacuum
Ambient air pump
Personal air pump
Ambient air pump
Vacuum
Vacuum
Tarpaulin
Chipping
Sample Type/Location111
Settled dust (Window well)
Settled dust (Floor)
Ambient air
Personal air (R&R workers)
Ambient air (Adjacent room)
Settled dust (Floor)
Settled dust (SSDC)
Settled dust (SSDC)
Debris deposit (Exterior)121
Paint chip (Interior Window Sash)
Paint chip (Exterior Window Sash)
TOTAL
Number
Proposed
3
9
1
2
2
9
6
1
3
3
3
42
Field PC Samples
When Sample
Was Taken
Pre-Activity
1-hr. Post-
Activity
Sample Collection
Method
Vacuum
Ambient air pump
Personal air pump
Vacuum
Sample Type/Location111
SBS settled dust (Floor)
Field blank
Field blank
Field blank
SBS settled dust (Floor)
SBS settled dust (SSDC)
TOTAL
Number
Proposed
1
1
1
1
1
1
6
(1) SSDC = stainless steel dustfall collector.
SBS = sample taken side-by-side with a regular sample.
(2) Samples archived; no analysis performed on these samples.
Immediately prior to the start of each window replacement, the field team collected one
vacuum dust sample within a one-square-foot area in each of the three floor sampling locations
represented by "P/P" in Figure 8B-1. The three locations are at 0, 3, and 6 feet perpendicular to
the window. Lead content within a pre-activity floor dust sample represented a baseline lead level
at the given location.
8-23
-------
Exterior
Tarp
Activity Window
Interior
(Not to interfere
with R&R workers)
3'
6'
P/P = locations where pre- and post-activity dust samples were taken from the floor surface.
SSDC = locations where stainless steel dustfall collectors were positioned.
Figure 8B-1. Settled Dust Sampling Locations at a Selected Window in the Window
Replacement Phase
A pre-activity ambient air sample was taken to obtain an estimate of baseline airborne lead
in the unit. As the ambient air samples took two hours to collect, the pre-activity ambient air
sample was sometimes taken on the day prior to window replacement.
Also prior to window replacement activities, the field team collected a vacuum dust sample
from the window wells of the selected windows. This sample was taken to assess the level of
available lead in dust from the given window, for a statistical evaluation of settled dust lead levels
resulting from the activity.
To evaluate how lead disturbance resulting from window replacement is associated with
lead levels in paint, paint chip samples were taken from the interior and exterior components of
the three windows being monitored. These samples were taken after the window was removed.
During window replacement activities, personal and ambient air samples were collected.
Personal exposure sampling was conducted on up to two R&R workers per dwelling unit to
8-24
-------
measure lead levels in air within the workers' breathing zones while they were performing window
replacement. Ambient air samples were collected in two rooms adjacent to one or more "activity
rooms" (i.e., rooms where window replacement took place) to measure lead levels in air dust that
could be inhaled by others present. No window replacement took place in rooms where ambient
air sampling occurred. When possible, doors to rooms where ambient air samples were being
collected were closed while window replacement was being performed.
Post-activity samples were taken of settled dust from floors and from one-square-foot
SSDCs positioned immediately prior to the start of the activity. At one hour following
completion of all window replacement activities (including cleanup), one vacuum dust sample was
collected in each of the three "P/P" sampling locations in Figure 8B-1, from an area adjacent to
the pre-activity vacuum floor dust sample area at that location. Results from this sample were
adjusted for the pre-activity sample result at that location. Also at one-hour post-activity, one
vacuum dust sample was collected from a SSDC positioned at 6 feet perpendicular to the
window.
A SSDC was also placed as close to each selected window as possible without disrupting
the normal activity of the R&R workers. To represent a "worst-case" scenario, a vacuum dust
sample was to be taken from this dustfall collector one hour following window replacement but
prior to the start of cleanup, or immediately prior to cleanup if the delay factor was unrealistic.
Finally, to determine whether lead "fallout" in settled dust increased after two hours
following the activity compared to one hour, a second SSDC was placed 6 feet from one of the
three selected windows in a unit. This SSDC was positioned adjoining the SSDC for the one-
hour post-activity vacuum sample. A vacuum dust sample was collected two hours following the
activity and cleanup. Results of comparison between one-hour and two-hour post-activity sample
lead loadings are found in Chapter 7.
The nature of the field work and the number of samples made it impossible to collect a
sample at precisely one hour or two hours following completion of window replacement. Settled
dust samples were collected in a time frame spanning 15 minutes before to 40 minutes after the
scheduled time. Analysis of the effect of the time of sampling on the lead loading results was
given in Section 7.2 of Chapter 7.
Of the 48 environmental samples specified in Table 8B-1 within each dwelling unit, six were
field quality control (QC) samples. One field blank was taken prior to the start of window
replacement for every sampling medium (vacuum dust, personal air, ambient air) to determine the
extent to which contamination was present during sample collection. At one window, side-by-
side vacuum dust samples were taken from the floor surface and from a SSDC at 6 feet from the
window to allow sampling and measurement error to be characterized and variability in results
between two adjoining sample areas to be observed.
More details on the sampling design, including the protocols used to collect the samples,
are found in the Quality Assurance Project Plan.
8-25
-------
8B-1.2 UNIT CHARACTERISTICS
Table 8B-2 lists the four dwelling units included in the window replacement phase, the
activity dates, and the rooms ("activity areas") within each unit that contained the three windows
selected for environmental sampling. The activity areas in two units were occupied, while the
activity areas of the other two units (Units 3-01 and 4-01) had been vacant and neglected for
some time.
Table 5-2 in Chapter 5 contains summaries of pre-activity lead loadings in these units.
Chapter 5 also contains discussion of worker characteristics.
Table 8B-2. Dwelling Units Included in the EFSS Window Replacement Phase
Study
Unit ID
1-01
2-01
3-01
4-01
Date of
Activity
8/24/93
9/14/93
9/15/93
12/22/93
Unit Description111
Two-story house;
Piketon, OH;
100 years old;
new siding;
1 968 remodeling;
Two-story house;
Columbus, OH;
no age estimate;
peeling ext. paint;
One-story business;
Richwood, OH;
1 50 years old;
brick/wood/concrete
exterior; activity area vacant
Second-floor apt.;
Plain City, OH;
100 years old;
peeling ext paint;
activity area vacant
Type of R&R
Workers
Professional
Contractor
Professional
Contractor
Self-Employed
Professional
Self-Employed
Professional
Activity
Areas121
Den
Kitchen
Bathroom
2 Bedrooms
Hall
Large Room
2 Living Rooms
Bedroom
(1)
All windows to be removed had painted wood frames.
(2) Except for unit 3-01, one window within each specified room (activity area) was selected for
monitoring dustfall and paint lead levels. Other windows may also have been replaced within
these rooms. Unit 3-01 was not divided into rooms, so all three selected windows were in the
same large room.
8-26
-------
8B-2.0 STUDY RESULTS FOR THE WINDOW REPLACEMENT PHASE
This chapter presents the results of the statistical analysis of environmental sample data
from the window replacement phase of the EFSS.
Analytical data were available for statistical analysis from all environmental field samples
that were collected in the window replacement phase. Table 8B-1 specifies 48 samples from each
of the four study units, for a total of 192 samples. Of these proposed samples, 186 were
successfully collected. The 6 uncollected samples included three tarpaulin samples and three floor
dust samples. In each unit, it was necessary to replace cassette samples during personal air
monitoring, resulting in additional cassette samples for analysis. One cassette had to be replaced
during ambient air sampling in unit 2-01 when a failed air pump was replaced, resulting in an
additional area air cassette sample. One additional tarpaulin sample and one additional exterior
paint chip sample were collected at unit 2-01, and one additional two-hour post-activity SSDC
sample at 6 feet from the window was collected at unit 3-01. (This latter sample was eliminated
from the statistical analysis because the SSDC was not placed until after cleanup). Analytical data
were available for a total of 193 regular and QC field samples in the window replacement phase.
A summary of the samples collected and data received is displayed in Table WR-1 of Appendix A.
This section presents results and conclusions of the statistical analysis of data from the
window replacement phase. It is organized according to the method of organizing statistical
objectives given in Section 6.1. Supporting discussion and materials are included as Appendices
A through D.
8B-2.1 CHARACTERIZE LEAD DISTURBANCE AND POTENTIAL LEAD EXPOSURE
Tables WR-2A through WR-6b in Appendix A contain summaries of lead concentrations
in personal and ambient air samples, and physical sample weights, lead loadings, and lead
concentrations in dust and paint chip samples. The summaries, presented for each study unit as
well as for the entire window replacement phase, include the arithmetic and geometric means, the
standard deviation of the log-transformed data, and the minimum and maximum observed values.
As discussed in Chapter 6, statistical analysis of dust lead levels focused primarily on log-
transformed lead loadings, or the amount of lead per unit surface area sampled.
The lead disturbance and exposure data observed in the window replacement phase are
characterized below, according to the sample type or component sampled. Summaries are
presented for personal exposure, area air, and dustfall samples. Paint chip sample results are
summarized in Tables WR-6a and WR-6b, and Figure WR-3 of Appendix A.
8B-2.1.1 Personal Worker Exposures
The duration of personal air monitoring across the workers in this study ranged from 181
to 492 minutes, with an arithmetic mean of 335 minutes. Monitoring was started immediately
before area preparation began, and it continued through window replacement, disposal, and
cleanup activities (if any). Two workers were monitored in each of the four study units, resulting
8-27
-------
in eight estimates of average personal air lead concentrations (in jig lead/m3 of air) over the
duration of activity (referred to as task-length averages (TLA)).
The task-length average lead concentrations from the personal air monitors are plotted for
each study unit along with lead concentrations from ambient air monitors in Figure 8B-2. All
concentrations represent instrument data above the detection limit. Descriptive statistics on
personal air lead concentrations are displayed in Tables WR-2a and WR-2b in Appendix A.
100.00
10.00
1.00
0.10
0.01
„,„,„, Personal Exposure
Samples
The dotted line at 50
1-01 2-01 3-01
Unit
0 0 0 Pre-Activity
Ambient Air
represents the OSHA PEL.
4-01
Post-Activity
Ambient Air
Figure 8B-2. Task-Length Average Personal Air Lead Concentrations, and Ambient Air
Sample Lead Concentrations, Within Each Study Unit in the Window
Replacement Phase
None of the eight monitored workers had personal air levels above the OSHA PEL of 50
|ig/m3. Personal air lead concentrations among the four study units ranged from 2.41 to 44.3
|ig/m3, with a geometric mean of 7.48 |ig/m3.
Figure 8B-2 illustrates that the variability among TLA personal exposure data from
different units is greater than the variability among the two data points within a unit, indicating
that workers within a unit are exposed to similar levels of lead while workers in different units are
generally exposed to much different levels of lead. Further statistical investigation of these two
components of variability is presented in Section 8B-2.2.4.
8-28
-------
8B-2.1.2 Potential Occupant Exposures to Airborne Lead
In each study unit, one ambient air sample was collected prior to the activity and two
ambient air samples were collected during the window replacement activity. Pre-activity samples
were collected over a 120-minute interval (±4 minutes), except for one unit in which the sample
was collected over a 100-minute period. The sampler flow rate for pre-activity ambient air
samples was 12.5 L/min in three of the four units. In the fourth unit (4-01), which had no
electricity, a personal exposure sampler with a flow rate of 4 L/min was used. The ambient air
samples obtained during the window replacement activities were collected over intervals ranging
from 256 minutes to 486 minutes. The flow rates for all 8 activity samples were approximately
4 L/min.
The amount of lead collected within the sample cassette, the air pump flow rate, and the
duration of monitoring were used to calculate lead concentration in |ig/m3 air for each sample (see
Figure 8B-2). Descriptive statistics on ambient air lead concentrations are displayed with the
personal worker exposure lead concentrations in Tables WR-2a and WR-2b of Appendix A. All
instrument data on ambient air samples were reported above the detection limit.
Figure 8B-2 illustrates the range of ambient air data observed during pre-activity periods
within each unit. Ambient air samples had lead levels that ranged from 0.10 to 2.86 |ig/m3. The
highest level occurred in the unit where the air flow rate was lowest. Lead levels for ambient air
samples collected during window replacement activity ranged from 0.29 to 4.16 |ig/m3, with a
geometric mean of 1.16 |ig/m3. The two highest ambient air levels were collected in the same unit
where the personal air levels were the greatest.
8B-2.1.3 Lead Disturbance and Potential Occupant Exposure to Lead in Dust
In the window replacement phase, settled dust was collected from three types of flat
surfaces to characterize lead disturbance and potential hazards to occupants from lead in dust.
These three surface types, and when they were sampled in the study, were as follows:
• floor surfaces, sampled pre- and post-activity
• stainless steel dustfall collectors (SSDCs), placed immediately prior to activity and
sampled post-activity
• window wells in activity rooms, sampled pre-activity.
Dust samples from each surface type were collected on and in the vicinity of three windows being
removed. Section 8B-1 provides more detail about the sampling procedures, including the
locations and timing of each sample collected.
The results of analysis on settled dust samples taken from each of the above three surface
types are summarized below.
8-29
-------
8B-2.1.3.1 Vacuum Dust Samples Taken Pre- and Post-Activity from Floor Surfaces
Vacuum dust samples were collected from one-square-foot areas on floor surfaces prior to
the start of activity and at one hour following completion of activity. Sample locations were at 0,
3, and 6 feet from a window being removed. At each location, one pre-activity and one post-
activity sample were taken from adjoining areas (Figure 8B-1). For each distance, results of
vacuum dust sampling (physical sample weight, lead loading, and lead concentration) directly
from the floor surface are summarized by unit and across all units in Tables WR-3a through WR-
31 in Appendix A. Boxplots of lead loadings and concentrations are included in Figures WR-la
and WR-lb for pre-activity vacuum floor samples and in Figures WR-2a and WR-2b for post-
activity vacuum floor samples.
Figure 8B-3 shows a scatterplot of the post-activity lead loadings versus the pre-activity
floor lead loading at the same window. The plotting symbols indicate the distance from the
window at which the samples were collected. Chapter 7 indicated that 9 of the 35 pre/post
sample pairs (non-QC samples) had lower lead loadings reported for the post-activity sample,
reflecting high spatial variability and/or analytical error in the data. The geometric mean lead
loadings (across all units) for regular and side-by-side QC samples taken pre- and post-activity
were 1920 and 3910 |ig/ft2 at 0 feet, 491 and 1290 |ig/ft2 at 3 feet, and 334 and 878 |ig/ft2 at 6
feet, respectively. Thus, in spite of high variability, the geometric means of the lead loadings
(plotted in Figure 8B-4) indicate that there is a two- to three-fold increase in the lead loadings
from pre- to post-activity samples. These results confirm the expectation that the settled dust lead
loading decreases (for both pre-activity and post-activity samples) as the distance from the
window increases, implying that proximity to the window influences baseline lead loadings as well
as the post-activity loadings.
8B-2.1.3.2 Vacuum Dust Samples from Stainless Steel Dustfall Collectors
In addition to the floor dust samples described above, vacuum dust samples were taken
from SSDCs as an alternative measure of lead disturbance. Immediately prior to the start of
window replacement, SSDCs were placed at 0 feet and at 6 feet from each window to be
removed, and dust samples were collected from them at one hour post-activity. At one window,
an additional SSDC was placed at 6 feet for sampling at two hours post-activity.
Tables WR-4a through WR-4d in Appendix A summarize physical weights, lead loadings,
and lead concentrations in samples from SSDCs collected at one hour and two hours following
completion of window replacement activities. The two-hour post-activity sample results were
compared with the one-hour results in Section 7.2 of Chapter 7. Also in Appendix A, boxplots of
lead loadings and concentrations for the one-hour post-activity dustfall collector samples are
presented in Figures WR-2a and WR-2b, respectively.
8-30
-------
1
1000000
100000
? 10000
1
I 1000
I 100
I
I 10
1
Stope=1
10 100 1000 10000 100000 1000000
Pro-Activity Floor Loading
Figure 8B-3. Lead Loadings for Post-Activity Versus Pre-Activity Floor Samples at 0, 3,
and 6 Feet from the Windows
4500
Pre-Activity ^ Post-Activityl
3 6
Distance from Window (ft)
Figure 8B-4. Geometric Mean Lead Loadings from Floor Surfaces as a Function of Distance
from the Window
8-31
-------
Lead loadings from dust collected on the SSDCs at 0 feet from the windows ranged from
3100 to 331,000 |ig/ft2, with a geometric mean of 24,700 |ig/ft2. At 6 feet from the windows, the
lead loadings among regular and side-by-side QC samples ranged from 15.4 to 4160 |ig/ft2, with a
geometric mean of 241 |ig/ft2.
As discussed in Chapter 7, the SSDC method yields a better measure of lead disturbed by
the window replacement activity than the difference between post- and pre-activity lead loadings.
Therefore, the SSDC measurements were used in creating an estimate of the lead disturbed by
window replacement that takes into account the "dustfall gradient," or the decrease in dust-lead as
distance from the activity increases. As introduced in Chapter 6, the dustfall gradient is
characterized by estimating the amount of lead (in jig) that settles within a 6-foot by 1-foot
rectangular region lying perpendicular to the window. Section C.5 of Appendix C presents the
approach to obtaining this estimate, its variability, and associated 95% confidence intervals.
Table 8B-3 presents estimates of the average amount of lead disturbed in the 6-foot by 1-
foot gradient. The table contains estimates across all units and for each unit. The average
estimated lead disturbed across all four units was 46,300 jig, with a 95% confidence interval of
(251; 92,300) jig. These results were obtained by a single fitting of the model in Section C.5 of
Appendix C to all dust-lead results across all units. Average total lead estimates for the
individual units ranged from 15,500 jig to 96,000 jig. Because only three data points per unit
were available to calculate average lead amounts, the results for individual units were based on
calculating a separate estimate of lead disturbance in the 6-foot by 1-foot gradient calculated for
each of the three windows in a unit and averaging these three estimates. This difference in
method from the approach used when considering all units did not allow confidence intervals to
be calculated for the individual units.
Table 8B-3. Estimates of the Average Amount of Lead Disturbed by Window Replacement
Activity in a 6' by 1' Rectangular Region Perpendicular to the Window
Unit ID
All 4 units
1-01
2-01
3-01
4-01
Lead Disturbed in a 6' by 1'
Gradient (//g)
46,300
15,500
51,800
96,000
66,000
95% Confidence Interval (//g)
Lower
251
Upper
92,300
Note: The methods for calculating lead amounts within each unit was slightly different from the method used to calculate
average lead amount across all units.
Further discussion of the dustfall gradient approach to comparing lead exposures among various
activities is given in Chapter 9.
8-32
-------
8B-2.2 ASSESS FACTORS OR MEASUREMENTS RELATED TO LEAD DISTURBANCE
To characterize the statistical relationship between lead exposure in various media and
external predictor variables, as well as unit-to-unit and within-unit variability in the lead exposure
data, statistical models were developed and fitted to lead loading data in this phase using analysis
of variance (ANOVA) techniques. The models were fitted to lead loading data across all study
units.
Appendix C contains a detailed presentation of the statistical models, including tables of
the parameter estimates and their standard errors. Models are fit to the personal-air, ambient-air,
and SSDC sample results as well as to the estimates of lead disturbed in a 6-foot by 1-foot
gradient from a removed window. Because pre-activity floor lead loadings often exceed post-
activity floor loadings, floor loading data were not modeled. Covariates considered include lead
loadings from all of the pre-activity samples (floor dust, window well dust, and paint chip) taken
during the window replacement phase.
In the window replacement phase, model fittings were performed on relatively small
numbers of data points. Thus, only one or two covariates were considered at any one time in
modeling lead loadings. In addition, because the few data points had relatively high variability,
the tests for significant covariate effects tended to have low power. Therefore, findings of
statistically insignificant results may indicate no effect or merely little power to detect the effect.
Conversely, the relatively large number of separate model fittings required by the small number of
data points leads to a concern over the effect of multiple tests on the true significance level of
statistically significant parameters. Due to the exploratory nature of the model fitting in this
phase, alpha levels for statistical significance were not adjusted for the effect of multiple tests.
ANOVA models also characterized error in log-transformed lead loading data from three
sources: unit-to-unit variability, within-unit (window-to-window) variability, and replicate-to-
replicate variability at a window. Unit-to-unit variability is estimable because sampling was done
at four study units. Within-unit variability is estimable because samples were collected at three
windows within each study unit. For some sample types, replicate-to-replicate variability,
representing variability in sampling at the same physical location, was estimable due to taking two
side-by-side samples at the same distance from a window.
A summary of the results of pre-activity vacuum dust samples from window well surfaces
is presented in Section 8B-2.2.1. Results of the model fits on during- and post-activity sample
data are presented in Section 8B-2.2.2. In addition, the relationship between dustfall lead
loadings and the distance from the windows is examined using the results obtained from SSDCs.
The results of fitting the model relating lead exposure to distance are contained in Section 8B-
2.2.3. Estimates of variance components for the log-loading data are presented in Section 8B-
2.2.4.
8-33
-------
8B-2.2.1 Vacuum Dust Samples from Window Well Surfaces Taken Pre-Activity
Pre-activity vacuum dust samples were collected from the window wells selected in each
unit. The samples were collected to help characterize the units with respect to the lead levels
found prior to window replacement activities, and to use as potential predictors of lead exposure
resulting from window replacement.
Tables WR-5a and WR-5b in Appendix A present data summaries for the pre-activity
window well dust samples. One sample was eliminated from the statistical analysis because its
observed lead concentration and loading were judged to be outliers (significantly smaller than and
inconsistent with the results of the other window well samples). Summary statistics shown in
Tables WR-5a and WR-5b are calculated both with and without the outlier. Figures WR-la and
WR-lb in Appendix A compare the lead loadings and concentrations, respectively, of the pre-
activity window-well samples with the pre-activity floor samples using boxplots.
The lead loadings of the window well dust samples ranged from 26,800 to 415,000 |ig/ft2
(after removal of the outlier), with a geometric mean of 135,000 |ig/ft2.
8B-2.2.2 Lead Disturbance as a Function of Pre-Activity Lead Loadings
Model (C-l) presented in Section C. 1 of Appendix C was fitted to data from the following
sample types collected during or after window replacement activities:
• lead concentrations in personal exposure samples
• lead concentrations in ambient air samples
• lead loadings in SSDC dust samples.
In addition, Model (C-l) was fitted to the estimated amount of lead disturbed in a 6-foot by 1-
foot gradient, discussed in Section 8B-2.1.3.2. Each fit of these models contained random effects
representing unit-to-unit and within-unit variability (no side-by-side QC sample results were
included), as well as one of the following fixed covariates:
#1. pre-activity lead loading on floor surfaces at 0 feet from the window.
#2. pre-activity lead loading on floor surfaces at 3 feet from the window.
#3. pre-activity lead loading on floor surfaces at 6 feet from the window.
#4. pre-activity lead loading on window well surfaces.
#5. lead loadings within paint chips collected from the interior sash/frame of the window.
#6. lead loadings within paint chips collected from the exterior sash/frame of the
window.
#7. pre-activity ambient air lead concentrations.
Table 8B-4 summarizes the approach to fitting Model (C-l), indicating those covariates whose
effects on the data were significant at the 0.05 level. Model (C-l) takes the form of models (WR-
1) through (WR-5) when fitting to a specific data type (Table C-2 of Appendix C). Only one of
the covariates was consistently significant when modeling the different sample types: pre-activity
8-34
-------
lead loading on floor surfaces at 3 feet from the window (covariate #2). Ambient air lead
concentrations during the activity were significantly associated with pre-activity floor lead
loadings at each distance from the windows, indicating that ambient air results were likely
dominated by the pre-activity level of dirtiness existing within the study units.
Table 8B-4. Results of Tests for Significant Covariates in the Model Fitting to Lead
Loading Data in the Window Replacement Phase
Modeled Data
Personal exposure
sample concentrations
Ambient air sample
concentrations (during
activity)
SSDC dust-lead sample
loadings at 0 feet from
window
SSDC dust-lead sample
loadings at 6 feet from
window
Estimated lead amount
in 6'x1 ' gradient from
window
Models Fitted
(Table C-2 of Appendix C)
Model (WR-1) fit seven times, each
considering one of the seven covariates.
Model (WR-2) fit six times, each
considering one of covariates #1 to #6,
with covariate #1 always included in the
model.
Model (WR-3) fit six times, each
considering one of covariates #1, #2, #4,
#5, #6, #7.
Model (WR-4) fit six times, each
considering one of covariates #2 to #7.
Model (WR-5) fit seven times, each
considering one of the seven covariates.
Covariates Significant at
the 0.05 Level in the Fitted
Models
Covariate #2
Covariates #1, #2, #3
Covariate #2
None
Covariate #2
The significance levels for the tests of significant covariate effects were not adjusted for the large
number of tests performed in this exercise.
More detailed results of the model fittings are presented in Section C.I of Appendix C.
8B-2.2.3 Lead Disturbance as a Function of Distance from the Windows
Lead disturbance resulting from renovation and remodeling activities was expected to vary
with the distance from the activity. Field sampling within the window replacement phase of the
EFSS was designed so that this relationship could be explored. Floor dust samples were collected
(pre- and post-activity) at 0, 3, and 6 feet from the windows being replaced. SSDCs were placed
at 0 and 6 feet from the same windows. The sample results can be used to determine how lead
loadings change over distance.
8-35
-------
In Section 8B-2.1.3.1, a summary of the results of the pre-activity and post-activity floor
dustfall samples was presented. The difference between post- and pre-activity floor lead loadings
represented lead disturbance resulting from window replacement activities. These observed
differences were not well-behaved: nine of the 35 differences were negative, and the variability of
the differences was very large (Figure 8B-3).
Figures 8B-5a and 8B-5b are plots of the lead loadings from pre-activity and post-activity
floor dust samples, respectively, versus the distance from the windows at which the samples were
collected. The plotting symbols indicate the unit from which the samples were collected. Lines
are drawn connecting observations taken at the same window. In both cases, measured lead
loadings tended to decrease as distance from the windows increased. Figure 8B-5c plots the
difference in lead loadings between adjoining post- and pre-activity floor samples versus the
distance at which the samples were collected. This figure clearly shows the effect of the negative
differences and the large variability in the differences. For this reason, statistical models in the
previous section were not fitted to floor sample lead loadings.
Figure 8B-6 is a plot of lead loadings for the one-hour post-activity samples taken from
SSDCs versus the distance from the windows at which the samples were collected. Although
only two distances per window were considered, this figure indicates that lead loadings from
SSDC dustfall samples have a more consistent decreasing trend in the lead loading as the distance
from the window increases, compared to trends seen in the other plots for floor dustfall samples.
Model (C-3) in Section C.3 of Appendix C was fitted to evaluate the relationship between
one-hour post-activity lead loading data (|ig/ft2) from SSDCs and distance from the window being
removed. The estimated slope parameter associated with distance was -0.732, which was
statistically significantly different than 0 at the 0.0001 level. This result indicates that as one
moves one foot farther away from the window, the log-transformed lead loading decreases by
approximately 0.732, or the untransformed lead loading is reduced by approximately (l-e"°732) *
100% = 52%.
Because of the significant distance effect on SSDC lead loadings, it was decided that the
amount of lead disturbed in a 6-foot by 1-foot gradient extending from the window would be a
better overall estimate of the lead disturbed by window replacement activities than estimates made
at a single distance.
8-36
-------
1000000
100000
10000
I
, 1000
100
10
1
01234567
Distance from Window (feet)
Figure 8B-5a. Lead Loadings (//g/ft2) for Pre-Activity Floor Samples Versus Distance from
the Windows at Which the Samples Were Taken
1000000
100000
10000
1000
100
10
1
01234567
Distance from Window (feet)
Figure 8B-5b. Lead Loadings (//g/ft2) for Post-Activity Floor Samples Versus Distance from
the Windows at Which the Samples Were Taken
8-37
-------
100000
-100000
-200000
-300000
-400000
01234567
Distance from Window (feet)
Figure 8B-5c. Lead Loadings (//g/ft2) for Differences Between Adjoining Post-Activity and
Pre-Activity Floor Samples Versus Distance from the Windows at Which the
Samples Were Taken
1000000
100000
10000
1000
100
10
01234567
Distance from Window (feet)
Figure 8B-6. Lead Loadings (//g/ft2) for One-Hour Post-Activity SSDC Samples Versus
Distance from the Windows at Which the Samples Were Taken
8-38
-------
8B-2.2.4 Estimating Variance Components
The sampling design for the window replacement phase specified sample collection over
multiple study units, multiple locations (or workers) within a study unit, and that settled dust
samples be taken in multiple areas within locations (i.e., side-by-side QC samples). This design
provided for key sampling components of total data variability to be isolated and characterized.
Therefore, when estimable, the magnitudes of the following variance components were estimated
for several sample types (where appropriate) using random-effects analysis of variance:
• "unit-to-unit" variability
• "location-to-location" (or worker-to-worker) variability within a unit (window-to-
window)
• "replicate-to-replicate" variability within a location.
Variance components were estimated for the following lead measurements:
• pre-activity floor dust lead loadings (separately at 0, 3, 6 feet)
• post-activity floor dust lead loadings (separately at 0, 3, 6 feet)
• 1-hour post-activity SSDC vacuum lead loadings (separately at 0, 6 feet)
• pre-activity window well dust lead loadings
• paint chip lead loadings (separately for interior, exterior)
• personal air lead concentrations
• pre-activity ambient air lead concentrations
• during-activity ambient air lead concentrations
• lead amounts in a 6' x 1' gradient.
Replicate-to-replicate variability could only be estimated for the lead measurements in the first
three items listed, because these were the only sample types where side-by-side QC samples were
taken.
It is assumed that total variability is the sum of the variabilities associated with the three
components listed above. Any additional variance components are confounded within one or
more of these components.
Model (C-2) presented in Section C.2 of Appendix C was fitted to the lead measurements
to estimate the variance components for the various sample types. Table 8B-5 contains the
estimated total variability in these (log-transformed) measurements, plus estimates of the
components of the total variability. The results represent variability in the log-domain, so they are
expressed in log units.
8-39
-------
Table 8B-5. Estimates of Total Variability and its Estimable Components in Log-Transformed Lead Measurements
By Sample Type in the Window Replacement Phase
Sample Type111
Pre-act. Floor Dust at 0 feet
Pre-act. Floor Dust at 3 feet
Pre-act. Floor Dust at 6 feet151
Post-act. Floor Dust at 0 feet
Post-act. Floor Dust at 3 feet
Post-act. Floor Dust at 6 feet151
1-hr. Post-Activity SSDC Dust (vacuum) at 0 feet
1-hr. Post-Activity SSDC Dust (vacuum) at 6 feet151
Pre-Activity Window Well
Interior Paint chip
Exterior Paint chip
Personal air
Pre-Activity Ambient air
Post-Activity Ambient air
Six-foot by One-foot Gradient
Model Estimate
of Mean Log
Measurement121
7.56
6.04
5.79
8.27
7.11
6.76
10.1
5.4
11.5
0.79
2.40
2.01
-1.21
0.15
10.8
Square Root of
Estimated Total
Variability131: o,0,
3.92
3.43
3.36
3.26
2.66
3.27
1.67
1.86
1.54
2.54
1.50
1.26
1.58
0.86
1.56
Square Roots of Estimated Variance Components141
Unit-to-unit (ou)
3.46 (77.8%)
3.23 (88.6%)
3.21 (91.5%)
2.90 (79.0%)
2.38 (80.4%)
3.11 (90.8%)
0 (0%)
1.14 (37.6%)
0.99 (41.6%)
2.46 (93.5%)
1.02 (46.5%)
1.11 (77.6%)
1.33 (71.4%)
0.72 (69.5%)
0 (0%)
Location-to-
location within
a unit (oj
1.52 (15.0%)
0.68 (3.9%)
0.60 (3.2%)
1.05 (10.4%)
0.31 (1.4%)
0.73 (5.0%)
1.52 (83.7%)
0.93 (24.8%)
1.17 (58.4%)
0.65 (6.5%)
1.10 (53.5%)
0.60 (22.4%)
0.84 (28.6%)
0.47 (30.5%)
1.56 (100%)
Replicate-to-
replicate within
a location (OR)
1.05 (7.2%)
0.94 (7.5%)
0.78 (5.4%)
1.06 (10.5%)
1.13 (18.2%)
0.67 (4.2%)
0.67 (16.3%)
1.14 (37.5%)
___
___
___
___
___
___
—
CO
o
(1) SSDC = Stainless steel dustfall collectors.
(2) Estimate of the intercept term from the random effects model.
(3) Total variability = otot2 = au2 + OL2 + OR2, where these parameters were estimated from the random effects model using restricted maximum likelihood.
(4) Number in parentheses is the percent of total variability (q2,,) represented by the given variance component (the square of the tabled value).
(5) Side-by-side samples were included with the floor dust and SSDC dust samples when estimating variance components.
-------
For pre-activity and post-activity floor samples at all three distances, over 75% of
variability in the lead loadings was attributed to results from different units; there was no clear
pattern in the within-unit and replicate-to-replicate variabilities. For samples collected from
SSDCs, the results were mixed: at 6 feet from the windows, all three sources of variability made
roughly equal contributions to the total variability; at 0 feet from the windows, within-unit
variability was about five times as large as the replicate-to-replicate variability. The total
variability of both pre-activity and post-activity floor sample data exceeds that of the SSDC
samples.
The estimated within-unit variance of window well loadings was about 1.5 times as large
as the estimated unit-to-unit variance. The unit-to-unit variance among the interior paint-chip
loadings was 14 times greater than the within-unit variability, while the unit-to-unit and within-
unit variabilities were roughly equal for exterior paint chip loadings. The unit-to-unit variability in
log-transformed personal air lead concentrations was four times greater than the within-unit
variability, indicating that workers involved in window replacement at a given site are exposed to
similar lead levels within their breathing zones. The results for ambient air samples are similar to
those of personal exposure samples, with the unit-to-unit variability 2.5 times greater than within-
unit variability for both pre-activity and post-activity samples.
8B-2.3 CORRELATIONS BETWEEN LEAD IN DIFFERENT SAMPLE MEDIA
To quantify the extent of the linear relationship between lead exposure estimates from
different sample types and sample collection approaches, Pearson correlation coefficients were
calculated among pairs of twelve variables that measure lead levels in various media through the
study. The variables of interest include:
• pre-activity dust lead loadings on floors (0, 3, 6 feet)
• pre-activity dust lead loadings on window wells
• 1-hour post-activity dust lead loadings on SSDCs (0, 6 feet)
• 2-hour post-activity dust lead loadings on SSDCs (6 feet)
• estimates of total lead disturbed in a 6-foot by 1-foot gradient
• lead concentrations in personal air and ambient air samples collected during window
replacement activity
• lead loadings in interior and exterior paint chips from window sashes and frames.
Post-activity floor dust lead loadings were not included in this analysis because of the previous
decision to use SSDC measurements as the preferred indicator of lead generated in settled dust by
the activity (see Chapter 7). For each variable, the geometric mean of the observed data was
calculated for the four study units. The estimated total lead in the 6-foot by 1-foot gradient was
8-41
-------
calculated as discussed in Section 8B-2.1.3.2. Pearson correlation coefficients associated with
these geometric means were calculated for each pair of variables. Table 8B-6 presents these
correlations. A low correlation between two variables does not necessarily imply the lack of a
relationship, as the relationship may not be linear in nature or the sample size may have not have
been large enough to detect such a relationship.
Because each of the correlations was calculated using only four observations, only
correlations that are greater than 0.95 are statistically significantly different from 0 at the 0.05
level. Correlation coefficients greater than 0.9 are statistically significantly different from 0 at the
0.10 level.
Table 8B-6 shows several strong correlations that are of particular interest:
• The strong correlations between personal air lead concentrations, lead loadings in
one-hour post-activity SSDC samples at 0 feet from the windows, ambient air lead
concentrations, and the estimated lead amount disturbed in a 6-foot by 1-foot
gradient indicate a probable relationship between a worker's exposure and the lead
generated as a potential exposure to occupants. This suggests that a single medium
may be considered in future studies as an indicator of both worker exposure and
potential occupant exposure
• The strong correlation between dust lead loadings from pre-activity floor samples at 0
feet from the windows and lead loadings in interior paint chip samples suggests that
the high levels of lead in dust on the floor near the windows may be due to high levels
of lead in paint on the interior surfaces of the windows rather than from lead tracked
or blown into the unit.
A strong correlation between personal worker lead concentrations and post-activity SSDC lead
loadings was also noted in the carpet removal phase. While correlation between personal worker
exposure, floor dust loadings, and ambient air concentrations was somewhat lower in the carpet
removal phase than that observed for window replacement, the small numbers of data points may
imply no statistical difference between phases.
8B-2.4 SUMMARY OF RESULTS
Following is a summary of the major results observed in the EFSS window replacement
phase:
• None of the lead concentrations within personal air samples exceeded the OSHA PEL
of 50 |ig/m3. Personal air lead concentrations ranged from 2.41 to 44.3 |ig/m3 with a
geometric mean of 7.48 |ig/m3. Estimated variability in these data were primarily the
result of sampling in different study units.
8-42
-------
Table 8B-6. Pearson Correlation Coefficients'11 of the Geometric Mean Lead Loadings Between Pairs of Sample
Types and Approaches in the Window Replacement Phase
Sample Type
Pre Floor D = 0
Pre Floor D = 3
Pre Floor D = 6
Window Well
Paint Chip Ext.
Paint Chip Int.
1-Hr. SSDC D = 0
1-Hr. SSDC D = 6
2-Hr. SSDC D = 6
Ambient Air
Personal Exp.
Pre Floor
D = 0(21
1.00
Pre Floor
D = 3
0.847
1.00
Pre Floor
D = 6
0.999
0.827
1.00
Window
Well
0.129
0.267
0.119
1.00
Paint
Chip Ext.
-0.495
-0.174
-0.511
0.769
1.00
Paint
Chip Int.
0.903
0.754
0.902
-0.296
-0.772
1.00
1-Hr.
SSDC
D=1
0.463
0.829
0.433
0.618
0.395
0.260
1.00
1-Hr.
SSDC
D = 6
0.744
0.292
0.767
0.101
-0.513
0.596
-0.110
1.00
2-Hr.
SSDC
D = 6
0.836
0.416
0.855
-0.050
-0.664
0.762
-0.062
0.972
1.00
Ambient
Air
0.420
0.832
0.387
0.452
0.309
0.302
0.978
-0.229
-0.141
1.00
Personal
Exp.
0.420
0.838
0.386
0.327
0.214
0.360
0.941
-0.264
-0.147
0.991
1.00
6' x V
Lead
Gradient
0.655
0.892
0.632
0.662
0.271
0.402
0.961
0.169
0.202
0.901
0.851
CO
CO
(1) Correlations calculated on geometric means calculated for each study unit (n =4). All correlations greater than 0.95 are significantly different from zero at the
0.05 level.
(2) Signifies distance from window (in feet).
-------
Lead concentrations within ambient air samples collected during window replacement
activity ranged from 0.29 to 4.16 |ig/m3 with a geometric mean of 1.16 |ig/m3.
The difference in lead loadings from floor dust samples taken pre- and post-activity
had high variability in this study. Across all floor dust samples, the geometric mean
lead loading increased from 2.0 to 2.6 times from pre- to post-activity at the various
distances from the window. Some pre-activity sample lead loadings were actually
larger than the loadings in the adjoining post-activity samples.
Lead loadings from SSDC samples had substantially lower variability than the
difference between post- and pre-activity floor lead loadings. This indicates that the
SSDC method may yield a more accurate and precise measure of the lead disturbed
by window replacement than the difference in pre- and post-activity floor dust
samples. The geometric mean of the lead loadings collected from SSDCs one hour
after completion of the window replacement was 24,700 |ig/ft2 at 0 feet from the
window and 241 |ig/ft2 at 6 feet from the window, indicating the significant effect of
distance from the activity on observed load levels. An estimate of the total amount
of lead disturbed in a 6-foot by 1-foot gradient was calculated for each window. The
estimate of the average amount of lead disturbed in a 6-foot by 1-foot gradient for
window replacement activities in this study was 46,200 jig.
Of the covariates providing information on pre-activity lead levels, only the pre-
activity lead loadings on floor surfaces at 3 feet from the window was consistently
significant at the 0.05 level when modeling during- and post-activity lead disturbance
in various media. Small sample size, high variability and multiple model fittings make
these results very tenuous, however, and only exploratory in nature.
The strong correlations between the lead loadings from personal worker exposure
samples, one-hour post-activity SSDC samples at 0 feet from the windows, area air
samples, and the estimated lead disturbed in a 6-foot by 1-foot gradient indicates a
probable relationship between a worker's exposure and the lead generated as a
potential exposure to the occupants.
8-44
-------
8C-1.0 STUDY DESIGN IN THE CED PHASE
This section presents the study design for the Controlled Experimentally-Designed (CED)
phase of the EFSS study. The primary objective of this phase was to estimate lead disturbances
and potential lead exposures associated with several target R&R activities and generic activities
which are difficult to isolate in an actual R&R job or are ingredients to larger R&R activities. In
addressing this objective, the CED phase offered a controlled, simulated R&R environment in
which R&R activities of interest were performed and environmental samples were collected.
The activities in the CED phase were classified into one of the following three primary
categories:
1. simulated R&R target activities (e.g., demolition)
2. generic R&R tasks (e.g., drilling and sawing)
3. a positive control activity (paint removal, for comparison purposes).
The components of each activity category are discussed in Section 8C-1.1.
The CED phase was conducted using skilled R&R workers and laborers who were
familiar with protective measures and methodologies used in the lead-abatement industry. These
R&R workers were contracted by the project team to perform the designated R&R tasks within
each study component, and they were instructed to perform the tasks as they are typically
conducted in an unregulated environment. The workers were protected according to the
procedures approved by the human subjects review committees of Battelle, MRI, and EPA.
The CED phase consisted of three "case studies" (or "sites"), each constituting one or
more related vacant buildings slated for subsequent gutting, gut rehab, or complete abatement and
clearance testing before reoccupancy. Two sites, each consisting of a single row house, were
located in Baltimore, Maryland. The third site consisted of a group of four dwelling units in
Denver, Colorado. It was expected that the unique characteristics of each site would cause
slightly differing approaches to applying the overall study and sampling design for each activity
occurrence. This was reflected in the decision to develop a separate sampling plan for each site.
Nevertheless, for the most part, the different R&R tasks within each study component were able
to be replicated across sites and were often replicated within sites. This allows for an assessment
of not only the potential lead exposure associated with each different CED activity, but also of the
variance components associated with the measures of those lead exposures.
8C-1.1 SAMPLING DESIGN FOR THE CED PHASE
The carpet removal and window replacement phases of the EFSS study (Sections 8A and
8B) were designed as observational studies in which only the types and locations of the samples
were specified in the sampling design. In contrast, the CED phase, as its title suggests, was
designed to control not only the type and location of samples, but also the type and conduct of
R&R activity.
8-45
-------
The sections which follow contain discussions of the two aspects of the sampling design
for the CED phase: descriptions of the activities that were included in the study, and the types
and locations of samples that were collected to monitor lead exposure as a result of performing
the activities.
8C-1.1.1 CED Activities
As mentioned above, three primary categories of R&R activities were included in the CED
phase: simulated R&R target activities, generic R&R activities, and a positive control activity.
In an attempt to replicate each R&R activity across the different "sites" in the study, the
actual work that was to be performed as part of the activity was defined prior to the start of data
collection. Because each study site was unique, separate test plans for each study site were also
prepared that presented specifications of where, when, and how each activity was to be performed
at each of the three sites. Following is a general description of activities conducted within each
activity category in the CED phase (more detailed descriptions are provided in the CED QAPjP
addendum).
Category #1: Simulated R&R Target Activities
1. Demolition In each study unit, three large structure removal activities were planned.
These activities consisted of demolishing one or more walls in the study units. In
most cases, a single plaster wall was removed. However, in some situations, one or
more walls consisting of drywall, wood, or plaster were removed. Removal of walls
that separated rooms were limited to one room, and the activity was planned so that
there was no damage to the wall of the room backing the wall that was demolished.
Removal of exterior walls was planned so that there was no damage to the underlying
structure of the wall (usually brick). In all cases, the end result of the demolition was
to be an exposed wall structure completely ready for the installation of new drywall.
2. HVAC Removal. In each study unit, one HVAC repair/ replacement activity was
planned. This activity consisted of removing several sections of the HVAC
ductwork. The activity was designed to prepare the HVAC system for the
installation of new ductwork. (Actual HVAC removal was performed only at the
Baltimore units).
3. Small Surface Disruption. The CED QAPjP Addendum lists several examples of
activities that fit into the category of small surface disruptions. The activity that was
chosen for the CED phase was Door Modification. This activity consisted of two
subtasks — trimming wood from the edge of an existing door (and sanding it
smooth), and drilling a hole for installation of a doorknob — which were designed to
mimic actual work that might be performed as part of an R&R job. Two door
modification activities were scheduled for each unit. Because it was expected that
the time required to perform the modification of a single door would be too short to
provide for reasonable samples, several doors were included in the door modification
8-46
-------
activity. The activity consisted of trimming an inch from one end of the door (top or
bottom), drilling a hole for a doorknob, and trimming an inch from the other end of
the door. The choice of tools for performing this activity was left to the discretion of
the worker. (Note: When this activity was conducted in the field, workers lost track
of the number of doors modified. Therefore, exposure results are presented as a
generic R&R task involving cutting and sanding rather than as an example of the
small surface disruption target activity.)
Category #2: Generic R&R Activities
An objective of the CED phase was to determine lead exposures associated with several
generic components of larger R&R activities. The planned generic activities were as follows:
1. Sawing. The sawing activities involved making a series of fifteen parallel cuts (5 feet
in length) into either plaster or wood substrates using either a rough blade or a fine
blade. Four activities were planned, using each combination of substrate and blade.
The cuts into plaster walls were separated by three inches in the vertical direction,
and the cuts into wood baseboards removed from walls were made at one-inch
intervals.
2. Drilling. The drilling activities involved drilling a lattice of holes into either plaster
or wood substrates using either a small or large drill bit. Four activities were
planned, using each combination of substrate and drill bit. Wood drilling was done
into walls (when available) or into doors, and plaster drilling was done into walls.
The lattice of holes for all drilling activities was 3 feet wide and 2 feet high (or vice-
versa). With the small drill bit, holes were drilled every inch in both horizontal and
vertical directions (925 holes), and with the large drill bit, holes were drilled every 3
inches (117 holes).
3. Building Component Removal. The CED QAPjP Addendum recognized the
possibility of monitoring additional R&R activities if the opportunity presented itself.
In the first dwelling unit in Denver, different building components needed to be
removed prior to conducting R&R activities on them. This opportunity allowed the
removal of lead-painted wood trim, baseboards, doors, and door-jams to be
monitored in this unit as a simulated R&R target activity.
4. Cleanup, Cleanup activity was monitored by personal air concentrations only (as
opportunity allowed). New filter cassettes were inserted in the personal exposure
monitors at the start of cleanup activity, so that cleanup exposures could be clearly
separated from exposures resulting from the earlier R&R activity.
The first two activities also provided a basis for comparison of different tools and different
substrates with regard to their effect on lead exposure.
8-47
-------
Category #3: Positive Control Activity
Paint removal was conducted at each CED study unit. This activity was performed at
one location per unit where two or three windows were available. Potential lead exposure
associated with paint removal has been studied and cited in the literature in detail; thus paint
removal acted as a positive control activity to which results of the other CED activities could be
compared. Paint was removed from the windows using abrasive techniques, thereby acting as a
"worst-case" positive control. Half of the available area had paint removed by hand scraping
and sanding while the remaining portion had paint removed using power sanding.
In all activities performed in the EFSS, vacuum attachments were not used on any tools.
In addition, no dust reduction methods, such as wetting, misting, or negative air, were employed.
Before any of the activities at a given building were performed, the building was prepared
according to normal abatement procedures, including creation of a decontamination area, tool
cleanup area, sealing of floors and rooms with poly, and other safety and activity preparations.
8C-1.1.2 Sample Types and Locations
The inclusion of different R&R activities at different sites in the CED phase required some
flexibility in sampling plans across activities and sites. In general the sampling plan for the
collection of environmental field samples was consistent within each type of CED activity.
Wherever possible, sampling was done in a consistent manner across CED activities. This section
presents the sampling plan and protocols that were consistent across activities and sites. Table
8C-1 presents the specific implementation of the sampling plan for each site and activity in the
CED phase.
The primary types of samples taken in the CED study included the following:
• Personal exposure samples taken during the activity
• Paint chip samples taken before the activity from plaster and painted-wood
surfaces disturbed by the activity
• Plaster samples taken before the activity from plaster walls used for the activity, to
measure the amount of lead in the substrate which can be disturbed during the
activity
• Settled dust samples taken via wipe techniquesyrom HVAC components, to
measure the amount of lead which can be disturbed during HVAC removal
• Settled dust samples taken via vacuum or wipe techniques following completion of
the activity from stainless steel dustfall collectors (SSDCs) placed on the floor
immediately prior to the activity, and at areas adjacent to and at varying distances
away from the activity.
8-48
-------
Table 8C-1. Sampling Design for Each CED Activity Within Each Study Unit
City/
Building No./
Activity No.
Baltimore
Building 1
Activity 1
Baltimore
Building 1
Activity 2
Baltimore
Building 1
Activity 3
Baltimore
Building 1
Activity 4
Baltimore
Building 1
Activity 5
Baltimore
Building 1
Activity 6
Baltimore
Building 1
Activity 7
Baltimore
Building 1
Activity 9
Baltimore
Building 1
Activity 10
Baltimore
Building 1
Activity 1 1
Baltimore
Building 1
Activity 1 2
Baltimore
Building 1
Activity 1 3
Baltimore
Building 1
Activity 14
Baltimore
Building 1
Activity 1 5
Baltimore
Building 1
Activity 81
Baltimore
Building 1
Activity 82
Activity
Drilling Wood
925 Holes with Yt" Bit
Drilling Plaster
925 Holes with % " Bit
Drilling Wood
1 17 Holes with 1 " Bit
Drilling Plaster
1 17 Holes with 1" Bit
Sawing Plaster
40 Feet with Circular Saw
Sawing Wood
75 Feet with Circular Saw
(Rough Blade)
Door Modification
Sawing Wood
75 Feet with Circular Saw
(Smooth Blade)
Door Modification
Doors from Baltimore 121
HVAC Removal
Abrasive Paint Removal
Demolition of East Wall
Demolition of West Wall
Demolition of Closet
Drilling Wood
925 Holes with % " Bit
Drilling Plaster
925 Holes with 1/4 " Bit
Day/Time
Day 1-1
07:16 - 08:10
Day 1-2
10:17 - 11:42
Day 1-1
15:34 - 15:54
Day 1-2
10:34 - 10:51
Day 1-1
08:04 - 08:34
Day 1-1
14:28 - 15:14
Day 1-2
08:08 - 09:39
Day 1-1
09:04 - 09:38
Day 1-1
14:18 - 15:30
Day 1-2
08:24 - 08:46
Day 1-2
14:30 - 15:53
Day 1-3
08:12 - 08:41
Day 1-3
09:34 - 11:28
Day 1-3
13:37 - 14:01
Day 1-3
15:18 - 15:59
Day 1-1
08:33 - 08:48
Worker
1A
1A
1B
1B
1B
1B
1A
1B
1A
1B 1C
1A 1B
1A 1B 1C
1A 1B
1A 1B
1A
1A
Room
Description
South Basement
2nd Floor Bathroom
North Basement
North Bedroom
2nd Floor
Foyer
South Bedroom
3rd Floor
North Bedroom
3rd Floor
South Bedroom
3rd Floor
North Bedroom
3rd Floor
Basement
South Bedroom
2nd Floor
Living Room
North Bedroom
2nd Floor
Basement
North Bedroom
3rd Floor
2nd Floor Bathroom
Samples
Collected
1 Personal Exposure
5 Settled Dust
1 Paint Chip
1 Personal Exposure
5 Settled Dust
1 Paint Chip/1 Plaster
1 Personal Exposure
5 Settled Dust
1 Paint Chip
1 Personal Exposure
5 Settled Dust
1 Paint Chip/1 Plaster
1 Personal Exposure
5 Settled Dust
1 Paint Chip/1 Plaster
1 Personal Exposure
5 Settled Dust
2 Paint Chip
1 Personal Exposure
5 Settled Dust
2 Paint Chip
1 Personal Exposure
5 Settled Dust
2 Paint Chip
1 Personal Exposure
5 Settled Dust
2 Paint Chip
2 Personal Exposure
10 Settled Dust
4 Interior Duct Wipes
3 Personal Exposure
8 Settled Dust
3 Paint Chip
3 Personal Exposure
10 Settled Dust
1 Paint Chip/1 Plaster
2 Personal Exposure
10 Settled Dust
1 Paint Chip/1 Plaster
2 Personal Exposure
10 Settled Dust
2 Paint Chip/1 Plaster
1 Personal Exposure
5 Settled Dust
1 Paint Chip
1 Personal Exposure
8-49
-------
Table 8C-1 (Continued)
City/
Building No./
Activity No.
Baltimore
Building 2
Activity 1
Baltimore
Building 2
Activity 2
Baltimore
Building 2
Activity 3
Baltimore
Building 2
Activity 4
Baltimore
Building 2
Activity 5
Baltimore
Building 2
Activity 6
Baltimore
Building 2
Activity 8
Baltimore
Building 2
Activity 9
Baltimore
Building 2
Activity 10
Baltimore
Building 2
Activity 1 1
Baltimore
Building 2
Activity 1 2
Baltimore
Building 2
Activity 1 3
Baltimore
Building 2
Activity 14
Baltimore
Building 2
Activity 1 5
Denver
Building 3
Activity 1
Denver
Building 3
Activity 2
Denver
Building 3
Activity 3
Activity
Drilling Wood
925 Holes with Yt" Bit
Drilling Plaster
925 Holes with Yt" Bit
HVAC Removal
Drilling Wood
1 17 Holes with 1" Bit
Drilling Plaster
1 17 Holes with 1" Bit
Sawing Wood
75 Feet with Circular Saw
(Rough Blade)
Door Modification
Sawing Plaster
47.5 Feet with Sawzall
Sawing Wood
75 Feet with Circular Saw
(Smooth Blade)
Door Modification
Abrasive Paint Removal
Demolition of
North and South Walls
Demolition of
North and South Walls
Demolition of East Wall
Building Component
Removal
HVAC Sampling
Demolition of West Wall
Day/Time
Day 2-1
09:11 - 10:34
Day 2-1
11:00 - 12:23
Day 2-1
14:30 - 15:44
Day 2-1
13:42 - 14:02
Day 2-1
15:05 - 15:26
Day 2-1
08:23 - 08:52
Day 2-1
08:08 - 09:39
Day 2-1
15:37 - 15:59
Day 2-1
11:06 - 11:35
Day 2-1
12:59 - 13:42
Day 2-2
08:14 - 10:24
Day 2-3
07:43 - 09:21
Day 2-2
13:15 - 14:02
Day 2-2
10:30 - 11:37
Day 3-1
09:11 - 11:52
12:59 - 13:58
Day 3-1
09:41 - 10:00
Day 3-1
14:16 - 15:47
Worker
2A
2A
2B 2C
2A
2A
2B
2A
2A
2B
2B
2A 2B
2B 2C
2A 2B
2A 2B
3A 3B
MRI
3A 3B
Room
Description
Kitchen
Bathroom
3rd Floor
Front Bedroom
2nd Floor
Kitchen
Bathroom
3rd Floor
Rear Bedroom
3rd Floor
Center Bedroom
3rd Floor
Bathroom
2nd Floor
Rear Bedroom
2nd Floor
Rear Dining Room
1 st Floor
Front Bedroom
3rd Floor
Bathroom
2nd Floor
Bathroom
3rd Floor
Kitchen
Front and Center
Bedrooms
Attic
Kitchen
Samples
Collected
1 Personal Exposure
5 Settled Dust
1 Paint Chip
1 Personal Exposure
5 Settled Dust
1 Paint Chip/1 Plaster
2 Personal Exposure
5 Settled Dust
3 Interior Duct Wipes
1 Personal Exposure
5 Settled Dust
1 Paint Chip
1 Personal Exposure
5 Settled Dust
1 Paint Chip/1 Plaster
1 Personal Exposure
5 Settled Dust
2 Paint Chip
1 Personal Exposure
5 Settled Dust
3 Paint Chip
1 Personal Exposure
5 Settled Dust
1 Paint Chip/1 Plaster
1 Personal Exposure
5 Settled Dust
2 Paint Chip
1 Personal Exposure
5 Settled Dust
3 Paint Chip
3 Personal Exposure
8 Settled Dust
3 Paint Chip
2 Personal Exposure
8 Settled Dust
3 Paint Chip/1 Plaster
2 Personal Exposure
8 Settled Dust
2 Paint Chip/2 Plaster
2 Personal Exposure
8 Settled Dust
2 Paint Chip/1 Plaster
2 Personal Exposure
5 Settled Dust
5 Interior Duct Wipes
2 Personal Exposure
6 Settled Dust
1 Paint Chip/1 Plaster
8-50
-------
Table 8C-1 (Continued)
City/
Building No./
Activity No.
Denver
Building 3
Activity 4
Denver
Building 3
Activity 5
Denver
Building 4
Activity 6
Denver
Building 4
Activity 7
Denver
Building 4
Activity 9
Denver
Building 4
Activity 10
Denver
Building 4
Activity 1 1
Denver
Building 5
Activity 1 2
Denver
Building 6
Activity 1 3
Denver
Building 6
Activity 14
Denver
Building 6
Activity 1 5
Activity
Demolition of
East and West Walls
Abrasive Paint Removal
Drilling Wood
925 Holes with 1/4 " Bit
Drilling Wood
1 17 Holes with 1" Bit
Sawing Wood
75 Feet with Circular Saw
(Smooth Blade)
Door Modification
Door Modification
HVAC Sampling
HVAC Sampling
Drilling Plaster
925 Holes with 1/4 " Bit
Demolition of
East and West Walls
Day/Time
Day 3-2
10:20 - 11:59
Day 3-2
13:07 - 14:45
Day 3-3
09:14 - 09:37
Day 3-3
09:44 - 09:53
Day 3-3
10:52 - 11:56
Day 3-3
13:20 - 14:32
Day 3-3
13:11 - 14:30
Day 3-2
10:55 - 11:16
Day 3-4
10:55 - 11:16
Day 3-3
14:50 - 15:06
Day 3-4
8:58 - 9:46
Worker
3A 3B
3A 3B
3B
3B
3A
3D
3A
MRI
MRI
3B
3A 3B 3C
Room
Description
Bathroom
South Dining Room
Unitd)
Unit(2)
Unit(4)
Unitd)
Unit(2)
Basement
Basement
Center Bedroom
Dining Room
Samples
Collected
2 Personal Exposure
6 Settled Dust
2 Paint Chip/2 Plaster
2 Pre-Activity Dust
2 Personal Exposure
1 2 Settled Dust
4 Paint Chip
2 Pre-Activity Dust
1 Personal Exposure
5 Settled Dust
1 Paint Chip
1 Personal Exposure
5 Settled Dust
1 Paint Chip
1 Personal Exposure
5 Settled Dust
2 Paint Chip
1 Personal Exposure
5 Settled Dust
3 Paint Chip
1 Personal Exposure
5 Settled Dust
3 Paint Chip
5 Interior Duct Wipes
4 Interior Duct Wipes
1 Personal Exposure
5 Settled Dust
1 Paint Chip/1 Plaster
3 Personal Exposure
6 Settled Dust
2 Paint Chip/2 Plaster
Unlike the carpet removal and window replacement phases, ambient air samples were not
collected in the CED phase because (1) multiple simulated activities being conducted in a single
building did not allow for suitable adjacent non-activity areas, and (2) ambient air samples in
previous phases had low lead levels.
Personal exposure sampling was conducted on each participating R&R worker to measure
lead concentrations in air within the workers' breathing zones while they were performing the
activities. Personal exposure monitoring started immediately prior to the beginning of the activity,
and stopped as soon as the activity was completed.
8-51
-------
Dust samples from SSDCs were collected approximately one hour after the completion of
each activity. Times could not be exactly one hour after completion of the activity because the
number of samples per activity was too great to allow simultaneous collection.
At selected locations within the Baltimore units, dustfall samples were taken from SSDC
surfaces via wipe techniques, in support of an EFSS study objective to compare results between
vacuum and wipe collection techniques. These wipe samples were collected adjacent to vacuum
sample areas. Comparisons of wipe and vacuum sample results were found in Section 7.3 of
Chapter 7.
Before CED field work started, a work plan was developed that assigned locations to each
CED activity to be conducted at a given site. An approximate spot was chosen where the activity
was to be performed (designated the activity area). Locations for dustfall samples were also
determined.
Table 8C-1 presents the proposed types and numbers of environmental field samples
collected during the CED field study. The major difference in sampling plans for the different
CED activities was in the layout of the settled dust sampling locations. For most of the CED
activities (drilling, sawing, door modification, and HVAC removal) the layout was fairly
consistent. One sample site was located adjacent to the surface being disrupted, and 4 others
located in a symmetric pattern at approximately 4-6 feet away from the surface being disturbed.
The layout for settled dust samples planned for the paint removal activity was similar, but
included more sample locations. The sampling design for the demolition activities did not include
any settled dust sample located adjacent to the surface being demolished due to the amount of
debris distributed adjacent to the surface.
In addition to the environmental field samples associated with CED activities, several field
quality control (QC) samples were included in the sampling design. Field blank samples were
taken for three types of samples: personal exposure, dust wipe, and dust vacuum. Field blanks
were collected at the beginning of each day for the type of sample that was scheduled to be
collected. Personal exposure and settled dust field blank samples were to be collected on each
day of work; wipe blank samples were to be collected only on the days when the HVAC activity
was to be performed.
More details on the sampling design, including the protocols used to collect the samples,
are found in the Quality Assurance Project Plan.
8C-2.0 STUDY RESULTS
This chapter presents the results of the statistical analysis of environmental sample data
from the CED phase of the EFSS.
8-52
-------
8C-2.1 STRUCTURE OF CEP DATA
The purpose of the statistical analysis is to evaluate the lead disturbance and lead exposure
that results from performing some prescribed R&R activity on a lead-contaminated component.
The models that were selected for presentation allow for comparison of different R&R activities
on the basis of lead exposure for workers and lead disturbance and potential exposure to
occupants.
All of the samples and their associated measurements in the CED phase can be identified
by activity, dwelling unit, subunit (i.e., room or portion of a room) within dwelling unit, and
replicate. As a result, an experimental unit (EUijk) is defined as the occurrence of activity (i)
within subunit (k) of dwelling unit (j). Each environmental field sample collected during the CED
phase can be linked to one or more experimental units.
The following paragraphs describe the variables or responses included in the statistical
analysis.
Activity. As discussed in Section 8C-1.2.1, the following eight different R&R activities
were investigated during the CED phase:
1. Drilling into wood or plaster surfaces using a small or large drill bit
2. Sawing into wood or plaster surfaces using a coarse or smooth blade
3. Abrasive Sanding and Scraping of a painted wood surface
4. Door Modification on wood doors
5. Building Component Removal of wood baseboards, doorjambs, etc.
6. Demolition of plaster and wood walls
7. HVAC removal in houses with interior lead-based paint
8. Cleanup after R&R activities.
The first six activities were performed on components that were covered with lead-based paint.
The components were identified and confirmed as containing lead-based paint prior to the study
using a portable XRF analyzer.
Personal Exposure Monitor. During the eight activities, workers who performed the
activity were monitored for exposure to airborne lead using personal air pumps. Total
activity monitoring included the entire length of the activity and tool cleanup. For each
worker, the amount of lead collected within the sample cassette(s), the duration of the
monitoring, and the air pump flow rate were used to calculate the worker's task-length
average (TLA) lead exposure in |ig/m3. As discussed in Chapter 6, a natural log-
transformation was applied to each TLA for use as a response variable in the statistical
models.
= Personal exposure monitor result (|ig/m3) measured on the 1th worker during
the occurrence of activity (i) within subunit (k) of dwelling unit (j).
8-53
-------
Settled Dust. Prior to each CED activity, a number of SSDCs were placed on the floor at
predetermined distances from the center of the activity. These plates were used to collect
the fallout of dust and debris generated by each occurrence of an R&R activity.
Approximately one hour after the completion of an activity, the dust located within each
plate was collected and then analyzed for lead. Since each SSDC had a surface area of
one square foot, the results of the settled dust field samples are reported as a dust-lead
loading in units of |ig/ft2. As discussed in Chapter 6, a natural log-transformation was
applied to each settled dust loading for use as a response variable in the statistical models.
In addition to the dust-lead loading, each SSDC has an associated distance, which
measures how far away each dust collector was placed from the surface being disturbed
during the activity.
Dustijkl = Dust lead loading result (|ig/ft2) measured on the 1th SSDC from the
occurrence of activity (i) within subunit (k) of dwelling unit (j).
Distance^ = Distance from the activity of the 1th SSDC from the occurrence of activity
(i) within subunit (k) of dwelling unit (j).
During the demolition, abrasive sanding, and component removal activities, some SSDC's
were placed in adjacent pairs in an effort to assess side-by-side variability within settled
dust results. For the demolition activity in particular, some of the side-by-side paired
samples were sampled using both dust vacuum and dust wipe collection methodologies.
With the exception of the demolition side-by-side vacuum/wipe samples, all other settled
dust samples from the CED phase were collected using the vacuum methodology.
Pre-Activity Lead Levels. To understand factors that may affect the potential lead
exposure that results from performing an R&R activity on a lead-contaminated
component, it is important to also characterize the amount of lead that was on the surface
of each disturbed surface prior to the activity. Thus, for each R&R activity, at least one
field sample was collected to characterize the pre-activity lead level. For activities which
disturbed lead-painted surfaces, the pre-activity lead level was measured using paint chip
samples and reported in units of mg/cm2. For the HVAC removal activity, a number of
dust wipe samples were taken from the interior of the duct-work to characterize the pre-
activity lead levels. The pre-activity dust wipe samples for the HVAC removal activity are
reported in units of |ig/ft2.
PaintjjH = Lead in paint result (mg/cm2) measured on the 1th surface during the
occurrence of activity (i) within subunit (k) of dwelling unit (j).
HVAC Wipejkl = Lead in wipe sample (|ig/ft2) obtained from the interior of the 1th duct
surface during the occurrence of the HVAC removal activity within
subunit (k) of dwelling unit (j).
For R&R activities which disturbed lead-painted plaster, additional pre-activity samples of
the plaster were collected and archived for possible future analysis.
8-54
-------
Table 8C-2 summarizes the number of field samples collected and available for statistical
analysis for the CED phase:
Table 8C-2. Number of Field Samples Collected and Available for Analysis
in the CED Phase
Activity
Type
Drilling
Sawing
Sanding
Door
Modification
Removal
HVAC
Demolition
Cleanup
Substrate
Plaster
Wood
Plaster
Wood
Wood
Wood
Wood
Duct
Plaster
Plaster
Wood
Exposure
Units
6
7
2
6
3
6
1
2
9
4
2
Personal
Exposure
Monitor
6
7
2
6
9
6
2
4
20
4
2
Settled
Dust
Vacuum
24
35
10
30
28
30
5
15
51
0
0
Settled
Dust Wipe
0
0
0
0
0
0
0
0
12
0
0
Pre
Activity
Paint
5
7
2
12
10
17
0
0
16
0
0
Pre
Activity
Dust
0
0
0
0
2
0
0
7
2
0
0
8C-2.2 CHARACTERIZING LEAD DISTURBANCE AND POTENTIAL LEAD EXPOSURES
The lead disturbance and exposure data observed in the CED phase are characterized
below, according to the sample or exposure type considered.
For each worker, the amount of lead collected within the sample cassette(s), the duration
of the monitoring, and the air pump flow rate were used to calculate the worker's
task-length average (TLA) lead exposure in |ig/m3. A discussion of the relationship between a
TLA and an 8-hour time-weighted average (TWA), on which the OSHA PEL is based, is given in
Section 9.1.2 of Chapter 9.
8C-2.2.1 Personal Worker Exposures
In the CED phase, each worker was monitored for airborne lead exposure for the duration
of each R&R task. Duration was defined as the time it took a worker to complete the task and
clean his tools. The duration of CED phase activities ranged from 9 to 185 minutes. Table 8C-3
gives a summary of task duration for each CED activity under investigation.
Figure 8C-1 is a plot of the task-length average personal exposure lead concentrations
(|ig/m3), or average exposures over the duration of activity, for each CED activity type. In this
plot, a horizontal line is drawn at 50 |ig/m3, which represents the OSHA PEL for lead exposure
8-55
-------
during an assumed eight hours of exposure per day. This plot demonstrates that all CED
activities, with the exception of the drilling activities, HVAC removal, and cleanup of plaster,
resulted in most worker TLAs above the OSHA PEL. All worker TLAs were at or above the
OSHAPEL for cleanup of wood, door modification, wood component removal, abrasive sanding,
and sawing activities. Nineteen of twenty TLAs were above the OSHA PEL for demolition. In
contrast, for drilling activities, all worker TLAs except one were at or below the OSHA PEL.
Further summary of the data in Figure 8C-1 can be found in Table CED-1 of Appendix A.
Random-effects model (CED-1) in Appendix C was fit separately to the TLA personal
exposure concentrations for each combination of target activity and substrate, in order to
characterize four components of variability in the log-transformed TLA concentrations. These
variance components are:
• unit to unit variability, or variability resulting from sampling from different study units
• variability resulting from performing the activity at different subunits within a study
unit
• variability among different workers
• replication variability, including measurement error.
Table 8C-3. Task Length for Each Combination of Target Activity and Substrate in the
CED Phase
Activity
Type
Drilling
Sawing
Sanding (Hand)
Sanding (Power)
Door
Removal
HVAC
Demolition
Cleanup
Substrate
Plaster
Wood
Plaster
Wood
Wood
Wood
Wood
Duct
Plaster
Plaster
Wood
n
6
7
2
6
6
3
6
2
4
20
4
2
Average
Duration
(minutes)
40
36
19
41
63
39
68
160
48
61
49
15
Minimum
Duration
(minutes)
15
9
16
14
58
22
43
135
21
25
14
12
Maximum
Duration
(minutes)
85
83
22
77
75
64
91
185
74
99
129111
18
The second-highest duration was 27 minutes.
8-56
-------
1000
100
10
o
o
o
o
o
o
$
8
«
o
s
o
_
Clean Clean Demo Door Drill Drill HVAC Comp Sanding Sow Sow
Piaster wood Mod Piaster wood Removal Removal Piaster wood
Task Type
oo Baltimore Unit 1 ° ° ° Baltimore Unit 2 * * * Denver Units
The dotted line at 50
represents the OSHA PEL for R&R workers.
Figure 8C-1. Scatterplot of Personal Exposure Loadings by Task Type and Unit ID
Table 8C-4 displays model estimates of the geometric means and variance components
(standard deviations) of worker personal exposure to airborne lead for each CED activity,
resulting from fitting model (CED-1). The actual number of variance components that can be
estimated from the CED data is different for each target activity and is based on the number of
observations per target activity and the CED experimental design. For example, if only two
observations of worker exposure to airborne lead are available, we can only estimate the
geometric mean and residual standard deviation for the given activity.
Table 8C-5 displays estimates of the 50th, 75th, and 95th percentiles of the distribution of
personal exposure concentrations for each target activity, with approximate 95% confidence
intervals. The approaches for calculating the percentiles and confidence intervals are found in
Appendix C. Figure 8C-2 presents a graphical display of the estimated 75th percentile (and
associated confidence interval). The vertical line in this figure represents the OSHA PEL of 50
|ig/m3. Results such as these percentiles, with their estimates of variability, can be used to
compare potential lead exposures for R&R workers across activities.
8-57
-------
Table 8C-4. Model Estimates of the Geometric Mean and Standard Deviation of Variance
Components of Worker Personal Exposure to Airborne Lead (Task Length
Average) for Each CED Activity Based on a Variance Components Model of
log(PEMijkl)
Activity
Drilling
Sawing
Sanding (Hand
Sanding (Power)
Door Modification
Component
Removal
HVAC
Demolition
Cleanup
Substrate
Plaster
Wood
Plaster
Wood
Wood
Wood
Wood
Duct
Plaster
Plaster
Wood
Number
of
Samples
6
7
2
6
6
3
6
2
4
20
4
2
Geometric Mean
Estimated
from Model
6.76
15.1
110
546
254
571
590
344
49.6
108
24.5
102
Residual
(oError)nl
0.313
1.29
1.11
0.380
0.589
0.847
0.0451
0.101
0.475
0.550
0.205
Worker-to-
Worker
(Ow.rk.rl
0.434
0.000
0.000
0.000
0.896
Unit-to-Unit
(Ouni.)
0.860
0.147
0.463
Task-to-Task
Within Unit
(OjasklUnit))
0.451
' Standard deviation of the log transformed data (expressed in log measurement units).
Table 8C-5. Estimates of the 50th, 75th, and 95th Percentiles for Worker Personal
Exposures, Along With 95% Confidence Intervals, by Activity in the CED
Phase
Activity
Drilling
Sawing
Sanding (Hand)
Sanding (Power)
Door
Modification
Component
Removal
HVAC
Demolition
Cleanup
Substrate
Plaster
Wood
Plaster
Wood
Wood
Wood
Wood
Duct
Plaster
Plaster
Wood
Number
of
Samples
6
7
2
6
6
3
6
2
4
20
4
2
50th Percentile
(jug/m3)
Estimate
6.76
15.1
110
546
254
571
590
344
49.6
108
24.5
102
95% C.I.
(3.00, 15.3)
(4.57, 50.2)
(0.01, 2. 32x1 O6)
(366, 813)
(23.7, 2720)
(42.9, 7600)
(93.5, 3730)
(229, 516)
(11.4, 216)
(26.6, 435)
(10.2, 58.7)
(16.2, 646)
75th Percentile
(jug/m3)
Estimate
9.70
36.3
232
705
513
1150
1360
354
56.0
185
35.5
118
95% C.I.
(5.71, 40.3)
(13.8, 214)
(11.9, 2. 40x1 O13)
(518, 1300)
(150, 63800)
(286, 170000)
(410, 35200)
(314, 995)
(37.7, 3070)
(95.7, 5500)
(19.9, 159)
(67.9, 12800)
95th Percentile
(jug/m3)
Estimate
16.3
127
681
1020
1410
3170
4480
370
66.6
403
60.5
143
95% C.I.
(9.47, 226)
(42.9, 2490)
(149, 2.15x10")
(726, 2890)
(447, 2. 03x1 O7)
(940, 5. 03x1 O7)
(1300, 1.89x106)
(348, 3680)
(53.0, 588000)
(186, 542000)
(34.0, 925)
(108, 4. 89x1 O6)
8-58
-------
Drill/Plaster
Clean/Plaster
Drill/Wood
HVAC Removal
ClearvWood
Demolition
Saw/Plaster
Component Removal
Hand Sanding
Saw/Wood
Power Sanding
Door Modification
1 10 100 1000 10000 100000 1000000
75* Percentile of Personal Exposure To Airborne Lead Cu.g/m3)
The clotted line at 50
represents the OSHA PEL for R&R workers.
NOTE: The upper bound for 'Saw/Plaster1 is 2 x 1013
Figure 8C-2. 75th Percentile and Associated 95% Confidence Interval for Personal
Exposure to Airborne Lead (//g/m3) from Each Combination of CED Activity
and Substrate
The results presented in Table 8C-4 demonstrate that for R&R activities that are
performed on both wood and plaster surfaces in the CED phase, the estimated median worker
exposure (as given by the geometric mean) is much higher for wood surfaces. This trend is
consistent for the drilling, sawing, and cleaning activities, and is possibly attributable to a higher
concentration of lead in paint on wood surfaces. Table 8C-6 provides parameter estimates (in the
log domain) for the effect of a wood substrate on personal exposure levels when performing
the same activity on wood versus plaster surfaces. These results are based on fitting Model
(CED-2) in Appendix C. (Note that the estimates are similar to, but not in complete
correspondence with those obtained from the geometric means in Table 8C-4. This difference is
caused by fitting different models to the data.)
8-59
-------
Table 8C-6. Estimates of the Increase in Personal Air Exposures that are Attributable to a
Substrate Effect (Wood versus Plaster) in the CED Phase
Activity
Drilling
Sawing
Cleanup
Parameter
Estimates of
Increase in
log(PEMijkl)
Associated with
Wood Over Plaster
0.798
1.60
1.70
Standard
Error of
Estimate
0.529
0.465
0.248
Degrees
of
Freedom
8
4
2
P-Value
for Test of
Significance
from Zero
0.17
0.03
0.02
Multiplicative Increase
in PEM Results When
the Activity is
Performed on Wood
Rather Than Plaster
2.2
5
5.5
The interpretation of the estimated increase in log(PEMijkl) with wood versus plaster
substrates is multiplicative. The final column in Table 8C-6 shows that drilling into wood results
in a 2.2 fold increase in the personal exposure level over drilling into plaster, while a 5 fold
increase in wood over plaster is seen with sawing, and a 5.5 fold increase in wood over plaster is
seen during cleanup. The increases associated with the latter two activities are significant at the
0.05 level.
8C-2.2.2 Lead Disturbance and Potential Occupant Lead Exposure
SSDC Approach to Settled Dust Sampling
Immediately prior to the occurrence of each CED activity, a number of SSDCs were
placed at various distances away from the surface about to be disrupted. These SSDC's measured
one square foot in area, and were designed to collect the fallout of dust and debris generated
during an R&R activity.
Approximately one hour following the conclusion of each CED activity, the settled dust
and debris were collected from each SSDC through vacuum sampling, and then chemically
analyzed for lead.* As discussed in Section 8C-2.1, each settled dust sample has an associated
measure of the lead-loading in units of micrograms of lead per square foot (jig/ft2) and a measure
of distance from the activity. Distance for each settled dust sample was measured as how far
away (in feet) its SSDC was from the nearest surface that was disturbed during the activity.
The goal of the statistical analysis of settled dustfall lead-loadings is to characterize the
potential lead exposure to occupants from dust and debris generated by each CED activity. The
settled dustfall loadings from each CED activity represent the amount of lead generated by the
performance of the activity at varying distances away from the surface being disturbed.
* A small number of settled dust samples were collected from the SSDC's using wipe sampling. These wipe sampling
results were not utilized in the determination of potential occupant exposures. They were used only for the vacuum/wipe
methodology comparisons presented in Chapter 7.
8-60
-------
For most of the CED activities (drilling, sawing, door modification, and HVAC removal)
the placement of SSDC's was fairly consistent, with one SSDC adjacent to the surface being
disrupted, and four other SSDC's placed in a symmetric pattern at approximately 4-6 feet away
from the nearest surface. The adjacent SSDC was expected to capture the fallout of both large
debris and dust generated by each activity, while the other four plates were expected to capture
mostly dust and smaller airborne particles. The sampling design for the abrasive sanding activity
was similar, but included more SSDCs. The sampling design for demolition activities did not
include any SSDC's located adjacent to the surfaces being demolished, for two reasons: (1) they
would have been an obstacle to the workers, and (2) the expected amount of large debris in a
sample collected by an adjacent SSDC would have been difficult to chemically analyze.
Relating Lead Loadings in Settled Dust to Distance from Activity
Exploratory analysis of the settled dustfall results demonstrated an approximate linear
relationship between log-transformed lead loadings and distance from the activity. Model (CED-
3) in Appendix C was fitted separately to log-lead loading data for each experimental unit (i.e.,
activity/unit/subunit combination) to estimate this linear relationship to distance. This model took
the form
log(dust) = P0 + PJ Distance + error.
Averages of the parameter estimates for each activity/substrate combination are given in Table
C-8 of Appendix C.
For most experimental units, the slope estimate within Model (CED-3) was negative,
which implied two things:
1. The highest settled dust lead-loading results are associated with SSDC's that are
positioned adjacent to the surface being disrupted.
2. The amount of lead found on each SSDC diminishes exponentially with respect to
distance, at a rate determined by the slope parameter.
After Model (CED-3) estimated the relationship between lead loading and distance from
activity for each experimental unit, it was then necessary to summarize these results across all
experimental units. Two different models, a two-stage model and a population model, were
examined to summarize the relationship across experimental units. The two models produced
similar results. (Details concerning the choice of model are provided in of Appendix C).
Therefore we chose to focus our conclusions on the population model, given by Model (CED-5)
in Appendix C. This model was fit separately for each combination of activity and substrate, to
model lead-loading as a function of distance.
8-61
-------
Estimating Lead Disturbance Within a 6' x T Gradient
As stated earlier, the goal of the statistical analysis of settled dust lead-loadings is to
determine the potential occupant exposure to lead that results from each CED activity. The result
of fitting the population model (CED-5) to lead loading data allows us to estimate an overall
average curve which explains the average amount of lead in settled dust that is expected to occur
at varying distances for each activity, and the components of variation about that curve. For a
given activity (i), the population average response curve has the following form:
Lead in dust (|ig/ft2) = exp(poi)- exp(pi;- Distance)
By integrating the area underneath the estimated curve for Activity (i) from 0 to 1 foot,
we obtain the expected lead-loading of an SSDC located adjacent to the surface of activity (i)
(denoted by "[0-1] foot SSDC"). Similarly, by integrating the area underneath the estimated
curve for activity (i) from 5 to 6 feet, we obtain the expected lead-loading of an SSDC located 5-
6 feet from the surface of activity (i) (denoted by "[5-6] foot SSDC"). The measure used to
characterize lead disturbance and potential exposure to occupants for each CED activity is the
amount of lead in a 6-foot by 1-foot gradient region, which is obtained by integrating the area
underneath the estimated curve from 0 to 6 feet.
Table 8C-7 displays the parameter estimates (and associated standard errors) for the
population average lead loading curve, and the calculated average lead loadings ([0-1] foot
SSDC, [5-6] foot SSDC, and 6-foot by 1-foot gradient; and associated standard errors) for each
CED activity. The average lead loadings are relative to the total amount of activity performed in
the CED phase. The standard errors of the average lead loadings for each activity were calculated
by using the Delta Method (Bishop, Fienberg, and Holland, 1975), which is based on asymptotic
normal theory. However, the sample sizes (number of experimental units) within each CED
activity are small, making these estimates only approximate. Details of the statistical
methodology are presented in Section C.5 of Appendix C.
The estimated 6-foot by 1-foot gradient lead loadings for each CED activity are displayed
in Figure 8C-3.
The estimated [0-1] foot SSDC and [5-6] foot SSDC loadings have different interpretations
with respect to each CED activity. The [0-1] foot SSDC loading is representative of the amount
of lead that is likely to fall directly underneath the surface being disrupted, while the [5-6] foot
SSDC is more representative of the amount of lead that becomes airborne in dust and small
particles. By comparing the estimates of the [0-1] foot SSDC to estimates for the entire gradient,
we can gain some perspective on the amount of lead disturbed that is likely to stay localized,
versus the amount of lead that is likely to become airborne for each activity. For example,
approximately 80% of the lead in the gradient remains localized in the [0-1] foot area for the two
drilling activities, while only 33% stayed localized in the abrasive sanding activity. Thus, a higher
percentage of the lead disturbed becomes airborne when performing abrasive sanding in
comparison to drilling. This type of information may be useful in the design of final cleanup
procedures for R&R workers to help protect occupants from lead exposure.
8-62
-------
Table 8C-7. Estimates of Lead Disturbance for Each CED Activity Based on the
"Population" Random Effects Modeling of Settled Dust Lead-Loading
Results'11
Target
Activity (i)
Drilling
Drilling
Sawing
Sawing
Abrasive
Sanding
Door
Modification
HVAC
Removal
Demolition
Substrate
Plaster
Wood
Plaster
Wood
Wood
Wood
Duct
Plaster
Number
of
EUiik's
4
7
2
6
3
6
2
9
Poi
(selPoi))
10.5
(1.10)
12.9
(0.602)
11.2
(1.20)
12.6
(0.474)
11.5
(0.737)
11.1
(0.707)
8.08
(2.21)
8.77
(0.839)
P,,
(selP,;))
-1.67
(0.344)
-1.47
(0.121)
-0.929
(0.247)
-0.668
(0.117)
-0.341
(0.047)
-0.417
(0.094)
-0.375
(0.226)
-0.263
(0.071)
[0-1] Foot SSDC
Mean (//g/ft2)
(Standard Error)
17400
(17500)
205000
(116000)
50000
(55700)
223000
(97500)
85500
(62800)
54000
(36300)
2690
(5690)
5690
(4630)
[5-6] Foot
SSDC
Mean (//g/ft2)
(Standard Error)
4.08
(5.40)
130
(51.9)
480
(330)
7900
(3450)
15500
(11500)
6700
(2630)
414
(516)
1530
(948)
6' x 1' Gradient
Mean
(//g/6 ft2)
(Standard Error)
21400
(20600)
266000
(143000)
82300
(81400)
449000
(167000)
257000
(188000)
145000
(82700)
7710
(14100)
19500
(14300)
1 Lead in Dust (fjg/ft2) = exp (Poi), exp (P,i Distance).
Dust Pb (ug)
400000
Figure 8C-3. Estimated Lead Fallout Gradient in Settled Dust from 0 to 6 Feet Away from
the Edge of the Activity Space
8-63
-------
The above results are based on the total amount of activity performed in the CED phase.
To facilitate comparability in results across phases, this analysis was repeated after adjusting the
amount of each activity to a "real-world" standard. The results of the standardized analysis are
found in Section 8C-2.2.3.
Exposure Potential for Wood Versus
Plaster Components in Drilling and Sawing
The estimated results of the lead disturbed in the 6-foot by 1-foot gradient region suggest
that CED activities which disturb lead-painted wood surfaces have higher loadings than activities
which disturb lead-painted plaster surfaces. This trend for the gradient lead loading is consistent
with the results from the personal exposure monitoring results. Table 8C-8 presents the results of
fitting statistical Model (CED-6) in Appendix C to estimate differences in slopes and intercepts
that result in considering wood substrate over plaster substrate. Results are presented for drilling
and sawing activities.
Although the differences in parameter estimates between wood and plaster surfaces are
not statistically significant at the 0.05 level, they suggest the following:
• the positive increase in the intercept term suggests that the amount of lead found
directly underneath a wood surface is greater than that for a plaster surface.
• the slight positive increase in the slope term, coupled with the negative slope estimates
for both wood and plaster, suggests that the exponential decline in lead amounts with
respect to distance is slower with wood than with plaster surfaces.
Table 8C-8. Estimates of the Log Increase in Intercept and Slope for the Relationship
Between log (Dustukl) and Distance^, that are Attributable to a Substrate
Effect (Wood versus Plaster)
Activity
Drilling
Sawing
Parameter111
Intercept
Slope
Intercept
Slope
Parameter Estimates of
the Increase in Parameter
Estimates for Wood
Versus Plaster
2.54
0.133
1.35
0.271
Standard
Error
1.18
0.321
1.34
0.254
Degrees
of
Freedom
9
9
6
6
P-Value
0.06
0.69
0.35
0.33
m "Intercept" represents the baseline increase in log(Dust) associated with disturbing wood
substrate versus plaster substrate. "Slope" represents the amount this baseline increase is
inflated by moving 1-foot closer to the activity.
8-64
-------
Another method of interpreting the substrate effect within these two CED activities is to
examine the differences between the 6-foot by 1-foot gradient lead loading estimates for wood
versus plaster. Table 8C-9 presents estimates of the log-transformed gradient region loadings
(and associated standard errors) for each substrate, along with the corresponding two-sample
t-test statistic. The lead gradient loadings are transformed to the natural log scale in an effort to
improve the assumption of normality that is required by the two-sample t-test.
Table 8C-9. Estimated Differences in the 6' x 1' Lead Loading Gradient Attributable to a
Substrate Effect (Wood verses Plaster)
CED
Activity
Drilling
Sawing
Wood
log(6' x 1' Gradient)
(Std. Error.)
12.5
(0.54)
13.0
(0.37)
Plaster
log(6' x 1' Gradient)
(Std. Error.)
9.97
(0.96)
11.3
(0.99)
t-Statistic
2.29
1.61
Degrees of
Freedom111
5.0
1.3
P-value
0.07
0.31
111 Determined by Satterthwaite's approximation.
While the observed difference between wood and plaster results were not statistically
significant at the 0.05 level, the power of the test is limited by the small sample sizes.
The estimated positive differences in measures of potential occupant exposure (intercept,
slope, and lead loading gradient) between wood and plaster surfaces for these two activities might
be attributable to the fact that the concentration of lead in paint on wood surfaces is generally
higher than the concentration of lead in paint on plaster surfaces. This issue will be further
investigated in Section 8C-2.3.2.
8C-2.2.3 Adjusting Estimates of Lead Disturbance for the Amount of R&R Activity
The generic activities of sawing and drilling, as designed in the CED phase, were not
intended to reflect sawing or drilling activities to the extent that they are performed in the R&R
industry. Rather, they were intended to provide sufficient activity to allow for estimating a task
length average worker exposure, as well as estimating the lead disturbed by the specified amount
of activity. However, the measure of the amount of lead disturbed by a CED activity is directly
related to the amount of activity performed. As a result, it is desirable to express settled dustfall
relative to a "standard unit of activity".
As they exist in this study, the simulated target activities of demolition and HVAC duct
removal represent a "real-world" standard unit of activity. For demolition, the standard unit of
activity is the demolition of interior wall(s) in a single room of a building. For HVAC work, the
standard is the removal of all duct work in a single room of a building. As a result, the unit of
activity will not be adjusted for these two task activities. In contrast, because the generic
activities of drilling and sawing, as conducted in this study, were not judged to be a reasonable
8-65
-------
representation of any "real-world" R&R activity, the amount of lead disturbed will be adjusted to
a "standard unit" of activity for these activities. For drilling, this standard unit represents 10 small
holes or 1 large hole. For sawing, this standard unit represents 1 linear foot cut. Subsequent
estimates of the average amount of lead disturbed by that activity can then be interpreted as the
amount of lead disturbed per unit activity.
Two activities could not be considered when defining standard units of activity. The
positive control activity, abrasive sanding, was conducted primarily to obtain personal worker
exposure measurements for comparison to other activities and cannot be accurately quantified
according to the "amount of activity." Likewise, the door modification task was originally
designed as a repeated task for a specified amount of time out of concern over obtaining
detectable levels of airborne lead exposure. Therefore, neither of these two tasks can be
adequately defined per unit of activity and are not included in the adjusted results.
Table 8C-10 presents the amount of lead in the [0-1] foot SSDC, [5-6] foot SSDC, and
6-foot by 1-foot gradient region for the "standard unit of activity" for each CED activity in which
a standard unit of activity could be determined. Figure 8C-4 demonstrates the amount of lead in
the gradient region per standard unit of activity. Note that by comparing Table 8C-7 with Table
8C-10, and Figure 8C-3 with Figure 8C-4, one can observe how adjusting for the "standard unit
of activity" qualitatively affects the comparison of lead disturbance across different activities.
Lead disturbance resulting from demolition and HVAC removal is more substantial relative to
other activities when these other activities are adjusted for the amount of activity performed.
Table 8C-10.
Estimates of the Potential Occupant Lead Exposure for Each CED Activity
Relative to a Standard Unit of Activity'11
CED
Activity (i)
Drilling
Drilling
Sawing
Sawing
Demolition
HVAC
Removal
Substrate
Plaster
Wood
Plaster
Wood
Plaster
Duct
System
Standard Unit
of Activity
1 0 1/4 " Holes
or a 1 " Hole
1 0 1/4 " Holes
or a 1 " Hole
1 Linear Foot
1 Linear Foot
1 Room
1 Room of
Duct Work
Estimated
Intercept
Bo;
5.85
(1.05)
8.25
(0.605)
7.53
(1.59)
8.31
(0.474)
8.78
(0.839)
8.08
(2.21)
Estimated
Slope
Pi,
-1.68
(0.343)
-1.47
(0.120)
-0.947
(0.305)
-0.668
(0.117)
-0.263
(0.071)
-0.375
(0.226)
[0-1] Ft SSDC
(//g/ft2) Mean
(Std Error)
168
(163)
2000
(1140)
1200
(1780)
2970
(1300)
5690
(4640)
2690
(5700)
[5-6] Ft SSDC
(//g/ft2) Mean
(Std Error)
0.04
(0.05)
1.27
(0.49)
10.6
(6.63)
105
(46.0)
1530
(948)
414
(516)
6' x V Gradient
(jug/6 ft2) Mean
(Std Error)
207
(191)
2590
(1400)
1970
(2560)
5990
(2230)
19500
(14300)
7710
(14100)
Lead in Dust = exp(poi) exp(P,i Distance)
8-66
-------
0
Demolition
HVAC Removal
Saw/Wood
Drill/Plaster
Figure 8C-4. Estimated Lead Fallout Gradient in Settled Dust, From 0 to 6 feet Away From
the Edge of the Activity Space - Adjusted to the "Standard Unit of Activity"
8C-2.3 FACTORS OR MEASUREMENTS RELATED TO LEAD DISTURBANCE
8C-2.3.1 Pre-activity Measures of Lead Contamination
This section provides descriptive characterizations on the three types of pre-activity sample
media considered in the CED phase. These characterizations include distributional qualities,
descriptive summaries and plots for paint chip samples, dust wipe samples from HVAC systems,
and pre-activity dust vacuum samples. Tables of descriptive summaries in this section present
number of observations, arithmetic and geometric means, log standard deviation and minimum
and maximum values.
8C-2.3.1.1 Results of Paint Chip Samples
A total of 62 paint chip samples were collected during the CED phase prior to the
occurrences of those activities which disturbed lead-painted surfaces. The lead loading (mg/cm2)
for each paint chip sample was calculated by dividing the measurement of lead in the sample (mg
lead) by the area of the sample in square centimeters. Table 8C-11 provides descriptive statistics
on the distribution of paint chip lead loadings from samples collected from within each CED
activity. The number of samples listed in Table 8C-11 total 69 (instead of 62) because samples
were collected from building components disturbed by more than one CED activity. Table 8C-12
provides descriptive statistics for the distribution of paint chip lead loadings based on substrate
(wood vs. plaster).
8-67
-------
Table 8C-11. Descriptive Summaries of Paint Chip Lead Loadings (mg/cm2) from Samples
Collected Within Each Activity in the CED Phase
Task Type
Demolition
Door/Wood
Drill/Plaster
Drill/Wood
Sand
Saw/Plaster
Saw/Wood
Number
16
17
5
7
10
2
12
Arithmetic
Mean
5.24
8.78
3.18
7.38
10.4
2.11
7.26
Geometric
Mean
0.840
4.90
0.639
5.51
5.03
0.0401
4.06
Log Std.
Dev.
3.22
1.50
2.54
1.08
1.61
6.58
1.71
Minimum
Value
0.0002
0.0587
0.0130
0.519
0.285
0.0004
0.0266
Maximum
Value
17.7
28.7
12.7
13.4
39.9
4.21
19.5
Table 8C-12. Descriptive Summaries of Paint Chip Lead Loadings (mg/cm2) by Substrate
in the CED Phase
Substrate
Wood
Plaster
Number
44
18
Arithmetic
Mean
8.65
4.78
Geometric
Mean
4.87
0.397
Log Std.
Dev.(1)
1.48
3.63
Minimum
Value
0.0266
0.0002
Maximum
Value
39.9
17.7
Exploratory plots are included in Figures 8C-5 and 8C-6. Figure 8C-5 is a plot of lead
loadings in paint chip samples, identified by the activity which disrupts the surface. This plot
demonstrates the variability for paint chip samples within each CED activity. Figure 8C-6 is a
scatterplot of paint chip lead-loadings separated by substrate (wood vs. plaster), with different
symbols representing different dwelling units. Although the geometric means for paint chip
loadings from wood and plaster surfaces were an order of magnitude apart, this plot indicates that
the difference in paint chip lead loadings between wood and plaster surfaces was primarily the
result of the difference in only one unit (Baltimore Unit 1) .
The differences in geometric mean loadings for paint chip samples collected from wood
versus those collected from plaster (Table 8C-12) may help to explain the higher personal
exposure and settled dust lead loadings observed for activities that disturb wood surfaces
compared to plaster-disturbing activities. This issue is examined further in Section 8C-2.3.2.
8-68
-------
100.0000
10.0000
1.0000
0.1000
0.0100
0.0010
0.0001
: : : it '-
6 :- :- :- :-
$ \ * \ t, \ * \ * \ \ °
O : 8 : : O : & : : ft
* : o : : : : ° : g
o ; ; ; ; ; ; o
o \ o ; o \ \ \ \
0 ; o ; : o : ; ;
; ; ° ; ; o ; :
o
* ! » ! : : : :
: : : : : : O
o ; ; 0 ; ; ; ;
; ; ; ; ; o ;
0 ; ; ; ; ; ;
Demo Door/W Drill/P Drill/W Sand Saw/P Saw/W
Task Type
000 Baltimore Unit 1 »<> * Baltimore Unit 2 * * * Denver Units
Figure 8C-5. Scatterplot of Paint Chip Loadings by Task/Unit in the CED Phase
S
100.0000
10.0000
1.0000
0.1000
0.0100
0.0010
0.0001
• 1 1
ill : '
• S
0 »
0 6
0
*
:
Wood Plaster
Substrate
• • • Baltimore Unit 1 * * * Baltimore Unit 2 * * * Denver Units
Figure 8C-6. Scatterplot of Paint Chip Loadings by Substrate in the CED Phase
8-69
-------
8C-2.3.1.2 Results of Dust Wipe Samples from HVAC Systems
A total of 21 wipe samples were collected from HVAC systems in the CED phase. Seven
wipe samples were taken from the HVAC systems in the two Baltimore housing units prior to
removing portions of these systems. The remaining 14 HVAC wipe samples were collected from
the HVAC systems from three housing units in Denver, where no HVAC removal was performed.
Table 8C-13 presents the descriptive summaries on lead loadings for HVAC samples across units
and within each unit. Geometric means suggest lead loadings in the Denver units (units 3 through
5 in Table 8C-13) were lower than the corresponding lead loadings in the Baltimore units.
Table 8C-13.
Descriptive Statistics for HVAC Wipe Sample Lead Loadings Across and
Within Units in the CED Phase
Unit ID
All
1
2
3
4
5
N
21
4111
3111
5
5
4
Arithmetic Mean
(//g/ft2)
5870
17900
6940
2060
1060
3850
Geometric Mean
(//g/ft2)
2900
14800
6880
2040
709
2700
Log
Std. Dev.
1.28
0.784
0.168
0.184
1.04
1.01
Minimum Value
(//g/ft2)
205
4880
5800
1610
205
1000
Maximum Value
(//g/ft2)
30900
30900
8120
2650
2140
8280
111 Samples taken before HVAC removal activities (Baltimore dwelling units).
The loading data summarized in Table 8C-13 are plotted in Figure 8C-7. This plot
illustrates the variability within and between housing units. It should be noted that the unit with
the lowest lead loadings contained an HVAC system attached to a furnace that was installed in
1985. These results provide an estimate of the amount of lead dust present in an HVAC system
that could be made available for human exposure if the system is disturbed.
8C-2.3.1.3 Results of Dust Vacuum Samples (Pre-Activity in Denver)
Only four pre-activity settled dust samples were collected from window sills and shelves
during the CED phase. These samples were taken from the same dwelling unit. Two samples
were taken before a demolition activity, and two samples were taken before an abrasive sanding
activity. Table 8C-14 presents the loadings and concentrations for these four samples.
8-70
-------
100000
i
f
10000
B. 1000
»
>
I
100
o ; ; ; ;
8 ! ! ! !
: o : : : o
0
; o ; ; ;
0 ; ; ; ; o
; ; o ; ;
; ; s ; s ;
: : o : :
; ; ; ; o
; ; ; ; o
: : : O :
: : : o :
; ; ; 0 ;
Bait. Unit 1 Bait Unit 2 Denv. Unit 1 Denv. Unit 2 Denv. Unit 3
Unit IDs (for HVAC Removal)
Figure 8C-7. Lead Loadings Within Wipe Dust Samples Collected from Inside HVAC
Ductwork
Table 8C-14. Pre-activity Settled Dust Results (Lead Loading and Sample Concentration)
from Window Sills and Shelves in the CED Phase
Activity
Sanding
Demolition
Surface
Window Sill
Window Sill
Window Sill
Lower Shelf
Loading
(//g/ft2)
9730
825
29100
129
Concentration
(//g/g)
31600
3240
19000
431
8C-2.3.2 The Relationship Between Airborne Lead, Lead in Settled Dust, and Paint Lead
Levels
In the previous sections, two hypotheses were made based on the outcome of statistical
analysis that investigated the relationship between airborne lead, lead in settled dust, and paint
lead levels:
8-71
-------
2.
Within a given CED activity, the observed substrate (wood vs. plaster) differences in
personal exposure and settled dust results can be attributed to differences in substrate
paint loadings.
The estimated lead loading in the [5-6] foot SSDC is representative of airborne dust
and smaller particles, while the estimated lead loading in the [0-1] foot SSDC is more
representative of larger debris that fell directly from the disturbed surface.
In this section, the validity of these two interpretations is tested by considering how paint lead
loadings affect observed substrate differences, how [5-6] foot SSDC results are correlated with
personal exposure results, and how [0,1] foot SSDC results are correlated with paint lead
loadings.
The Effect of Paint Loadings
If the first hypothesis is valid, then adding a fixed effect for paint lead loadings to the
statistical models that characterize personal exposure and settled dust lead loadings should
substantially reduce any remaining substrate effect on these data. Tables 8C-15 (personal
exposure concentrations) and 8C-16 (settled dust lead loadings) summarize the estimated
substrate effects before and after adjusting for paint lead loadings, and are obtained by fitting
Models (CED-7) and (CED-8) in Appendix C, respectively.
Table8C-15.
Estimated Substrate Effect on Log-Transformed Personal Exposure
Concentrations (log(PEM)), Before and After Adjusting for Paint Lead
Loading, in the CED Phase
Activity
Drilling
Sawing
Model Adjusted
for Paint Lead
Loadings?111
Unadjusted
Adjusted
Unadjusted
Adjusted
Parameter121
Wood vs. Plaster
Wood vs. Plaster
Paint
Wood vs. Plaster
Wood vs. Plaster
Paint
Estimate
0.798
0.374
0.109
1.60
1.56
0.00888
Standard
Error
0.529
0.602
0.0642
0.465
0.597
0.0606
Degrees
of
Freedom
8
7
7
4
3
3
P-Value
0.17
0.55
0.13
0.03
0.08
0.89
111 Indicates whether an effect for paint lead loading is included in the model.
121 "Wood vs. Plaster" represents the increase in log(PEM) associated with disturbing wood substrate versus plaster
substrate. "Paint" represents the increase in log(PEM) associated with a unit increase in paint lead loading.
The results in Table 8C-15 show that by including the effect of paint lead loading into the
model, the magnitude of the multiplicative substrate effect in the drilling activity has been reduced
from a 2.2 to a 1.45 fold increase when drilling into wood versus drilling into plaster (ea798 = 2.2,
e0374 = 1.45). However, the increases are not statistically significant at the 0.05 level given the
observed data and sample sizes. Essentially no reduction in the multiplicative substrate effect
occurs from adjusting the model for the sawing activity.
8-72
-------
The results presented in Table 8C-16 demonstrate that adjusting for the effect of paint
lead-loading causes a large reduction in the substrate effect on the intercept term. The substrate
effect for the slope of distance remains relatively stable between the two models. Moreover, the
paint lead loading effect is statistically significant (p=0.03) for both activities, while the effects of
wood versus plaster substrates are not.
Table 8C-16. Estimated Substrate Effect on Settled Log-Transformed Dust (log(Dust)),
Before and After Adjusting for Paint Lead Loading, in the CED Phase
Activity
Drilling
Sawing
Model
Adjusted for
Paint Lead
Loadings?111
Unadjusted
Adjusted
Unadjusted
Adjusted
Parameter121
Intercept
Distance
Intercept
Distance
Paint
Intercept
Distance
Intercept
Distance
Paint
Estimate
2.54
0.133
0.730
0.160
0.258
1.35
0.271
0.594
0.268
0.148
Standard
Error
1.18
0.321
1.329
0.412
0.114
1.34
0.254
1.359
0.278
0.066
Degrees
of
Freedom
9
9
8
9
33
6
6
5
6
24
P-Value
0.06
0.69
0.60
0.71
0.03
0.35
0.33
0.68
0.37
0.03
(2)
Indicates whether an effect for paint lead loading is included in the model.
"Intercept" represents the baseline increase in log(Dust) associated with disturbing wood substrate versus plaster
substrate. "Distance" represents the amount this increase in inflated by moving one foot closer to the activity. "Paint"
represents the increase in log(Dust) associated with a unit increase in paint lead loading.
Overall, the data collected in the CED phase is not sufficient to reach a conclusion on
whether the difference in lead paint concentration is responsible for differences in lead loadings
observed between activities performed on wood and plaster.
The Difference Between [0-1] Foot and [5-6] Foot Samples
If the second hypothesis is valid, then it is reasonable to assume that the estimated [5-6]
foot SSDCs would be positively correlated with worker personal exposure result within each
experimental unit, while the estimated [0-1] foot SSDCs would be positively correlated with the
observed average paint lead loadings from within each experimental unit. The estimated [0-1] and
[5-6] ft SSDCs within each experimental unit are calculated by integrating the area underneath the
exponentiated regression lines from 0 to 1 feet and 5 to 6 feet, respectively, based on parameter
estimates from Model (CED-3) in Appendix C.
8-73
-------
Table 8C-17 gives the estimated correlations between PEM^, 6-foot by 1-foot lead
loading gradient^, [0-1] foot SSDCp, [5-6] foot SSDQjk, and Paint^, where "ijk" represents the
same experimental unit. Values above the diagonal are based on the original scale, and values
below the diagonal are based on the log-transformed values within each experimental unit.
Table 8C-17. Correlations of PEM Lead Concentrations, Settled Dust Lead Loadings, and
Paint Lead Loadings, Within an Experimental Unit, in the CED Phase
PEMiik
Gradientiik
[0-1]SSDCiik
[5-6]SSDCiik
Paintijk
PEMiik
0.39
0.28
0.83
0.21
61 x r
Gradientiik
0.35
0.98
0.46
0.38
[0-1] Foot
SSDCiik
0.25
0.98
0.32
0.32
[5-6] Foot
SSDCiik
0.46
0.42
0.25
0.25
Paintiik
0.26
0.14
0.09
0.23
Note: Correlations above the diagonal are based on untransformed data. Correlations below the diagonal are based on log-
transformed data.
The above correlation table shows that the log(PEM) concentrations are most highly
correlated with the log lead amounts in the [5-6] foot region. This result supports the contention
that the estimated lead loading in the [5-6] foot SSDC is representative of airborne dust and
smaller particles. A high correlation (0.98) is also observed between the log lead amount within
the 6-foot by 1-foot dust gradient and the log lead amount within the [0-1] foot SSDC region on
both the log-transformed and untransformed scale, indicating that the gradient measure is being
driven primarily by the high lead loading levels observed near the point of activity.
8C-2.4 THE POTENTIAL FOR CROSS CONTAMINATION WITHIN R&R FIELD STUDIES
The experimental design of the CED phase specified several different R&R activities
within the same housing unit. Additionally, there were potentially two different CED activities
being conducted simultaneously (in different rooms) at any given time. Thus, it is possible that
the measures of potential lead hazard associated with one CED activity were cross-contaminated
by lead generated by another CED activity. This possibility was recognized in the experimental
design of the CED phase, and was made even more apparent during an activity in which a worker
cut into plaster with a circular saw causing plaster dust to be distributed throughout the dwelling
unit. The following precautionary measures were taken to minimize the effects of cross
contamination on the study results:
1. In preparation for the CED activities, the floors in each room were sealed with clear
plastic sheeting (held down with duct tape). This was done to reduce the effects of
settled dust that was present in the housing units prior to the CED phase.
8-74
-------
2. Whenever possible, the entrances and exits to a room in which a CED activity was
performed were sealed off with clear-plastic. This precautionary measure was used
to minimize the effects of cross contamination from one activity to another.
3. If there was any visible dust present in a room prior to the occurrence of a CED
activity, the entire room was HEPA-vacuumed and wet-wiped prior to the start of
the activity.
4. Tools that were used during a CED activity were cleaned appropriately following the
completion of that activity to reduce the potential for cross-contamination from one
activity to another.
We can assess the effect of cross-contamination on the results of the CED phase through
an investigation of some specific settled dust lead loading results in two ways:
1. One of the CED activities (drilling into wood) was performed on the same surface
twice. The first occurrence of this activity was performed directly before the sawing
into plaster activity. The settled dust samples from this first occurrence (collected
after the sawing into plaster) were very likely to contain settled dust and debris from
both the drilling into wood and the sawing into plaster. The second occurrence of
the drilling activity was isolated from all other activities. Thus, a comparison
between these two occurrences will give us an assessment of the extent of cross
contamination that had occurred during the sawing into plaster. These data were
investigated by comparing the two regression lines which predicted log(Dustijk]) as a
function of distance. The two intercept estimates were 10.4 and 13.5 while the slope
estimates were -1.19 and -1.73. There was no statistically significant difference
between the two regression lines at the 0.05 level of significance.
2. The third floor of the first housing unit had two bedrooms whose entrances were
separated by a narrow hallway. Two activities (sawing wood and door modification)
were performed simultaneously, one in each bedroom. The entrances to each
bedroom were sealed using the clear-plastic doorways. An SSDC was placed in the
hallway that separated these two bedrooms just prior to the start of these two
activities. This settled dust sample was specifically designed to measure the amount
of lead that is likely to escape through the clear-plastic doorways. The lead loading
on this SSDC was 432 |ig/ft2.
In the first assessment, we find that the fallout of fine dust and debris from the sawing/plaster
activity did not add a significant amount of lead to the settled dust samples in the drilling wood
activity. The second assessment demonstrates that a substantial amount (432 |ig/ft2) of lead may
escape the protective containment barriers. We feel that some of the lead that settled on this
SSDC is attributable to the lead that was carried out into the hallway on the clothing of the
workers. We also feel that it is unlikely that a substantial amount of lead-contaminated dust
would have exited through the plastic doorway from one room and entered through the plastic
doorway of another room during the course of the study.
8-75
-------
Clearly the potential for cross-contamination of environmental field samples existed in the
CED phase. Given the substantial amount of lead disturbed by the CED activities under
investigation, it is unlikely that the amount of cross-contamination that occurred during the CED
phase had a qualitative impact on the overall results.
8C-2.5 DISCUSSION OF THE RESULTS OF THE CED PHASE
Table 8C-18 presents estimates of observed lead levels associated with worker personal
exposure and lead in settled dust, for each CED activity. The model estimate of the geometric
mean of the personal exposure lead concentrations, and the estimated amount of lead in settled
dust within a 6-foot by 1-foot region from the activity are summarized in this table.
The major result of the CED phase becomes clear when studying Table 8C-18:
The amount of lead disturbed by the
CED activities was substantial.
Table 8C-18.
Statistical Estimates, with Standard Errors, of Worker Personal Exposure to
Airborne Lead, and 6' x 1' Gradient Dustfall Lead Amounts for Each CED
Activity
Activity
Substrate
Model Estimates
Geometric Mean of PEM
Concentration (//g/m3)111
(with LOG Standard Error)
Lead Amount in 6' x 1 '
Gradient (//g)121
(with Standard Error)
Target Activities in the CED Phase
Demolition
HVAC Removal
Plaster
Dust
108 (0.325)
49.6 (0.116)
19500 (14300)
7710 (14100)
Generic R&R Tasks
Door Modification
Sawing
Sawing
Drilling
Drilling
Abrasive Sanding
Component Removal
Cleanup
Cleanup
Wood
Wood
Plaster
Wood
Plaster
Wood
Wood
Wood
Plaster
590 (0.574)
546 (0.155)
110 (0.783)
15.1 (0.489)
6.76 (0.255)
254 (0.424) (Hand)
571 (0.602) (Power)
344(0.0319)
102 (0.145)
24.5 (0.275)
145000 (82700)
449000 (167000)
82300 (81400)
266000 (143000)
21400 (20600)
257000 (188000)
—
—
--
111 Geometric mean results from exponentiating the estimated intercept term of Model (CED-1) in Appendix C. The log
standard error is associated with the intercept estimate. These estimates are also found in Table 8C-4.
(21
Lead amount is estimated by fitting Model (CED-5) to SSDC settled dust lead loadings, then integrating under the curve
from 0 to 6 feet distance. This approach is detailed in Section 8C-2.3.2 and Appendix C.
8-76
-------
The estimated geometric mean worker personal exposure to airborne lead over the duration
of activity was above the OSHA PEL of 50 |ig/m3 in every CED activity with the exception of
drilling into either plaster or wood, HVAC removal, and cleaning up after activities which
disturbed lead painted plaster. The average lead-loading in the gradient region was greater than
1000 |ig/ft2 for all CED activities as they were performed.
Secondary results of the CED phase are:
1. CED activities which disturb lead-painted wood surfaces are generally associated
with higher personal exposure to airborne lead and higher lead levels in settled dust
when compared to CED activities which disturb lead painted plaster surfaces. This
effect is confounded with the fact that wood surfaces generally had higher
concentrations of lead-based paint in comparison to plaster surfaces.
2. Settled dust found in interior HVAC duct work had high levels of lead. This lead can
be exposed to workers and occupants if the HVAC systems are disturbed.
3-77
-------
8D-1.0 INTRODUCTION AND OBJECTIVES OF THE CLEANUP INVESTIGATION
As presented in Section 1.1, one of the primary technical objectives of the R&R study was
to determine the extent to which R&R workers disturb lead and create a lead-based paint
exposure to occupants or other exposed individuals while conducting R&R activities. The three
study phases of the EFSS (Sections 8A through 8C) collected lead loading data from settled dust
samples before cleanup in support of this objective. These data do not address the extent to
which typical cleanup procedures reduce lead levels in dust that settles following the activity. In
response to this concern, the EFSS included the design and conduct of a field study addressing the
effect of typical cleanup procedures on potential occupant lead exposures.
The EFSS cleanup investigation had the following technical objectives:
• Compare post-activity lead loading results within settled dust collected before
cleanup versus results in dust collected after cleanup, and determine how the
outcome of the comparison is a function of the cleanup method used.
• Characterize lead loadings in settled dust that is likely to remain after R&R work and
typical cleanup procedures are completed.
The cleanup investigation in the EFSS was designed as an independent follow-on to the
CED study phase presented in Section 8C. As in the CED phase, the cooperation of the Denver
(CO) Housing Authority (DHA) was solicited to provide vacant housing units for conducting
simulated R&R activities in a controlled environment. An apartment complex previously used in
the CED phase (and thereby screened for the presence of lead-based paint hazards) was the site
for the cleanup investigation. As they did in the CED phase, the DHA provided trained abatement
workers to conduct the R&R activities in these units. The workers used the same protocols
established for the CED study phase.
8D-2.0 STUDY DESIGN IN THE CLEANUP INVESTIGATION
To address the objectives for the cleanup investigation, the study design considered two
different methods of cleanup that were applied after performing two different types of generic
R&R activities. The two cleanup methods were selected as those typically used by many R&R
workers:
• dry (broom) sweeping, and
• use of an industrial non-HEPA vacuum cleaner (Shop-Vac®).
The cleanup procedures were performed as if the R&R worker had completed carpentry work in a
private residence and wanted to clean the work area prior to the completion of the job.
The two R&R activities, both conducted on wood door surfaces covered with lead-based
paint, were:
8-78
-------
• drilling (using a Vi" drill bit), and
• abrasive (power) sanding.
Protocols for conducting these activities were the same as those used in the CED phase (see
Section 8C-1.1.1). Each activity was scheduled to be no longer than one hour in duration. Each
combination of R&R activity and cleanup method was replicated three (3) times, for a total of
twelve (12) field study experiments. These experiments were labeled "experimental units".
As a means of minimizing cross-contamination across units, the study directed that the
twelve experimental units be conducted in separate apartments within the complex. However, as
only nine apartments were made available to this study, four of the experimental units were
conducted in different rooms of a single apartment, while the remaining experimental units were
conducted in separate apartments. Other precautionary measures taken to minimize cross-
contamination from other sources included:
• The floors, vents, and entrances of the activity room were sealed with clear plastic
sheeting (held down with duct tape), to reduce the effects of settled dust present in
the apartment prior to the investigation and to minimize the transfer of dust from one
activity room to another.
• To characterize the level of any potential contamination that could have resulted
from the demolition of buildings within the complex that occurred simultaneously
with the investigation, two paired dust-vacuum samples were collected from SSDCs
that were left overnight in previously unused activity rooms.
As was the practice throughout the EFSS, no attempt was made to minimize dust generation
during the activity.
8D-2.1 SAMPLING DESIGN IN THE CLEANUP INVESTIGATION
The sampling design of the cleanup investigation was established to characterize the
following:
• levels of lead in the paint on surfaces disturbed by the R&R activity
• lead loadings in dust that settles following completion of the R&R activity, taken at
varying distances from the activity both prior to and following cleanup
• lead loadings in dust that settles up to one day following the activity and cleanup
• how lead loadings differ between dust-vacuum samples and dust-wipe samples
• how lead loadings differ between dust-vacuum samples taken at adjacent locations
("side-by-side").
8-79
-------
Within an experimental unit, an area was identified where the specified R&R activity and
subsequent cleanup method would be performed. The door on which the activity would be
performed was confirmed to contain lead-based paint by use of a portable X-ray fluorescence
(XRF) analyzer. In addition, a single paint chip was taken from the door surface for laboratory
analysis to more accurately characterize lead levels in the paint that would be disturbed as a result
of the activity.
Immediately prior to the start of the activity within an experimental unit, a clean 4'x8'
piece of new linoleum sheeting (mounted on plywood) was placed on the floor at the base of the
activity. The linoleum represented a pristine floor surface that became contaminated only by the
conduct of the R&R activity. The design associated with collecting settled dust samples from the
linoleum surface is presented in Figure 8D-1.
Approximately one hour after the completion of the activity, but prior to the start of
cleanup, dust-vacuum samples were collected from three one-square-foot areas of the linoleum
surface, each at a specified distance from the activity (0, 3, and 6 feet). The three samples
represented the total amount of lead that was disturbed by the activity and settled at the given
distances from the activity.
After the three pre-cleanup dust samples were collected, the linoleum surface was cleaned
by the method assigned to the experimental unit. Then, six additional one-square-foot areas were
identified on the linoleum surface from which post-cleanup settled dust samples were collected
(Figure 8D-1). Four of the six samples consisted of two pairs of dust-vacuum/dust-wipe samples
(i.e., one sample within the pair collected by vacuum techniques, the other collected by wipe
techniques), where the two samples within a pair were taken from adjoining locations (i.e., side-
by-side). The remaining two samples were a side-by-side pair of dust-vacuum samples. Wipe
samples were taken in order to compare results to lead clearance testing standards. Once the
post-cleanup dust samples were taken, two stainless steel dustfall collectors (SSDCs) were placed
on the linoleum surface. A dust-vacuum sample was collected from each SSDC on the morning
of the next day. These samples were taken to characterize the amount of airborne lead that is
likely to settle after cleanup has been performed.
Table 8D-1 presents the proposed types and numbers of environmental field samples (both
regular and QC samples) collected within each experimental unit in the cleanup investigation.
Methods used to collect paint and dust samples followed the same field sampling protocols
applied throughout the R&R study.
8D-3.0 STUDY RESULTS FOR THE CLEANUP INVESTIGATION
Analytical data were available for statistical analysis from all environmental field samples
that were collected in the cleanup investigation. These data included lead loadings in paint chip
samples (expressed in mg/cm2) and settled dust samples (expressed in |ig/ft2). Data are
summarized by sample type and collection method in Tables Cl-1 through Cl-4 in Section A.4 of
Appendix A.
8-80
-------
0'
r
2'
3'
4'
5'
6'
7'
R'
DOOR
PRE
POST
(V)
POST
(W)
PRE
POST
(V)
POST
(V)
SSDC
"*^
PRE
POST
(V)
POST
(W)
SSDC
WALL
^ 8' x 4'
new linoleum
surface
1' 2' 3' 4'
Note: Positioning of samples within a given distance from activity differed among the experimental units.
Legend:
PRE
POST
(V)
POST
(W)
SSDC
= Pre-cleanup dust-vacuum sample location
= Post-cleanup dust-vacuum sample location
= Post-cleanup dust-wipe sample location
= Next-day SSDC dust-vacuum sample location
Figure 8D-1. Settled Dust Sampling Locations Specified by the Sampling Design for the
Cleanup Investigation
8-81
-------
Table 8D-1. Numbers and Types of Environmental Samples (Regular and QC) Proposed in
the EFSS Cleanup Investigation
Regular Samples
When Sample Was
Taken
Pre-Activity
Post-Activity, Pre-
Cleanup
Post-Cleanup
Next Day
Sample Collection
Method
Chipping
Vacuum
Vacuum
Wipe
Vacuum
Sample Type/Location
Paint chip (wood door)
Settled dust (linoleum
surface)
Settled dust (linoleum
surface)
Settled dust (linoleum
surface)
Settled dust (SSDC)
TOTAL
Number
Proposed per
Exp. Unit111
1
3
4
2
2
12
Total
Number
Proposed
12
36
48
24
24
144
Field PC Samples
When Sample Was
Taken
Pre-Activity or Pre-
Sampling
Sample Collection
Method
Paint Chip
Vacuum
Wipe
QC Sample Type
Field blank
Field blank
Field blank
Proposed
Frequency of
Sampling
1 per activity
day
1 per vacuum
sampling day
1 per wipe
sampling day
TOTAL
Total
Number
3
4
3
10
111 See Figure 8D-1 for position of dust-vacuum and dust-wipe sample locations relative to the R&R activity.
8D-3.1 LEAD LEVELS IN PAINT TO BE DISTURBED
To characterize lead levels in paint to be disturbed as a result of R&R activity, paint chip
samples were taken from each of the twelve wood doors on which the activities would take place.
Three paint chip results were associated with each of the four combinations of activity and
cleanup method, one result for each time the activity/cleanup combination was performed in this
investigation. The measured lead levels in paint chips are summarized in Table Cl-1 of Appendix
A (Section A.4). The combination with the highest observed lead levels in paint was abrasive
sanding/vacuum cleanup, where all three lead levels exceeded 12 mg/cm2. For drilling/vacuum
cleanup and abrasive sanding/broom sweeping, two of the three paint chip lead results exceeded
1.0 mg/cm2. Only one of the three results exceeded 1.0 mg/cm2 for drilling/broom sweeping.
Even in situations where low results were observed, however, conduct of the activity resulted in
high levels of lead in post-activity settled dust samples as illustrated in Table Cl-2 in Appendix A
and discussed below.
8-82
-------
8D-3.2 PRE- VERSUS POST-CLEANUP RESULTS
In characterizing the difference in post-activity settled dust lead loadings between pre- and
post-cleanup, loadings from post-cleanup dust-wipe samples were compared to the pre-cleanup
dust-vacuum samples at a given distance from the activity. Wipe samples rather than vacuum
samples were chosen to characterize post-cleanup lead loadings, as only small amounts of dust
tended to remain following cleanup. Wipe sampling techniques are typically more efficient at
measuring lead loadings from smooth surfaces with small amounts of dust, while vacuum
techniques are more appropriate when large amounts of dust are to be collected. This is evident
from Table Cl-3 in Appendix A (Section A.4), where dust-wipe lead loadings were consistently
higher than vacuum-wipe lead loadings at the same distance from the activity for the less-dirty
drilling activity, while the highest lead loadings among the more-dirty abrasive sanding activity
were observed among vacuum-dust samples.
For each target activity and cleanup method, Table 8D-2 presents the geometric mean lead
loading in pre-cleanup dust-vacuum samples, along with the average ratio of post-cleanup dust-
wipe sample loading to the pre-cleanup sample loading observed on the same surface at the same
distance from the activity. At each distance, pre-cleanup loadings tended to be higher for abrasive
sanding than for drilling. Reasons for this finding could include higher lead loadings in the paint
associated with abrasive sanding activities, as well as differences in the activities. At the zero-feet
distance, post-cleanup lead loadings averaged one percent or less of the corresponding pre-
cleanup lead loadings for each activity and cleanup method. Based on results of a paired t-test on
log-transformed lead loadings, these declines are statistically significant at the 0.05 level. In
contrast, little if any reduction is observed at the six-feet distance. For all but abrasive
sanding/vacuum cleanup, the average ratio of post-cleanup to pre-cleanup results on the same
floor surface exceeded one. In each case, however, the paired t-test results indicated that the
average ratio did not statistically differ from one. This implies that the cleanup methods
conducted in this investigation do little to cause statistical changes in dust lead levels at this
distance. Therefore, one can conclude that the efficiency of cleanup, as measured by a percent
reduction of lead loadings in remaining dust, declines as distance from the activity increases.
This conclusion holds for both activities considered in the investigation, even though pre-cleanup
lead loadings tended to be higher for one activity.
The same statistical methods used in the CED phase to characterize lead levels in settled
dust were applied in the cleanup investigation for both pre-cleanup and post-cleanup sample lead
loadings. This approach involved fitting a regression model for each combination of R&R activity
and cleanup method to express log-transformed lead loadings as a function of distance from
activity and experimental unit. A form of the "population" random effects model (model
8-83
-------
Table 8D-2. Geometric Mean of Pre-Cleanup Lead Loadings, and the Average Ratio of Post-
to Pre-Cleanup Lead Loadings on the Same Floor Surface
R&R Activity
Drilling
Drilling
Abrasive Sanding
Abrasive Sanding
Cleanup
Method
Broom
Vacuum
Broom
Vacuum
0 Feet from Activity
Geometric
Mean Loading
(Pre-Cleanup)
26,700
73,500
653,000
203,000
Average Ratio of
Post- to Pre-Cleanup
Lead Loadings on a
Floor Surface
0.013"
0.008"
0.007"
0.010"
6 Feet from Activity
Geometric
Mean
Loading (Pre-
Cleanup)
65.0
146
1380
491
Average Ratio of
Post- to Pre-Cleanup
Lead Loadings on a
Floor Surface
3.65
1.12
1.77
0.69
Note: Geometric means in this table are rounded to three significant figures.
The average difference (taken across experimental units) between (log-transformed) post-cleanup result and
pre-cleanup result on a given floor surface is significantly different from 0 at a 0.05 level.
(CED-5) of Appendix C) was fit separately for pre- and post-cleanup sample lead loadings and for
each combination of activity (i) and cleanup method (j):
log(Dustm) =
0(ij)
*>mDistancem + R^
Rmf>istance(
where
Dust(ij)kl
is the dust lead loading (|ig/ft2) measured on the 1th sample from the ktt
occurrence of activity i and cleanup method j
Piffl
iffl)
Distance(ij)kl is the distance (ft) from the activity of the 1th sample from the kth occurrence
of activity i and cleanup method j
P0(ij) is the baseline average log-loading for samples associated with activity i and
cleanup method j
is the slope relating log-loading to distance from activity for samples
associated with activity i and cleanup method j
is a random effect which represents how the baseline average log-loading
for the kth occurrence of activity i and cleanup method j differs from the
overall average P0(ij)
is a random effect which represents how the slope associated with the kth
occurrence of activity i and cleanup method j differs from the overall slope
Pi®
is the error term.
e
(ij)kl
8-84
-------
The goal of fitting this model to the observed dust-lead loading data was to estimate average lead
loading in settled dust, expressed as a function of distance from the activity. A curve quantifying
this relationship was determined for each combination of activity and cleanup method, and for
pre- and post-cleanup lead loadings. The curve expresses (untransformed) average lead-loadings
as follows:
Lead-loading([tg/ft2) =
where the slope and intercept terms were estimated from the model fit. By integrating the area
underneath this curve from distance=0 to 1 foot, we obtain the expected lead-loading of an area
within one foot of the activity. Similarly, by integrating this curve from distance=5 to 6 feet, we
obtain the expected lead-loading of an area located 5-6 feet from the activity. The measure that
was chosen in the R&R study to represent lead disturbance and the potential hazard to occupants
from lead in settled dust is the lead loading within a 6-foot by 1-foot gradient region extending
from the activity. This measure was determined by integrating the area underneath the estimated
curve from distance=0 to 6 feet. See Section C.5 of Appendix C for more detailed discussion of
the gradient lead-loading.
The estimated average pre- and post-cleanup lead loadings from 0-1 feet and 5-6 feet from
the activity, and the estimated gradient lead loading are presented in Table 8D-3 for each
combination of activity and cleanup method. Also included in this table are the standard errors
associated with these estimates, which were obtained by the Delta Method (Bishop, Fienberg, and
Holland, 1975) (see Section C.5 of Appendix C). It should be noted that while the sample sizes
(i.e., number of experimental units) within each activity/cleanup method combination are relatively
small, the Delta Method is based in asymptotic normal theory. As a result, the standard errors are
likely only rough approximations.
The estimated lead loadings in the 0-1 foot area and 5-6 foot area have different
interpretations. The 0-1 foot area loading is representative of the amount of lead that is likely to
fall directly underneath the surface being disrupted, while the 5-6 foot area loading is more
representative of the amount of lead that becomes airborne in dust and small particles and settles
at a distance from the activity. By comparing these two estimates, one gains some perspective on
the amount of disturbed lead that is likely to stay localized, versus the amount that is likely to
become airborne.
For both activities in the cleanup investigation, approximately 65% of the lead in the
gradient settled within one foot of the activity, even though higher lead loadings (both pre- and
post-cleanup) were associated with abrasive sanding than with drilling. Therefore, while
differences in lead loadings may exist among the activities, the relative distribution of lead within
6 feet of the activity was consistent. This type of information may be useful in the design of final
cleanup procedures for R&R workers to minimize occupant exposures resulting from the activity.
8-85
-------
Table 8D-3. Estimated Lead Loadings (//g/ft2, with Standard Errors) for Pre- and Post-
Cleanup, and Percent Reduction from Pre- to Post-Cleanup Periods, As
Estimated from Statistical Modeling Procedures
R&R
Activity
Drilling
Drilling
Abrasive
Sanding
Abrasive
Sanding
Method of
Cleanup
Broom
Vacuum
Broom
Vacuum
Time of Sample
Collection
(Post-Activity)
Pre-Cleanup
Post-Cleanup
% Reduction
Pre-Cleanup
Post-Cleanup
% Reduction
Pre-Cleanup
Post-Cleanup
% Reduction
Pre-Cleanup
Post-Cleanup
% Reduction
6' x 1' Gradient
(//g/6ft2)
22,400 (27,900)
885 (661)
96.0%
54,300 (126,000)
1450 (1250)
97.3%
305,000 (608,000)
5800 (6350)
98.1%
269,000 (566,000)
3350 (1240)
98.8%
Lead Loadings
0 to 1 foot from
Activity (//g/ft2)
14,200 (18,400)
166 (122)
98.8%
35,100 (86,400)
361 (382)
99.0%
196,000 (426,000)
1070 (1120)
99.5%
171,000 (365,000)
809 (282)
99.6%
5 to 6 feet
from Activity
(//g/ft2)
94 (128)
130 (99)
(38.3%)
197 (243)
147 (94)
25.4%
1155 (1738)
865 (999)
25.1%
1129 (2101)
357 (192)
68.4%
On an average basis across the investigation, Table 8D-3 shows that for both activities and
under both cleanup methods, post-cleanup lead loadings were more than 95% reduced from pre-
cleanup levels over the entire gradient region, as well as within one foot of the activity. However,
at 6 feet from the activity, only marginal reductions in lead loadings were observed on a
percentage basis. There is evidence from Table 8D-3, however, that vacuum cleanup methods
may perform better than dry sweeping methods at the 6-foot distance when considering the
percent reduction in lead loadings from pre- to post-cleanup. These results were consistent for
both activities and with the results of the paired analysis summarized in Table 8D-2.
8D-3.3 POST-CLEANUP RESULTS
Post-cleanup lead loadings for samples collected by wipe sampling techniques were used
to characterize the amount of lead remaining following the activity and cleanup to which
occupants may be exposed. Wipe sample results were directly comparable to the EPA interim
health-based standard of 100 |ig/ft2. As a result, wipe sample results were more relevant than
vacuum sample results for making conclusions on lead levels relative to potential occupant
exposure.
8-86
-------
As seen in Tables 8D-2 and 8D-3, post-cleanup lead loadings were substantially reduced
from pre-cleanup levels. However, the post-cleanup values often remained above EPA's internal
health-based standard of 100 |ig/ft2, even at 6 feet from the activity. For both cleanup methods,
post-cleanup lead loadings tended to be higher for the abrasive sanding activity than for the
drilling activity. From these results, one may conclude that the typical cleanup methods
considered in this investigation reduce available lead loadings on surfaces, but potentially more
efficient methods (or repeated applications of the typical methods) are necessary to clean surfaces
to meet indicated standards.
8D-3.4 NEXT-DAY SSDC LEAD LOADING RESULTS
To characterize the amount of airborne lead that settles after cleanup, dust-vacuum
samples were collected the morning following activity and cleanup from two SSDCs positioned
on the linoleum surface upon conclusion of cleanup. The SSDCs were placed at 0 and 6 feet from
the activity (see Figure 8D-1). Table Cl-4 of Appendix A (Section A.4) provides descriptive
statistics of dust lead loading at both distances for each combination of activity and cleanup
method. The lead levels in these samples were low to moderate; the geometric mean exceeded
100 |ig/ft2 in only one situation (abrasive sanding, broom sweeping, at 5-6 foot distance).
Interpreting the results of the next-day SSDC dust samples was complicated by the
potential for contamination by sources other than the R&R activity, such as the demolition of
neighboring buildings within the apartment complex. To characterize the amount of
contamination from outside sources, four vacuum samples were collected from SSDCs that were
left overnight in previously unused activity rooms. The lead loadings in these four samples ranged
from 28.0 to 311 |ig/ft2. Because these samples show high potential for cross-contamination from
outside sources, it is possible that lead levels in the next-day SSDC dust samples were affected by
these sources. However, the settled dust results from pre- and post-cleanup samples were likely
not affected by outside sources, as they were collected within one hour of activity or cleanup.
8D-3.5 CONCLUSIONS
The following conclusions can be made as a result of the findings determined from the
cleanup investigation:
• Both cleanup activities considered in this investigation led to substantial reductions
in lead loadings in settled dust. This result was more apparent immediately adjacent
to the activity, where post-cleanup lead loadings were more than 95% reduced from
pre-cleanup levels. However, the efficiency of cleanup declined as distance from
the activity increased. The extent of decline in efficiency, however, was more
apparent for dry sweeping methods than for vacuum cleanup methods.
8-87
-------
For purposes of planning cleanup strategies, it was determined that approximately
65% of the lead in the estimated 6-foot by 1-foot gradient for both activities settled
within one foot of the activity. This result held regardless of the magnitude of lead
loadings existing within 6 feet of the activity.
Despite performing the cleanup procedure, post-cleanup lead loadings were
observed to exceed EPA's internal health-based standard of 100 |ig/ft2, even at 6
feet from the activity. Post-cleanup lead loadings tended to be higher for the
abrasive sanding activity than for the drilling activity for both cleanup methods. As
a result, potentially more efficient cleanup methods (or repeated applications of the
methods considered in this investigation) are necessary to clean surfaces to at or
beyond indicated standards, despite the marked decrease in lead loadings that can
result from conducting cleanup procedures.
8-88
-------
8E-1.0 DESCRIPTION OF RELEVANT DATA LOCATED FROM OTHER SOURCES
The approach taken to uncover other extant sources of data that relate R&R activities to
lead exposures is discussed in detail in Section 2.4 of Chapter 2.
The minimum requirements for accepting a source of extant data as applicable to the
objectives of the EFSS included the following:
1. The source needed to provide data related to activities that can be considered
representative of renovation and remodeling work as it is currently being conducted
in an unregulated environment.
2. The source needed to provide either task-length personal exposure measurements
or settled dust measurements related to a well-specified R&R target activity,
generic task, or combination of R&R activities.
3. The activities needed to be conducted in an environment with lead-based paint
present. (Specifically, the lead in paint chips measurements from surfaces in the
activity area needed to be above 0.5% by weight or 1.0 mg/cm2.)
4. The sampling and analysis methods needed to follow standard protocols and meet
minimum quality assurance standards.
Eight studies were identified as having collected data that met these requirements. The
studies included data on cleanup, component removal, demolition, exterior siding, surface
preparation, and window replacement. Surface preparation was the only activity not monitored in
the EFSS for which sufficient data are available from other sources to allow for estimating
average exposure levels. All studies collected task length personal exposure data and provided
information on lead levels in paint on surfaces to be disturbed by the activity. In addition, one
study included data from settled dust samples collected on 1 ft2 tiles. Descriptions of the eight
sources of data are given below:
CONSAD Research Corporation
Under contract with OSHA, CONSAD Research Corporation in Pittsburgh, Pennsylvania,
collected personal exposure data from a series of construction projects throughout the United
States. The exposure information was collected to provide data for OSHA's Interim final
standard for lead in construction. The documented studies ranged from large industrial paint
removal projects to residential window replacement. Three studies, performed between March
1991 and January 1993, involved R&R activities targeted in the EFSS. The first study involved
surface preparation — removing lead-based paint from metal door jams and from walls and
ceilings by dry scraping. The second study, characterized as component removal, documented the
removal of interior wood trim using a hammer and claw-bar. The third study monitored the
demolition of interior walls and ceilings in a home using hammers and claw-bars. These studies
were performed in situations which reflected a typical R&R setting. Although some of the
8-89
-------
workers were familiar with lead-dust minimization practices, no such practices were used during
the monitored studies.
NIOSH, Ohio University Cleanup
During a three-day period in early May 1992, NIOSH conducted a comparison study of
lead paint cleanup methods on the campus of Ohio University in Athens, Ohio. The building in
which the study was conducted was built in 1873, had been unoccupied for many years, and had
much loose and peeling paint on the walls. Although the study was classified by NIOSH as a
study of cleaning methods, it is considered a combined study of both surface preparation and
cleanup relative to the EFSS definition of target activities. Three methods of surface preparation
and cleaning were compared:
• dry scraping followed by dry broom sweeping
• misting painted surfaces with water and scraping (wet scraping) followed by high-
efficiency particulate air-filtered (HEPA) vacuuming
• wet scraping followed by HEP A vacuuming, with a HEP A-filtered air-filtration
device (AFD) placed in the room to exhaust room air to the outside.
Eighteen rooms were cleaned by three crews of two workers, with each crew employing the three
cleaning methods in two rooms each. All the work was performed on one day. All 18 rooms had
detectable lead concentrations in paint chips, but nine did not have mean levels above the federal
lead-based paint criterion of 0.5% lead by weight. The nine that did have lead levels in paint chips
above 0.5% had levels ranging from 2.8% - 19% lead by weight (NIOSH, Ohio University, May
1993). Since the building as a whole was considered to be highly contaminated with lead, all
results were included in the analysis.
To compare the personal exposure data collected in the Ohio University study to data
from the EFSS, the three cleaning methods were subsetted into two categories: dry methods and
wet methods. The dry methods included only the dry scraping with dry sweeping, while the wet
methods included wet scraping followed by HEPA vacuuming, as well as wet scraping followed
by HEPA vacuuming and use of a negative-air AFD.
National Association of Home Builders (NAHB)
The National Association of Home Builders (NAHB) provided personal exposure
information for a case study of the airborne lead produced when painted wooden shingles were
removed from a garage in Atlantic City, New Jersey. The garage had two layers of shingles, with
only the top layer painted with lead-based paint. For approximately three hours, a home owner
removed shingles from a 260 square-foot section of his garage using a flat pry bar to remove the
shingles and a hammer to remove the nails. During the process, some of the shingles split,
causing dust to be generated. While only one personal exposure data point was reported for this
8-90
-------
activity, this is the only source of information on the possible lead exposures generated when
exterior siding is removed.
New York State Department of Health
This study was conducted by the New York State Department of Health in Albany, New
York, during February 1994. The purpose of the study was to characterize the personal
exposures residential house painters encounter while performing various R&R activities. The
activities monitored included manual scraping with a putty knife, manual sanding, and window
removal, with many of the activities occurring both inside and outside of the homes. The
activities were classified into R&R target activities as follows: manual scraping and manual
sanding as surface preparation (exterior and interior), and window removal as window
replacement.
One of the two personal exposure samples collected during window replacement was
taken in an environment where the lead concentration in paint chips was reported as 0.22%.
However, the corresponding personal exposure result was 151 |ig/m3, indicating significant
exposure even in the absence of high levels of lead in the paint (New York SDH, February 1994).
Because of the information on potential lead exposure offered by this sample, the result was
included in the analysis despite the lead concentration in paint chips being below 0.5%.
NIOSH, Cincinnati - "People Working Cooperatively"
From June 1993 through June 1994, NIOSH recorded the personal exposures of
individuals working in the "People Working Cooperatively" program in Cincinnati, Ohio. The
workers in this program provide home repair, emergency repair, and weatherization services to
low-income elderly, disabled, and other qualified single-family homeowners. In this study, the
homes visited were generally built before 1960. Target R&R activities for which personal
exposures were observed included exterior wet and dry scraping and window replacement. The
exterior wet and dry scraping activities were considered representative of surface preparation.
Reported data for other target activities, such as demolition, were excluded from analysis because
the lead concentration in paint chips was either not reported or less than 0.5%.
U.S. Army Environmental Hygiene Agency
The U.S. Army Environmental Hygiene Agency conducted a study of lead exposure levels
generated by five different surface preparation methods performed over a five-day period in April
1992. The study was conducted in four military buildings in Maryland scheduled to be
demolished: three housing buildings and one office building. During the five-day period, personal
exposure levels were collected from six maintenance workers on three-hour shifts, using one of
five surface preparation methods. The surface preparation methods were wet scraping, dry
scraping, wet wire brushing, dry wire brushing, and metal wheel brushing. The purpose of the
study was to evaluate lead exposures generated by the different surface preparation methods and
to determine whether regulations needed to be implemented to prevent high exposures. In order
to compare results with the EFSS, the results from this study were identified by sample method
8-91
-------
(wet/dry) and location (interior/exterior). The metal wheel brush method was categorized as a
dry interior method.
U.S. Air Force
The Occupational and Environmental Health Dictorate of Armstrong Laboratory, Brooks
Air Force Base, Texas, conducted a study to determine if R&R work performed on Air Force
Military Family Housing (MFH) resulted in personal air exposures above the OSHA action level
of 30 |ig/m3. R&R work was performed on three Air Force bases throughout the country. The
activities performed were characterized as renovation, maintenance, and deleading. Deleading
activities were classified as component removal, since the actual deleading of the components was
performed off-site, and the exposure samples were collected while the components were being
removed. During the study, more than 200 personal air samples were collected. Unfortunately,
most of the personal exposure samples were associated with activity areas in which levels of lead
in paint chips were not above 0.5% concentration by weight or 1 mg/cm2. This eliminated over
75% of the available data points according to the minimum acceptance criteria stated earlier.
Activities observed for samples that did meet these criteria were classified into targeted R&R
activities as follows:
• Cleanup: Renovation cleaning
• Component Removal: Renovation baseboard removal and door frame removal
• Exterior Surface Preparation: Exterior renovation by hand sanding and scraping
• Interior Surface Preparation: Interior maintenance hand sanding and scraping and
renovation hand sanding.
Massachusetts Department of Health
The Massachusetts Department of Health, Childhood Lead Poisoning Prevention
Program, conducted a study in 1993 and 1994 to assess worker exposure to lead during surface
preparation activity. This activity prepared surfaces for abatement by encapsulation. Personal air
samples and dust wipe samples were collected at four housing units during interior surface
preparation, using both dry and wet methods. Dry methods included dry sanding and/or scraping
and feathering of edges by hand sanding, followed by a wash down with tri-sodium phosphate
(TSP). Wet methods included misting of the surface during sanding and/or scraping and
feathering of edges by hand sanding, followed by a wash down with TSP. A total of 25 specific
individual tasks were monitored, with one personal exposure sample taken for each task. Of
these, five had results less than the limit of detection (LOD), which ranged from 30 |ig/m3 to 90
|ig/m3. The very high LODs for these samples were a combination of low sample volume and
relatively high instrument LODs. Because the range of possible values for these samples was so
large, it was felt that they contained no quantitative information on worker exposures, an opinion
shared also by the Massachusetts field team industrial hygienist and the laboratory chemist.
Therefore, these five results were not included in any analysis of the data.
8-92
-------
In addition to the task-length average (TLA) worker exposure samples, settled dust
samples were collected from 1 ft2 ceramic tiles. At all units, three tiles were placed in the center
of the activity room and sampled for settled dust by wipe methodology at three times: prior to
start of the surface preparation activity, immediately after activity, and at least ten hours after the
activity. In three of the four units, an additional tile was placed in the center of the room after
completion of the activity and cleanup by HEP A vacuuming. A wipe dust sample was collected
from this tile approximately ten hours after placement. In one unit, one additional tile was placed
in each of two adjacent non-activity rooms after completion of the surface preparation activity and
HEPA vacuuming. These two tiles were sampled approximately ten hours after placement.
Table 8E-1 contains a summary of the target activities, lead percentages in paint, XRF
measurements, average task lengths, percent of personal exposure samples above the detection
limit, and the personal exposure sample sizes for each of the eight studies. Also included are
whether the activities took place inside or outside of the building, and whether wet or dry
methods were used.
Table 8E-1. Summary of the Presence of Lead-Based Paint, Task Lengths, Personal
Exposure Sample Sizes, and Percent of Personal Exposure Samples Above
Detection Limit by Study and Target Activity
Study
CONSAD
NIOSH, OU
NAHB
New York
NIOSH,
Cincinnati
U.S. Army
Hygiene
Brooks AFB
Massachusetts
DPH
Target
Activity
Component Removal
Demolition
Surface Preparation
Cleanup/Surface Prep
Cleanup/Surface Prep
Exterior Siding
Window Replacement
Surface Preparation
Window Replacement
Surface Preparation
Surface Preparation
Cleanup
Component Removal
Surface Preparation
Surface Preparation
Location/
Method
Interior/Dry
Interior/Dry
Interior/Dry
Interior/Dry
Interior/Wet
Exterior/Dry
Interior/Dry
Interior/Dry
Exterior/Dry
—/Dry
Exterior/Dry
Exterior/Wet
Interior/Dry
Interior/Wet
Exterior/Dry
Exterior/Wet
Interior/Dry
Interior/Dry
Exterior/Dry
Interior/Dry
Interior/Dry
Interior/Wet
Lead Levels in Paint
Average %
Paint Lead
0.52
13.0
0.53
3.2
4.8
2.69
3.2
6.2
11.6
15.7
20.4
5.3
5.6
2.0
4.25
3.18
0.83
0.92
8.2
1.9
___
—
XRF
(mg/cm2)
___
___
___
___
___
___
___
___
___
___
___
___
___
___
___
___
4.3
4.2
9.8
5.2
6.6
9.8
Personal Exposure Monitoring111
Average
Task
Length
(minutes)
252
178
430
27
33
137
139
152
240
181
265
329
168
170
184
182
43
75
246
198
101
109
% Samples
Above
Detection
Limit
100
100
100
100
100
0
100
100
100
100
100
100
100
67
0
0
0
38
17
20
100
100
Sample
Size
(N)
5
2
4
12
24
1
2
1
5
8
15
7
8
15
12
12
3
8
6
15
3
17
Only data included in data analyses within this section are included in summaries within this table.
8-93
-------
8E-2.0 SUMMARY OF PERSONAL WORKER EXPOSURE DATA FROM OTHER
SOURCES
Summary statistics of personal exposure data from the other data sources are listed in
Table 8E-2 according to specific R&R activities and tasks. For comparison purposes, summary
statistics are also included for data collected in the EFSS. In calculating summary statistics for
this table, all sample results reported below the limit of detection were set to one-half the LOD.
(Setting not-detected results to 0 or to the LOD had a negligible effect on the calculated summary
statistics.) All results represent task length averages (TLAs) in |ig/m3.
8E-2.1 COMPARISON OF PERSONAL WORKER EXPOSURE DATA FROM OTHER
SOURCES WITH EFSS RESULTS
Window Replacement
Figure 8E-1 presents personal worker exposure results for window replacement from the
NIOSH-Cincinnati study, the New York State Department of Health study, and the EFSS. The
NIOSH and EFSS results, both based on eight monitored workers, were very similar. In contrast,
the two workers monitored in the New York study have much higher exposures, both more than
twice the OSHA permissible exposure limit (PEL) of 50 |ig/m3. The New York data are in line
with the estimated distribution of exposures for window replacement from the EFSS data (as
presented in Table 9-1 of Chapter 9), and therefore indicate that the potential for achieving lead
exposures above the OSHA PEL during window replacement activities is not negligible.
Demolition
Figure 8E-2 compares personal worker exposures during demolition for the CONSAD
study and the EFSS. During the CONSAD study, two samples were collected using a Marple
Cascade Impactor, which involves a series of filters that separate particles by particle size. The
average estimated exposure using this collection method was 976 |ig/m3, which was much higher
than the levels associated with two study samples collected using a 37 mm filter and NIOSH
protocol 7300. Because CONSAD reports that their experience indicates that the impactor
collection method tends to produce higher results (for which they have no explanation), the
impactor results were not included in Table 8E-2 or Figure 8E-2.
Component Removal
Personal worker exposure estimates associated with component removal are presented in
Figure 8E-3 for the Brooks AFB study, CONSAD study, and the EFSS. The activity in all three
studies included removing lead-painted wood trim, baseboards, doors and door jams. The EFSS
results are markedly higher than those obtained in the Brooks AFB and CONSAD studies.
Personal exposure results from the Brooks AFB study (over all monitored activities) tend to be
low compared to data from other studies (see Table 8E-2). This might be related to
characteristics of the military family housing in which the monitoring took place in that study.
The average lead concentration in paint chips in both the CONSAD study and the Brooks AFB
8-94
-------
study was low: 0.92% and 0.52% lead by weight respectively. In contrast, lead loading
associated with component removal activities in the EFSS averaged 7.4 mg/cm2, and the activities
were conducted by two workers in a home that had been vacant for some time. However, this
alone does not explain such high exposure levels in the EFSS.
Table 8E-2. Summary of Personal Worker Exposure Levels as Measured by a Task Length
Average (//g/m3) for the Other Sources of Data and for Data From the EFSS
Study
Location/
Method
N
Arithmetic
Mean TLA
Geometric
Mean TLA
Log
Std
Dev
Minimum
Value
Maximum
Value
R&R Target Activities
Window
Replacement
Demolition
Surface
Preparation
Exterior
Siding
New York
NIOSH, Cincinnati
EFSS
CONSAD
EFSS
Brooks AFB
New York
NIOSH, Cincinnati
U.S. Army Hygiene
Brooks AFB
CONSAD
New York
U.S. Army Hygiene
Massachusetts DPH
NIOSH, Cincinnati
U.S. Army Hygiene
U.S. Army Hygiene
Massachusetts DPH
NAHB
—
Interior/Dry
Exterior/Dry
Interior/Dry
Exterior/Wet
Interior/Wet
Exterior/Dry
2
8
8
2
20
6
5
15
12
15
4
1
8
3
7
12
15
17
1
144
6.81
14.0
271
153
5.50
29.8
27.8
All < LOD
8.07
12.5
1270
94.1
2000
17.4
All < LOD
88.3
254
All < LOD
144
5.44
7.48
166
107
2.82
24.7
9.16
All < LOD
2.69
11.7
1270
72.5
605
6.75
All < LOD
22.0
164
All < LOD
0.07
0.72
1.19
1.52
0.74
1.48
0.68
1.84
___
1.48
0.43
___
0.82
1.77
1.71
___
2.28
0.98
—
137
2.00
2.41
56.5
33.6
0.50
12.0
0.20
All < LOD
0.50
6.98
1270
23.0
70
0.70
All < LOD
< 2.00
40
All < LOD
151
16.0
44.4
485
947
10.0
63.0
120
All < LOD
40.0
17.3
1270
190
5350
63.0
All < LOD
280
820
All < LOD
Generic R&R Tasks
Cleanup
Component
Removal
Brooks AFB
EFSS - Plaster
EFSS - Wood
Brooks AFB
CONSAD
EFSS
Interior/Dry
Interior/Dry
3
4
2
8
5
2
All < LOD
27.701
103.44
9.06
7.54
344.01
All < LOD
24.5
102
6.28
7.47
344
___
0.55
0.21
0.91
0.15
0.045
All < LOD
14.6
88.5
< 2.00
5.96
333
All < LOD
53.3
118
30.0
8.78
355
Combination R&R Tasks
Cleanup/
Surface
Preparation
NIOSH, OU
NIOSH, OU
Interior/Dry
Interior/Wet
12
24
100.0
48.7
83.1
27.3
0.65
1.03
29.0
5.00
205
360
Note: Observations less than the detection limit were set to one-half the level of detection (LOD) for this summary.
8-95
-------
CO
E
TO
0)
160
140
120
100
80
S 60
o
ijj 40
E
8 20
O
0
New York
NIOSH, Cincinnati
R&R/EFSS
Figure 8E-1. Comparison of Geometric Mean Task-Length Average Personal Worker
Exposures (//g/m3) for the EFSS and Other Data Sources During Window
Replacement
200
150
CO
E
c 100
CD
(D
E
o
03
0
50
CONSAD
R&R/EFSS
Figure 8E-2. Comparison of Geometric Mean Task-Length Average Personal Worker
Exposures (//g/m3) for the EFSS and Other Data Sources During Demolition
8-96
-------
CO
CD
400
350
300
250
200
150
-344-
K:::K::S
o
o 100
E
S 50
O
Note: All sample results reported below the limit of detection (LOD) were set to one-half the LOD.
Figure 8E-3. Comparison of Geometric Mean Task-Length Average Personal Worker Lead
Levels (//g/m3) for the EFSS and Other Data Sources During Component
Removal
Cleanup Activities
Personal worker exposures related to cleanup activity from the Brooks AFB study are
compared to EFSS results in Figure 8E-4. Only data for the CED phase of the EFSS (not the
cleanup investigation) are included in Figure 8E-4. The Brooks AFB study results are much
lower than the EFSS, especially concerning cleanup following activities that disturb wood
surfaces. All three data points entering into Figure 8E-4 for the Brooks AFB study represent
results below the detection limit.
Cleanup/Surface Preparation Combinations
As discussed in Section 8E-1.1, the NIOSH-Ohio University study contains task-length
average personal worker exposures for tasks that were a combination (according to definitions
established in the R&R study) of surface preparation and cleanup. Figure 8E-5 presents personal
worker exposure estimates for dry scraping and broom sweeping in the NIOSH study compared
to cleanup results from the Brooks AFB study and the EFSS, and to interior dry surface
preparation results from the Brooks AFB, CONSAD, U.S. Army, and Massachusetts studies.
The results from the combined activity in the NIOSH study appear reasonable when compared to
the set of results for each individual activity.
8-97
-------
120
100
CO
E 80
ra
o>
d>
E
o
d>
O
60
.^ 40
20
Wood
102
HHBHHBH!
••••••••••••i
••••••••••••
Note: All sample results reported below the limit of detection (LOD) were set to one-half the LOD.
Figure 8E-4. Comparison of Geometric Mean Task-Length Average Personal Worker Lead
Levels (//g/m3) for the EFSS and Other Data Sources During Cleanup Activities
<
l-
00
E
^>
^
c.
(C
0>
5
o
'C
•4-1
0}
E
o
0)
O
1200
1000
800
600
400
200
0
1
2.69 11-7
-]^^^^£~——^^^^^^^=
270
fmmm
••••
••••
••••
••••
••••
••••
••••
••••
••••
••••
••••
••••
••••
•••I
•EEi
••••
••••
••••
••••
••••
••••
••••
••••
••••
••••
••••
••••
••••
••••
••••
••••
••••
••••
••••
••••
••••
••••
••••
••••
uu
*
605
72.5
Wood
g3 1 Plaster 1 Q2
OH 1^0 24.5 |U|
Brooks CONSAD New York* U.S. Army Mass. NIOSH
AFB Hygiene DPH OU
| Surface Preparation 1
*Sample Size = 1.
Note: All sample results reported below the limit of detection (LOD) were set to one-half the LOD.
Brooks
AFB
| Clean - up 1
R&R/
EFSS
R&R/
EFSS
Figure 8E-5.
Geometric Mean Task-Length Average Personal Worker Exposures (//g/m3)
During Cleanup and Dry Interior Surface Preparation for Various Studies
Compared to the Combination Activities of the NIOSH-OU Study
8-98
-------
8E-3.0 RESULTS OF THE ANALYSIS OF SURFACE PREPARATION DATA FROM
OTHER SOURCES
No data were collected in the EFSS on exposures related to surface preparation activities,
partly because data were expected to be available from other sources. Six such data sources were
located, and summary results from these sources were included in Table 8E-2 of Section 8E-2.0.
Surface preparation methods in each study were subdivided into interior and exterior surface
preparation and wet and dry methods. In general, the surface preparation activity consisted of
sanding, scraping, and cleaning activities. However, the extent to which each of these generic
activities was performed differed among the studies. This reflects an inherent difficulty in defining
surface preparation, which may range from very slight touch-up of surfaces in overall good
condition to heavy scraping and sanding of surfaces in poor condition. The wide variety of
activity that may be included under surface preparation is reflected in the variability in exposure
estimates from the different studies.
Figures 8E-6 through 8E-9 summarize personal worker exposure results from the six
studies, with one figure for each of the four combinations of interior and exterior surface
preparation and wet and dry methods. The variability in results between studies is most
pronounced in Figure 8E-6 for interior surface preparation using dry methods. A large number of
factors may be contributing to this variability based on documentation of the individual studies
and results obtained in the EFSS. The factors include the concentration of lead in the paint being
disturbed, the condition of the paint, the percent of total activity time spent sanding or scraping,
the size and air-flow characteristics of the activity room, the equipment used, personal work
habits, and small numbers of samples.
1400
^ 1200
M—
o
CO
E
^
c.
1000
800
600
o
E 400
E
200
1270*
Z70
fOOKS/
11.7
605
72.5
J.S. Army Hygiene
Note: All sample results reported below the limit of detection (LOD) were set to one-half the LOD.
"Sample Size = 1.
Figure 8E-6. Geometric Mean Task-Length Average Personal Worker Exposures (//g/m3) for
Other Data Sources During Dry Interior Surface Preparation
8-99
-------
CO
E
C.
m
-------
Figure 8E-9.
NIOSH, Cincinnati U.S. Army Hygiene
Note: All sample results reported below the limit of detection (LOD) were set to one-half the LOD.
Geometric Mean Task-Length Average Personal Worker Exposures (//g/m3)
for Other Data Sources During Wet Exterior Surface Preparation
8E-3.1 ANALYSIS TO ESTIMATE AVERAGE PERSONAL WORKER EXPOSURE LEVELS
ACROSS STUDIES FOR SURFACE PREPARATION
Meta-analytic statistical methods are used to summarize data across studies when only
summary statistics are available. Using this approach, an estimate of the mean personal worker
exposure from surface preparation and a measure of the variability about the mean are obtained
for each study and combined to form an overall mean exposure level, taking into account both
within-study and between-study components of variation. As the raw data were available for all
studies identified in the search for other data sources, within-study and between-study variability
were estimated directly from the raw data and incorporated into overall mean exposure level
estimates and associated confidence intervals. Other meta-analysis issues and concerns that were
addressed included:
1. The assumption that the studies are a random sample from a population of studies.
The validity of confidence intervals based on a random-effects analysis depends on
this assumption. The studies included were judged to be independent and
representative of exposures during different occurrences of surface preparation.
The lack of bias (e.g., publication bias) in selection of studies to be included.
identified studies with relevant data were included in the analysis.
All
8-101
-------
3. An assessment of the quality of individual studies. Each study was assessed from a
quality assurance standpoint (use of standard protocols, a certified laboratory,
experienced study personnel, etc.) and determined to meet minimum quality
assurance standards.
The statistical analysis to estimate overall mean exposure levels across studies was
conducted only on data from dry methods for two reasons: (1) these data were considered
representative of surface preparation as it is currently being conducted in an unregulated
environment, and (2) insufficient data existed for the wet methods.
Summary statistics for the data used in estimating overall mean worker exposure levels for
interior and exterior dry surface preparation are presented in Table 8E-3. The results of the U.S.
Army study of exterior dry surface preparation showed all twelve personal exposure observations
below the limit of detection (LOD). In order to more accurately represent the variability of
measurements less than the LOD in that study and the other studies, all measurements less than
the LOD in any study were simulated by generating a random deviate from a uniform distribution
on the interval (0, LOD). The assumption of a uniform distribution assigns equal probability to
any value below the LOD, and allows for a measure of variability based on that assumption. This
approach to handling non-detected results differs from the approach used to obtain summaries in
earlier tables and figures in this section, where the result was replaced by one-half the detection
limit.
Table 8E-3. Summary Statistics for Data Used in the Statistical Analysis of the Surface
Preparation Data (Dry Methods Only)
Location
Exterior
Interior
Study
Brooks AFB
New York
NIOSH, Cincinnati
U.S. Army Hygiene
Brooks AFB
CON SAD
Massachusetts DPH
New York
U.S. Army Hygiene
Geometric Mean
(//g/m3) of TLA
1.94*
24.7
9.16
0.89*
2.25*
11.7
605
1270
72.5
Log Standard
Deviation
1.77*
0.68
1.83
0.56*
1.61*
0.43
2.17
___
0.82
N
6
5
15
12
15
4
3
1
8
* For observations less than the limit of detection (LOD), random deviate(s) generated from a uniform
distribution on the interval (0, LOD) were used to calculate these statistics.
Model (OS-1) in Section C.7 of Appendix C was fitted to the log transformed lead
concentrations (|ig/m3) from individual personal worker exposure samples collected in each of the
studies listed in Table 8E-3. Model (OS-1) was fitted separately for interior and exterior dry
surface preparation. Further details on the statistical model are provided in Section C.7 of
Appendix C.
8-102
-------
Table 8E-4 displays model estimates of the geometric means, 95% confidence intervals on
the geometric mean, and variance components (standard deviations) of worker personal exposures
for interior and exterior dry surface preparation, resulting from fitting model (OS-1). While the
geometric mean exposure is much higher for interior dry surface preparation, the 95% confidence
intervals indicate that the two estimates do not differ statistically. The nature of the diverse
results across studies for interior dry surface preparation is noted by a large study-to-study
variability compared to worker-to-worker variability.
Table 8E-4. Model Estimates of the Geometric Mean, 95% Confidence Interval and
Standard Deviation of Variance Components of Worker Personal Exposures
(Task Length Average in //g/m3) for Interior and Exterior Dry Surface
Preparation
Location
Interior
Exterior
Number of
Studies
5
4
Total Number
of PEM
Samples
31
38
Geometric
Mean Estimates
from Model
58.2
4.33
95% Confidence
Interval of
Geometric Mean
(2.27, 1490)
(.407, 46.0)
Square Root of Estimated
Variance Components
Study to Study
(Ostudv)
2.49
1.41
Worker to Worker
(Ow.rk.rl
1.40
1.40
To investigate the nature of study-to-study differences, Figures 8E-10 and 8E-11 provide
a comparison of the individual study estimates and associated confidence intervals with the overall
estimate and its confidence interval for interior and exterior dry surface preparation, respectively.
Overall
U.S. Army Hygiene
NIOSH, Cincinnati
New York
Brooks AFB
H 1—i—i i i i 11
H 1 1—I I I I I
H 1 1—I I I I I
0.1
1.0 10.0
Geometric Mean (ug/m3)
100.0
Figure 8E-10. Comparison of Results from Statistical Analysis of Geometric Mean Task
Length Average Personal Worker Exposures (//g/m3) for Exterior Surface
Preparation: Overall Mean Estimate and Individual Study Mean Exposures
with 95% Confidence Bounds
8-103
-------
Overall
New York'
Mass DPH
U.S. Army Hygiene
CONSAD
Brooks AFB
-M-
i i 11
i i 11
-M-
i i 11
i i 11
i i 11
0.1 1.0 10.0 100.0 1000.0 10000.0 100000.01000000.0
Geometric Mean (ug/m3)
* Sample Size = 1.
Figure 8E-11. Comparison of Results from Statistical Analysis of Geometric Mean Task
Length Average Personal Worker Exposures (//g/m3) for Interior Surface
Preparation: Overall Mean Estimate and Individual Study Mean Exposures
with 95% Confidence Bounds
8E-3.2 SETTLED DUST SAMPLES IN THE MASSACHUSETTS DEPARTMENT OF HEALTH
STUDY ASSOCIATED WITH SURFACE PREPARATION
As described in Section 8E-1.0, the Massachusetts study also collected settled dust
samples from 1-ft2 ceramic tiles that were placed in rooms (activity areas) where surface
preparation was being conducted, at various times in the study. Five types of ceramic tile
samples, all using wipe collection methods, were collected in the Massachusetts study:
1. Tiles placed in the activity area and sampled before the activity started (pre-activity)
2. Tiles placed in the activity area before the activity started and sampled immediately
following the activity (during activity; sampled 0 hrspost)
3. Tiles placed in the activity area before the activity started and sampled at least 10
hours after the activity was finished (during activity; sampled 10 hrspost)
4. Tiles placed in the activity area after all surface preparation activity was finished and
after the area was HEP A vacuumed. These were sampled at least 10 hours later (after
activity and cleanup; sampled 10 hrspost)
8-104
-------
5. Tiles placed in rooms adjacent to the activity area after all surface preparation activity
was finished and after the activity area was HEPA vacuumed. These were sampled at
least 10 hours later (after activity and cleanup; adjacent area).
Samples of type 4 above (after activity and cleanup; sampled 10 hours post) were collected in
only three of the four study units. Samples of type 5 above (after activity and cleanup; adjacent
area) were collected in only one of the four study units. Like the settled dust samples in the
EFSS, the first three types of samples in the Massachusetts study provide an estimate of the total
amount of lead disturbed by the activity and made available in the occupant's environment before
cleanup. The last two sample types, however, represent lead exposures that may be expected to
settle and remain in an occupant's environment after cleanup.
Results for 13 of 51 settled dust samples collected in the Massachusetts study were
reported as right-censored data (such as "> 2000 jig/ft2"). These represented samples in which
the dilution process was terminated, since a right-censored result already indicated an
unacceptable amount of lead for the purposes of that study. These results are included as the
right-censored value in the summary statistics presented below. For this reason, inferences
concerning average differences between sample type 2 (during activity; sampled 0 hrs post) and 3
(during activity; sampled 10 hrs post), should not be made based on these summary statistics since
many of these sample types were right-censored. For example, the samples for one room resulted
in a measurement of > 120,000 |ig/ft2 for the type 2 sample and > 42,100 |ig/ft2 for the type 3
sample, obviously providing no information or misleading information on the difference between
the two samples.
The wipe dust sample results from this study are quantitatively comparable to vacuum
dust sample results obtained in the EFSS for other R&R activities using stainless steel dustfall
collectors. The difference in sampling methodology (wipe versus vacuum) as well as the fact
some data values were right-censored, as discussed below, should be taken into account in any
such comparison.
Summary statistics for the different types of settled dust samples for interior dry and wet
surface preparation are given in Table 8E-5 below. These results indicate extremely high lead
levels in dust that settles during and up to 10 hours following the activity. While dust that settles
after activity and cleanup has much lower lead levels (interior/wet methods), these levels tend to
remain over the EPA interim health-based standard of 100 |ig/ft2.
Field personnel in the Massachusetts study estimated that most activity areas (rooms)
were small, approximately 10 by 10 feet or 12 by 12 feet. All tiles were placed in the center of an
activity area when possible. (Some activity areas such as hallways had different placement.) The
tile samples may roughly be considered to represent samples collected, on average, 3 feet from the
source of activity.
8-105
-------
Table 8E-5. Summary Statistics for the Different Types of Settled Dust Samples Collected
in the Massachusetts DOH Study for Interior Dry and Wet Surface Preparation
Method
Interior/
Dry
Interior/
Wet
Type of
Sample
#1 Pre-activity
#2 During Activity;
Sampled 0 Mrs Post
#3 During Activity;
Sampled 10 Mrs Post
#1 Pre-activity
#2 During Activity;
Sampled 0 Mrs Post
#3 During Activity;
Sampled 10 Mrs Post
#4 After Activity and
Cleanup; Sampled 10
Mrs Post
#5 After Activity and
Cleanup; Adjacent Area
N
3
3
3
10
10
10
10
2
% Reported as
Right-Censored
0
33
33
0
60
50
0
0
% < LOD
0
0
0
10
0
0
0
0
Geometric Mean
(jug /ft2)
15.5
7980
5960
13.7
1270
1230
162
194
Minimum
(jug /ft2)
12.5
980
990
3.7
380
391
53.2
110
Maximum
(jug /ft2)
17.6
> 120,000
> 42,100
66.1
> 2,000
> 2,000
615
342
* Sample may have been contaminated as the surface preparation worker used this area to sharpen his scraper with a
grinder intermittently throughout the activity.
Note: 1. Samples reported as less than the limit of detection (LOD) were included in the analysis using one-half the LOD.
2. Samples reported as "greater than" a specified value (i.e. right-censored) were included in the analysis at the
right-censored value.
8-106
-------
9.0 OVERALL RESULTS
Chapter 8 in this report discussed lead disturbance and exposure results specific to
individual activities and tasks monitored in the EFSS or in other studies. These results were
presented to address the three sets of data analysis objectives presented in Section 6.1:
Objective A Characterize lead disturbance and potential lead exposures
Objective B Assess factors or measurements related to lead disturbance
Objective C Evaluate different media as indicators of lead exposure.
This chapter presents overall lead exposure and disturbance results across all activities. The
content of the chapter is related to the three sets of objectives described below.
Objective A
• key summary statistics for each activity characterizing lead exposure and disturbance,
and graphs and tables allowing for comparisons across activities
• a methodology for constructing 8-hour time-weighted average exposure for different
worker groups
• a methodology for adjusting estimates of the amount of lead deposited in settled dust to
a "standard unit" of activity
Objective B
• an assessment across activities of the observed effect of predictor variables (such as a
measure of the amount of lead in the paint) on lead disturbance and exposure
Objective C
• the extent of agreement between results collected from different media (air and settled
dust).
Observed limitations and data gaps related to addressing the above objectives with the collected
data are also presented in this chapter.
9.1 CHARACTERIZE LEAD DISTURBANCE AND POTENTIAL LEAD EXPOSURES
The type of environmental data collected in this study varied to some degree from one
R&R activity to another. All activities, however, included data on:
• personal air lead exposure monitored during the activity
9-1
-------
• lead in settled dust disturbed by the activity and collected from stainless steel dust
collectors (SSDCs) after completion of the activity.
These measures represent the primary raw data used in summarizing results. The personal air
monitoring results estimate airborne exposure for R&R workers performing the activity. The
dust-lead levels generated by the activity reflect the amount of lead disturbed by the activity and
distributed in the living or working environment of the occupants.
9.1.1 Personal Worker Exposures
Table 9-1 presents summary statistics characterizing the distribution of task-length average
(TLA) personal worker lead concentrations (|ig/m3), or average exposure over the duration of
performing the activity, for all monitored target activities and generic tasks. Results presented
include the 50th, 75th and 95th percentiles of the distribution. As a measure of uncertainty, a
two-sided 95% confidence interval about each estimated percentile was calculated for each
activity. These intervals permit an assessment of whether there is a statistical difference from a
specified level (e.g., the OSHA permissible exposure limit) or a statistical difference between the
different activities. The estimates and confidence intervals are based on the assumption that the
logarithms of the air lead concentrations within each activity are normally distributed. The
components of measured variation were taken into account in the confidence interval calculations
to deal with correlations among the measures. The method of calculation of the estimates and
associated confidence intervals varies, depending upon the number of components of variation
assumed. The methods are presented in Section C.6 of Appendix C. Results for dry surface
preparation, both interior and exterior, are based on the data collection from other sources as
discussed in Section 8E of Chapter 8. All other results are based on data collected in the EFSS.
Note that paint removal by abrasive sanding is included in Table 9-1 in two forms: sanding
with a power tool and sanding by hand. Paint removal was conducted as part of the Controlled,
Experimentally-Designed (CED) phase for use as a positive control measure against which to
compare personal exposure levels from other activities, since health effects associated with paint
removal have been previously documented (as discussed in Chapter 2).
Table 9-1 presents TLAs which are not directly comparable to the eight-hour time-
weighted averages (TWAs) on which OSHA exposure limits are based. Section 9.1.2 includes a
detailed discussion of the relationship between TLAs and TWAs.
Figure 9-1 illustrates a ranking of worker personal exposure across activities based on the
75th percentile of TLAs. A pair of activities with non-overlapping confidence intervals can be
judged to have significantly different 75th percentiles.
From Figure 9-1, it appears that the activities roughly form three groups: the four activities
with the highest 75th percentiles, the five with the lowest, and the middle four. The four highest
are generally distinguishable from the five lowest, with the middle four overlapping both groups.
The 75th percentiles for the four highest activities, along with demolition, are all significantly
above the OSHA PEL of 50 |ig/m3.
9-2
-------
Table 9-1. Estimated Percentiles and Confidence Intervals of Personal Worker Exposures
Measured by a Task Length Average for R&R Activities and Tasks
Estimated
50th
Percentile
TLA
(//g/m3)
95%
Confidence
Interval
for 50th
Percentile
Estimated
75th
Percentile
TLA
(//g/m3)
95%
Confidence
Interval
for 75th
Percentile
Estimated
95th
Percentile
TLA
(//g/m3)
95%
Confidence
Interval
for 95th
Percentile
R&R Target Activities
Carpet Removal
Window
Replacement
Demolition
HVAC
Dry Surface
Preparation111
(Interior)
Dry Surface
Preparation111
(Exterior)
7.54
7.48
108
49.6
58.2
4.33
(1.74, 32.6)
(1.13, 49.3)
(26.6435)
(11.4, 216)
(2.27, 1490)
(.408, 46)
24.9
17.5
185
56.0
398
16.5
(7.37, 208)
(5.13, 488)
(95.7, 5500)
(37.7, 3070)
(38.8, 88700)
(3.81, 1930)
138
59.2
403
66.6
6350
114
(34.0, 4680)
(16.7, 28500)
(186, 542000)
(53.0, 588000)
(473, 1.1x108)
(19.2, 1.45x106)
Generic R&R Tasks
Door
Modification
Sawing into
Wood
Sawing into
Plaster
Drilling into
Wood
Drilling into
Plaster
Sand - Hand
Sand - Power
590
546
110
15.1
6.76
254
571
(93.5, 3730)
(366, 813)
(0, 2.32x106
(4.57, 50.2)
(3.00, 15.3)
(23.7, 2720)
(42.9, 7600)
1360
705
232
36.3
9.70
513
1150
(410, 35200)
(518, 1300)
(11.9,
2.40x1013)
(13.8, 214)
(5.71, 40.3)
(150, 63800)
(286, 170000)
4480
1020
681
127
16.3
1410
3170
(1300, 1.89x106)
(726, 2890)
(149, 2.15x1027)
(42.9, 2490)
(9.47, 226)
(447, 2. 03x1 07)
(940, 5.03x107)
(1) Summary of data collected from other sources. Surface preparation consisted of a wide variety
of activities including wet and dry scraping, feathering of edges, and wet and dry sanding to
prepare a surface for repainting.
9-3
-------
Ranking of R&R Activities According to Worker Exposure Results
Ranking
Activity or Task
Lowest 4-
-* Highest
Door Modification
Sanding (Pawarf
Sowing into Wood
Sanding (Hand}*
Dry Suiface Preparation (Interior)
Sawing into Plaster
Demolition
HVAC Removal
Drilling into Wood
Carpet Removal
Window Replacement
Dry Surface Preparation (Exterior)
Drilling into Plaster
10 100 1,000 10,000 100,000 1,000,000
75th Percentile TLA (ug/rrf)
• Represents 75th percentile of PEM distribution
I- -I Represents the 95* confidence intervol obout the 75th percentile
Activities with non-overlopping confidence intervals hove significantly different ranks
• Actual upper confidence bound is 2.40J5E 13
it These activities represent positive controls
Figure 9-1. Ranking of R&R Activities and Tasks Based on the Estimated 75th Percentile of Worker Exposures Measured by a
Task Length Average
-------
Table 9-2. The Percent of Workers Whose TLA Personal Air Lead Concentration Would
Exceed the OSHA PEL (50 //g/m3), as Estimated from the Observed
Distributions of TLAs in this Study
Number of
Workers
Monitored
(N)
Estimated Percent of
Workers Who Will Have
a TLA Above the OSHA
PEL (50 //g/m3)
95% Confidence Interval
for the Estimated Percent
of Workers Above the
OSHA PEL
R&R Target Activities
Carpet Removal
Window Replacement
Demolition
HVAC Removal
Dry Surface Prep. (Interior)111
Dry Surface Prep. (Exterior)111
14
8
20
4
31
38
14%
7%
83%
48%
52%
11%
(3% - 43%)
(0% - 50%)
(40% - 99%)
(10% - 90%)
(23% - 80%)
(0% - 49%)
Generic R&R Tasks
Door Modification
Sawing into Wood
Sawing into Plaster
Drilling into Wood
Drilling into Plaster
Sanding (Hand)
Sanding (Power)
6
6
2
7
6
6
3
98%
99%
76%
18%
0%
94%
99%
(58% - 100%)
(99% - 100%)
(15% - 99%)
(4% - 51%)
(0% - 21%)
(41% - 100%)
(48% - 100%)
Based on data from other sources. Surface preparation consisted of a wide variety of activities including wet and dry
scraping, feathering of edges, and wet and dry sanding to prepare a surface for repainting.
Table 9-2 presents estimates of the percentage of workers who would exceed the OSHA
PEL of 50 |ig/m3 if the activity were repeatedly executed under similar conditions of this study,
and 95% confidence intervals for this estimate, for each R&R activity and task.
9.1.2 Conversion of Task Length Averages (TLAs) into 8-Hour Time Weighted Averages
(TWAs)
The personal worker exposure data summarized above was collected as task length
averages (TLAs) for specific R&R activities in accordance with the focus of the EFSS on
activities rather than worker groups. The advantage of reporting TLAs for a variety of R&R
9-5
-------
activities is that TLAs can be used to estimate 8-hour time weighted averages (TWAs) for any of
the numerous R&R worker groups, once a profile of the activities that a worker group performs
is determined. Therefore, the EFSS has provided the components for constructing worker
exposure estimates based on profiles for a wide variety of worker groups. On the other hand, the
amount of activity being performed in a normal 8-hour workday must be taken into account when
comparing lead exposures across R&R activities or tasks, or when comparing exposures of
individual R&R activities or tasks to OSHA standards which are based on an 8-hour TWA. For
the purpose of evaluating and ranking activities, the TLA is equivalent to an 8-hour TWA only if:
1. the task is expected to be performed for eight hours in a day
2. the rate of exposure (|ig/m3) for eight hours of activity is equal to the rate measured
in a shorter time period.
To further illustrate, the 75th percentile of TLA exposures for drilling into wood is estimated to
be 36 |ig/m3 (Table 9-1). However, if this activity were only conducted for 1/4 hour in an eight-
hour work shift, and the exposure in the remaining 7-3/4 hours was zero, then a corresponding 8-
hour TWA calculated according to OSHA specifications would be reduced by a factor of 32,
resulting in an 8-hour TWA of approximately 1 |ig/m3.
Figure 9-2 is presented to help the reader assess the conversion of the TLA to an 8-hour
TWA for each activity. This figure presents the 75th percentile of an 8-hour TWA, expressed as
a function of the 75th percentile of the TLAs and the length of the activity or task. This
relationship is given by the following formula:
75th Percentile 8-hour TWA = (75th Percentile TLA) * (# Hours Conducted)
(Note that any summary statistic of the estimated TLAs, not just the 75th percentile, could have
been used for this illustration.) This formula, as mentioned above, assumes a constant rate of
exposure during conduct of the activity, regardless of the length of the activity, as well as zero
exposure when the activity is not being conducted. This is consistent with how 8-hour TWAs are
currently calculated by OSHA. Each hypothetical 8-hour TWA in Figure 9-2 depends only on the
amount of the specified activity assumed to be conducted in an 8-hour workday. Figure 9-2
indicates that just one hour of sanding or sawing into wood (and seven hours of no exposure)
would result in the 75th percentile of 8-hour TWAs being greater than the OSHA PEL (50 |ig/m3)
for those activities. On the other hand, eight hours of drilling, carpet removal, exterior surface
preparation, or window replacement would still not result in the 75th percentile of 8-hour TWAs
being greater than the OSHA PEL.
9-6
-------
Hypothetical 8 -Hour TWA for the 75th Percentile of Observed TLA
on the Amount of Activity Conducted in an 8 -Hour
Workday that was Otherwise Devoid of Exposure
Activity or Task 0.1
1.0
10.0 100.0 1000.0 10000.0
Sontong ftwwotj
Sowing into Wbod
Sanding (Hand)
l/ty Suffita Avyuiunoik fffttonitfj
Sowing into J%utar
HVAC Bomoval
Drilling into Vfood
Gtiiji&t /bxnovui
Diy Sutytu*& Av^MUwukn (Isxtozwzj
X/nuuitt uiw PlQ&tttf
I I I 1 1 1 1 1 1
0.1
1.0 10.0 100.0 1000.0 10000.0
Hypothetical 8-Hour TWA (ug/m3)
Time activities represent positive controls
Legend: ' ' ' ' ' ' ' ' '
15 mln 1 hr 2 hr 3 hr 4 hr 6 hr 6 hr 7 hr 8 hr
Amount of Activity
Figure 9-2. Hypothetical 8-Hour Time-Weighted Average Worker Exposures for Each Activity/Task and for Various Activity
Durations
-------
Figure 9-2 demonstrates the effect of the amount of activity conducted on an 8-hour TWA
when a worker performs only one type of activity in a day that causes lead exposure. In the more
common occurrence of different activities being conducted in a single work day, an estimated 8-
hour TWA can be constructed by averaging the TLAs for each activity, where each TLA is
weighted by the duration of activity. Let TLA; represent the TLA for the ith activity conducted in
an 8-hour work day, and Amt; the time in hours for which the activity was conducted. Then the
8-hour TWA is estimated by:
TLA * Amt
For example, suppose a certain R&R worker group typically conducts the following activities for
the specified period of time:
Activity
Sawing into Wood
Drilling into Wood
Window Replacement
No exposure activities
Time Conducted
1/2 hour
1/4 hour
6 hours
1-1/4 hour
Then by Table 9-1, the 75th percentile of the 8-hour TWA for this worker group would be
estimated by:
(705*.5) + (36.3*.25) + (17.5*6) + (0*1.25) _ _a - , , 3
9.1.3 Limitations of the Data and Additional Information Required for an
Exposure Assessment for Workers
Section 4.5 of Chapter 4 presents a general discussion of the limitations of the data
collected in the EFSS from a standpoint of representativeness and completeness. In order to
conduct a more complete exposure assessment of different worker groups than what could be
performed in the EFSS, the following additional information is required:
1. To what extent are the activities in question performed during R&R operations?
This addresses the TLA versus TWA issue discussed above. The percentage and
length of time R&R workers perform the studied activities must be characterized. In
addition, the percentage and types of R&R workers that perform each of these
activities must also be determined.
9-8
-------
2. How frequently are R&R operations conducted in lead-contaminated
environments? The HUD National Survey of Lead-Based Paint in Housing (U.S.
EPA, 1995) estimated that 64 million homes built before 1980 contain lead-based
paint. The issue is how frequently do different R&R worker groups work in these
residences or other lead-contaminated buildings.
Information was collected in the WCBS data collection effort of the R&R study to address
these questions and is presented in the WCBS technical report.
9.1.4 Summary Statistics for Lead Disturbance and Potential Exposure to Occupants
The potential lead exposure to occupants resulting from a specific R&R activity or task
was characterized by estimating the average amount of lead disturbed (or deposited) by the
activity along a 6-foot line (lead gradient) emanating from the activity. For each activity, a model
of dust-lead level as a function of distance from activity was fitted to lead loading data from
SSDCs (Model (C-3) of Appendix C). The average amount of lead exposed was estimated by
determining the area under the fitted curve from 0 to 6 feet from the activity. This measure is
equivalent to estimating the average amount of lead exposed in a 6-foot by 1-foot gradient region
extending from the activity. The uncertainty associated with this estimate is assessed by
calculating an approximate 95% confidence interval. Further details on the gradient region are
presented in Chapter 6 and Section C.5 of Appendix C. In contrast to the characterization of
personal worker exposures in Table 9-1, the 75th percentile and 95th percentile of the distribution
of estimates of lead disturbed were not estimated because of the minimal number of data points
and the amount of variability in the distribution.
For the majority of R&R activities, settled dust samples were collected from SSDCs
placed directly adjacent to the activity and at specified distances away from the activity. The
samples taken from SSDCs located directly adjacent to the activity captured the fallout of both
large debris and dust generated by an activity. Samples that were collected from SSDCs located
further away from the activity were expected to capture mostly dust and smaller airborne
particles.
For the majority of activities, the estimated 6-foot by 1-foot gradient lead loading
represents the average total amount of lead that is expected to settle in a 6-foot by 1-foot region
extending lengthwise away from the component being disturbed. However, this estimate has a
slightly different interpretation for carpet removal and demolition, as explained below.
The layout of settled dust sampling locations in the carpet removal activity only included
samples located adjacent to the surface being disturbed. This made it impossible to determine the
functional relationship between settled dust and distance. In all other activities studied, the
average lead loading decreased with increasing distance away from the activity. The most
conservative estimate of the functional relationship would assume that the amount of lead would
stay constant across the 6-foot by 1-foot region. Therefore, the (conservative) estimated average
gradient lead loading for carpet removal represents the amount of lead found in the 6-foot by
1-foot region assuming that the distribution of lead across that region is uniform.
9-9
-------
The layout of settled dust sampling locations in the demolition activity excluded samples
adjacent to the activity (due to the large amount of debris), but did include several samples at
varying distances away from the activity. Therefore, we are able to determine the functional
relationship between settled dust and distance for the demolition activity, but this relationship
does not take into account the amount of lead that settles at a location directly adjacent to the
activity. Since the settled dust samples associated with the demolition were all located at a
distance from the activity space, the estimated 6-foot by 1-foot gradient lead loading in the
demolition activity is interpreted as being the amount of lead found in the 6-foot by 1-foot region
that was airborne in dust and smaller particles, rather than the total amount of lead disturbed.
9.1.4.1 Adjustment to a Standard Unit of Activity
The R&R activities monitored in the EFSS consisted of a number of different target
activities as well as several generic activities. In all cases, the estimate of the amount of lead
disturbed in a 6-foot by 1-foot gradient region is highly dependent on the amount of activity
conducted. In the case of the generic activities (sawing and drilling), the amount of activity was
not intended to reflect normal levels of sawing or drilling that are performed in the R&R industry.
Rather, they were intended to provide sufficient activity to allow for estimating a TLA worker
exposure as well as the lead disturbed by the specified amount of activity. Therefore, the amount
of lead in the gradient region was adjusted to a "standard unit of activity". The standard unit of
activity selected for each studied activity does not represent an "average" amount of activity.
Rather, the estimated amount of lead was scaled to a unit of activity that seems reasonable for a
"real world" R&R activity. Moreover, the scaling seeks to produce estimates of disturbed lead
amounts which may be readily re-scaled by the reader. For example, in the CED phase, sawing
entailed cutting 75 linear feet of surface. The chosen standard unit of activity, on the other hand,
is 1 linear foot, which may be more easily rescaled.
The choice of the standard unit of activity for all activities except window replacement and
carpet removal is discussed in Section 8C-2.3.3 of Chapter 8 and presented in Table 9-3. The
standard unit of activity for window replacement is one window. The standard unit of activity for
carpet removal is 100 ft2. Table 9-3 presents estimates of the amount of lead disturbed in the
gradient region for the standard unit of activity for each R&R activity except abrasive sanding and
door modification, where no standard unit of activity could be determined. Figure 9-3 depicts the
lead loading gradient for activities as conducted in the EFSS. Figure 9-4 depicts the lead loading
gradient for the standard unit of activity.
Note that the gradient lead loading for window replacement, demolition, and HVAC
removal remain unchanged between Figures 9-3 and 9-4. This is due to the fact that these
activities, as performed in the EFSS, already represented the "standard unit of activity."
Otherwise, the chosen amount of activity has a considerable effect on estimates of the amount of
lead disturbed in the gradient region.
9-10
-------
100000
** T ^*-*^^CJf
Saw/Wbod r~3*^<
DrWWoodI J!r
Saw/Waster
Window
Distance^
DemoMton
HVAC Removal , ,
Carpet Removal °
Figure 9-3. Estimated Distribution of Dust Lead in a 6' x 1' Gradient for Various Target
Activities and Tasks, Based on Total Amount of Activity Performed in the EFSS
Window
Distance (ft)
DrtVPIaster T
Carpet Removal 6
Figure 9-4. Estimated Distribution of Dust Lead in a 6' x 1' Gradient for Various Target
Activities and Tasks, Based on Performing the Standard Unit of Activity
9-11
-------
Table 9-3. Estimated Lead Amounts (//g) Distributed Within a 6' x 1' Gradient
Region, with Standard Errors and 95% Confidence Intervals, in the
EFSS
Standard
Unit of
Activity
6' x 1' Gradient
Mean Lead Amount
(Std Error)
[95% Cl]
[0-1] ft SSDC
Mean Lead Amount
(Std Error)
[95% Cl]
[5-6] ft SSDC
Mean Lead Amount
(Std Error)
[95% Cl]
R&R Target Activities
Carpet
Removal*
Window
Replacement
Demolition*
HVAC Removal
100 ft2 of
Carpet
1 Window
1 Room
1 Room
102*
(37.2)
[46.3, 223]
46300
(20900)
[17100, 125000]
19500*
(14300)
[3610, 106000]
7710
(14100)
[0, 9.09x1013]
16.9
(6.2)
[7.7, 37.1]
26000
(13000)
[8730, 75000]
___
2690
(5700)
[0, 1.24X1015]
___
482
(217)
[179, 1300]
1530
(948)
[366, 6390]
414
(515)
[0, 3.1 x109]
Generic R&R Tasks
Drilling into
Plaster
Drilling into
Wood
Sawing into
Plaster
Sawing into
Wood
10 Holes
10 Holes
1 foot
1 foot
207
(191)
[11, 3900]
2590
(1400)
[690, 9740]
1970
(2560)
[0, 3.04X1010]
5990
(2230)
[2300, 15600]
168
(163)
[8, 3680]
2000
(1140)
[497, 8030]
1210
(1780)
[0, 1.6x10"]
2980
(1300)
[967, 9150]
0.04
(0.05)
[0.00, 2.81]
1.27
(0.49)
[0.50, 3.26]
10.6
(6.6)
[0, 29600]
105
(46)
[34, 324]
* 6' x 1' gradient lead loading calculated differently as explained in text.
Figure 9-5 illustrates, for each generic R&R task, the estimated increase in lead disturbed
as a function of the amount of activity performed. The amount of activity performed is presented
as a multiplicative increase in the standard unit of activity. For example, the standard unit of
activity for drilling is specified as 10 holes. The x-axis in Figure 9-5 ranges from a multiplicative
increase of 1 (i.e., the standard unit amount of activity) through a multiplicative increase of 20.
For drilling, this range corresponds to 10 to 200 holes. Figure 9-5 makes an assumption that the
amount of lead disturbed in the gradient region is related in a multiplicative fashion to the amount
of activity performed.
9-12
-------
cp
CO
Estimated 6' x 1' Gradient Lead Loading
Based on the Amount of Activity Performed
Activity/Task • Standard Unit
.2
150,000
120,000
a
p 90,000
x
82.
60,000
30,000
O)
Sawing into Wood • 1 toot
DrlllnQ Into Wood • 10 hate
Sawing into Piaster • 1 foot
Drtling into Plaster • 10 hate
unit 5 10 15 20
Multiplicative Increase in Standard Unit of Activity
Figure 9-5. Estimated 6' x 1' Gradient Lead Amounts as a Function of Increases in the Standard Unit of Activity for Each
Generic R&R Task
-------
9.1.5 Estimated Effect of Clean-up on the Total Lead Disturbed
Section 8D of Chapter 8 presents the design and results of an EFSS investigation on how
typical post-R&R activity cleanup procedures (dry broom sweeping, vacuuming with Shop-
Vac®) contribute to reducing lead loadings in settled dust to which occupants can be exposed.
This investigation was conducted to provide additional information on procedures that can
potentially affect lead exposures to occupants during R&R activities.
Two R&R activities (drilling, abrasive sanding) were considered in the cleanup
investigation, each conducted on wood door surfaces covered with lead-based paint. Settled dust
samples were collected at pre- and post-cleanup within each incidence of activity and cleanup.
The results indicated that both cleanup methods performed well in reducing lead loadings in the
dirtiest areas for the two activities being considered. On average, for each activity and cleanup
method, the post-cleanup lead loading was no more than one percent of the pre-cleanup lead
loading among dust deposited immediately adjacent to the activity (i.e., within one foot), where
approximately 65% of the total lead loading within six feet of the activity existed. However, the
extent of decline in lead loadings was less pronounced when the distance from the activity
approached six feet, where less dust is generated, or where some cross contamination from
cleanup closer to the activity may be taking place. Some evidence was observed that vacuum
cleanup methods may be more effective than broom sweeping in reducing lead levels at the 6-foot
distance, although any difference between pre- and post-cleanup results at the six-foot distance
was not statistically significant.
In addition to evaluating the efficacy of cleanup methods based on the percentage
reduction from pre-cleanup levels, it is also useful to examine whether levels remaining after
cleanup satisfy current clearance testing criteria. In summarizing lead loadings associated with
post-cleanup dust samples, the cleanup investigation determined that in most instances, post-
cleanup levels remained above EPA interim health-based standards of 100 |ig/ft2. This result was
evident throughout the six-foot region extending from the activity.
The results of the cleanup investigation showed that cleanup methods available to most
workers can substantially reduce the lead exposures that result from R&R activities. However,
the effort put into conducting these methods can influence the extent to which any remaining lead
levels are below pre-determined clearance standards.
9.1.6 Limitations of the Data on Lead Disturbance and Additional Information Required for
an Exposure Assessment for Occupants
Section 4.5 of Chapter 4 presents a general discussion of the limitations of the data
collected in the EFSS from a standpoint of representativeness and completeness. In order to
conduct a more complete exposure assessment for occupants, the following additional information
is required:
9-14
-------
1. To what extent are the activities in question performed during R&R operations?
Information collected in the WCBS to address this question is discussed in the
WCBS technical report and also presented in Table 8 of the R&R Summary Report.
2. What work practices and clean-up activities are regularly employed at R&R
operations? Work practices, such as installing polyurethane sheeting to collect the
dust generated, and the extent of clean-up will determine the amount of lead
disturbed that is actually left in the occupants environment after completion of "real-
world" R&R jobs. Information collected in the WCBS to address this question is
discussed in the WCBS technical report and also presented in Table 8 of the R&R
Summary Report.
3. To what degree does environmental lead contamination in a building translate
into an adverse health effect for occupants? An important question is how
environmental lead levels relate to human lead exposures. In the case of exposures
caused by an R&R activity, the additional consideration of the nature of a "transient"
exposure versus a long-term exposure will also need to be taken into account in
assessing occupant health effects.
9.2 ASSESSING FACTORS OR MEASUREMENTS RELATED TO LEAD DISTURBANCE
Two primary factors may be expected to affect lead exposure or lead disturbance resulting
from an R&R activity:
1. The amount of activity performed
2. The amount of lead contamination in the activity area.
Section 9.1 discussed ways to adjust the measurements of lead exposures and disturbances
for the amount of R&R activity that was performed. The adjustment is based on a simple
multiplicative model that implies, for example, performing twice as much activity will result in
twice as much lead disturbance. This assumed model was offered as a way of interpreting the
results pertaining to lead exposure and lead disturbance, and was not verified using data from the
study.
In contrast, the relationship between the amount of lead contamination in the activity area
and associated lead exposure and disturbance from the activity was examined in detail using data
from the study. Several statistical models were applied to data from the EFSS to investigate the
relationship between measures of lead exposure and disturbance, and different measures of lead
contamination in the activity area. The measures of lead contamination in the activity area
included pre-activity measures of lead in the following sample media: paint chip, ambient air, and
settled dust on floors, carpets, window sills and furnace ductwork. Not all sample media were
collected for each activity. In general, no strong statistical relationships with these lead
contamination measures were revealed by this investigation as indicated by Tables 8A-4, 8B-4,
and 8C-16 in Chapter 8. We attribute this lack of statistically significant results to two factors:
9-15
-------
1. It is difficult to characterize the amount of lead contained in the component(s) being
disturbed by an activity. For example, many different lead-containing components
may be disturbed during carpet removal including; lead dust from the carpet, the
floor underneath the carpet, walls, window sills and wells, and lead paint from the
baseboard.
2. The combination of small sample size and large variability in the measures of lead
disturbance resulted in low power to investigate this relationship.
The lack of a significant statistical relationship imposes a limitation on the interpretation of both
the worker and occupant exposure measures, in that it is not yet possible to adjust these exposure
measures for the amount of lead contained in the components being disturbed.
9.3 EVALUATING DIFFERENT MEDIA AS INDICATORS OF EXPOSURE
A strong linear relationship appears to exist between measures of airborne lead and
measures of lead in settled dust generated by the different R&R activities (as described in Chapter
8). In the carpet removal phase and the window replacement phase, positive correlation
coefficients of 0.98 and 0.94 respectively were found between settled dust measurements (1 hr.
SSDC vacuum samples located adjacent to the activity) and personal exposure measurements. In
the CED phase of the study, a correlation coefficient of 0.83 was found between personal
exposure measurements and estimated settled dust lead loadings located at six feet away from the
activity.
Despite the strong linear relationships that were found between lead in settled dust
measurements and personal exposure measurements, each individual R&R activity must be
studied in greater detail in order to evaluate the mechanism by which it can result in lead exposure
for either workers or occupants. For example, in the window replacement phase, window
replacement resulted in low levels of worker exposure and high levels of potential occupant
exposure. This phenomenon may be related to two different factors:
1. The percentage of time that the workers actually spend removing the windows (i.e.,
disturbing the lead contaminated components) in this activity is very small, while the
amount of lead disturbed by this activity is quite high.
2. The windows remained open for periods of time when the worker was not in the
room. During these periods of time, pre-activity or activity generated lead
contaminated dust may have been blown around the rooms under investigation,
resulting in the high levels of lead in settled dust.
Caution must therefore be exercised before using airborne lead to predict occupant exposures or
lead in settled dust to predict worker exposures.
9-16
-------
10.0 CONCLUSIONS
The conclusions presented in this section are based solely on the results of the EFSS
component of the R&R study. The R&R Summary Report integrates the results of the WCBS
with those of the EFSS to allow additional conclusions to be drawn concerning worker and
occupant exposures. Conclusions for the EFSS are related to three topical categories: exposure
assessment, recruitment, and sampling methodology. The major conclusions of the EFSS are
stated below.
EXPOSURE ASSESSMENT
1. The amount of lead distributed into the environment by all monitored target
activities and generic R&R tasks, with the exception of drilling into plaster and
exterior surface preparation, was substantial enough to indicate at least
occasional worker exposure above the OSHA PEL.
2. The EFSS provides exposure estimates for each studied target activity and
generic R&R task. These estimates can be combined with worker profile
information to characterize worker exposure associated with various worker
groups.
3. The EFSS provides estimates of the amount of lead disturbed and distributed in
the living or working environment of the occupants for each studied target
activity and task. These estimates can be combined with information on the types
and durations of activities conducted, the types of work practices and clean-up
activities and the link between environmental measurements and an adverse
health effect to characterize occupant exposures associated with R&R activities
and worker groups.
RECRUITMENT
1. Recruitment of unregulated contractors for participation in a study that monitors
their "real-world" work is very problematic and must be carefully considered in
any future study design.
SAMPLING METHODOLOGY
1. The stainless steel dustfall collector (SSDC) methodology represents a viable
alternative for measuring exposures due to a short-term activity with noted
advantages over the previous method of taking pre- and post-activity floor
samples.
2. Evidence suggests that lead disturbed by a specific activity and distributed in the
air continues to settle to the ground for a period longer than one hour.
3. It is possible that the order in which the samples are collected in side-by-side
settled dust samples affects the lead loading results.
10-1
-------
11.0 REFERENCES
Amatai, Y., Graef, J. W., et al. "Hazards of 'Deleading' Homes of Children with Lead
Poisoning," American Journal of Diseases in Children, July 1987, 141:758-760.
Amitai, Y., Brown, M. J., Graef, J. W., Cosgrove, E. "Residential Deleading: Effects on the
Blood Lead Levels of Lead-Poisoned Children", 1991, Pediatrics. 88(5):893-897.
Battelle Memorial Institute, "Comprehensive Abatement Performance Pilot Study (Volume
I)-Draft Final Report to USEPA," September 1994.
Battelle Memorial Institute and Kennedy-Kreiger Institute, "Quality Assurance Project Plan for
the Kennedy-Krieger Institute Lead-Based Paint Abatement and Repair and Maintenance
Study-Draft Report to USEPA," June 1992.
Bishop, Y. M. M., Feinberg, S. E., and Holland, P. W., Discrete Multivariate Analysis: Theory
and Practice. Cambridge, MA: The MIT Press, 1975, p. 492.
Centers for Disease Control and Prevention, "Preventing Lead Poisoning in Young Children,"
October, 1991.
Charney, E., Kessler, B., Farfel, M., Jackson, D. "Childhood Lead Poisoning: a Controlled Trial
of the Effect of Dust-Control Measures on Blood Lead Levels," New England Journal of
Medicine, 1983, 309(18): 1089-1093.
Childhood Lead Poisoning Prevention Program, Department of Public Health and Midwest
Research Institute, "Quality Assurance Project Plan for the Industrial Hygiene Field Study
to Assess Worker Exposure to Airborne Lead Dust During Surface Preparation
Activities," November 1992.
Chisolm, J. J., Mellits, E. D., Quaskey, S. A. "The Relationship Between the Level of Lead
Absorption in Children and the Age, Type, and Condition of Housing," Environmental
Research, 1985,38:31-45.
Clark, C. S., Bornschein, R. L., et al. "Condition and Type of Housing as an Indicator of
Potential Environmental Lead Exposure and Pediatric Blood Lead Levels," Environmental
Research, 1985, 38:46-53.
Colorado Department of Health, U. S. Department of Health and Human Service, "Leadville
Metals Exposure Study," April 1990.
David C. Cox & Associates, "Applicability of RCRA Disposal Requirements to Lead-Based
Paint Abatement Wastes-Draft Report to USEPA," October 1992.
11-1
-------
Curran, J. P., Nunez, J. R., "Lead Poisoning During Home Renovation," New York State Journal
of Medicine, December 1989, 89(12):679-680.
Farfel, M. R., Chisolm, J. J. "Health and Environmental Outcomes of Traditional and Modified
Practices for Abatement of Residential Lead-Based Paint," American Journal of Public
Health, October 1990, 80(10): 1240-1245.
Farfel, M. R., Chisolm, J. J. "An Evaluation of Experimental Practices for Abatement of
Residential Lead-Based Paint: Report on a Pilot Project," Environmental Research, 1991,
55:199-212.
Grondona, C. "Lead Revisited: A Case Study on Lead Exposed Painters," Journal of the
American Association of Occupational Health and Nutrition, January 1993, 41(l):33-38.
Landrigan, P. J., Baker, E. L., et al. "Exposure to Lead from the Mystic River Bridge: the
Dilemma of Deleading," New England Journal of Medicine, 1982, 306:673-676.
Marino, P. E., Franzblau, A., et al. "Acute Lead Poisoning in Construction Workers: the Failure
of Current Protective Standards," Archives of Environmental Health, May-June 1989,
44(3): 140-145.
Marino, P. E., Landrigan, P. J., et al. "A Case Report of Lead Paint Poisoning During Renovation
of a Victorian Farmhouse," American Journal of Public Health, October 1990,
80(10):1183-1185.
National Association of Home Builders, "A Private Sector Strategy for Reducing Lead Exposure
in Children-Report by the NAHB Lead Exposure Task Force," April 1992.
National Association of Home Builders, "What Remodelers Need to Know and Do About
Lead-An Interim Guide," December 1992.
National Association of the Remodeling Industry, "NARI 1990 Membership Survey."
National Institute for Occupational Safety and Health, "Health Hazard Evaluation Report-HUD
Lead-Based Paint Abatement Demonstration Project," HETA 90-070-2181, February
1992.
National Institute for Occupational Safety and Health, "Health Hazard Evaluation Report —
Ohio University, Athens, Ohio," HETA 92-095-2317, May 1993.
National Institute for Occupational Safety and Health, "Interim Report, Cincinnati, Ohio," HETA
93-0818.
Rabinowitz, M., Needleman, H. "Demographic, medical, and environmental factors related to
cord blood lead." Biol. Trace Element Res., 1984, 6:57-67.
11-2
-------
Rabinowitz, M., Leviton, A., Bellinger, D. "Home Refmishing, Lead Paint, and Infant Blood
Lead Levels," American Journal of Public Health, April 1985, 75(4):403-404.
Rabinowitz, M., Leviton, A., Needleman, H., Bellinger, D., Waternaux, C. "Environmental
Correlates of Infant Blood Lead Levels in Boston", 1985, Environmental Research 38:96-
107.
Reagan, P. L., "Comments of the Minnesota Lead Coalition: Analysis of the Lead-Based Paint
Guidelines", report to U.S. Department of Housing and Urban Development, April 26,
1991.
Rempel, D. "The Lead-Exposed Worker", Journal of the American Medical Association,
262:532-534.
Remodeling Magazine, "1992 State of the Remodeling Industry," Hanley-Wood Inc., 1992.
Roberts, J. W., Camann, D. E., Spittle, T. M. "Reducing Lead Exposure from Remodeling and
Soil Track-In in Older Homes", Air and Waste Management Association, Presented at
84th Annual Meeting and Exhibition, Vancouver, British Columbia, Jun 16-21, 1991.
Rundle, S. A., Duggan, M. J., "Lead Pollution from the External Redecoration of Old Buildings,"
Science of the Total Environment, December 1986, 57:181-190.
Schneitzer, L., Osborn, H. H. et al. "Lead Poisoning in Adults from Renovation of an Older
Home," Annals of Emergency Medicine, April 1990, 19(4):415-420.
Shannon, M.W., and Graef, J.W. "Lead Intoxication in Infancy," Pediatrics, January 1992,
89(1):87-90.
Society for Occupational and Environmental Health, "Protective Work Practices for Lead-Based
Paint Abatement-Draft for Final Review."
State of Minnesota, "Minnesota Lead Law Statues and Regulations, 1992".
University of Cincinnati, "Cincinnati Soil-Lead Abatement Demonstration Project-Draft Final
Report to USEPA," April 1992.
U.S. Census Bureau, 1987. Census of Construction Industries, Expenditures on Residential
Upkeep and Improvement, C-50.
U.S. Department of Health and Human Services, "Strategic Plan for the Elimination of
Childhood Lead Poisoning," February, 1991.
11-3
-------
U.S. Department of Housing and Urban Development, "Guidelines for the Evaluation and
Control of Lead-Based Paint Hazards in Housing." Office of Lead-Based Paint
Abatement and Poisoning Prevention. HUD-1539-LBP, July 1995.
U.S. Department of Housing and Urban Development. (1990) "Lead Based Paint: Interim
Guidelines for the Hazard Identification and Abatement in Public and Indian Housing."
Office of Public and Indian Housing. September 1990.
U.S. Environmental Protection Agency, 1995. "Report on the National Survey of Lead-Based
Paint in Housing: Base Report." EPA Report No. 747-R95-003. Office of Pollution
Prevention and Toxics. April, 1995.
U.S. Environmental Protection Agency, "Air Quality Criteria for Lead Volume I - IV," 1986,
EPA Report No. EPA-600/8-83-028CF. Environmental Criteria and Assessment Office,
Office of Research and Development, Research Triangle Park, NC.
Weitzman, M., Aschengrau, A., et al. "Boston Lead-in-Soil/Lead Free Kids Demonstration
Project-Draft Final Report to USEPA," December 1991.
11-4
-------
50272-101
REPORT DOCUMENTATION
PAGE
1. REPORT NO.
EPA 747-R-96-007
3. Recipient's Accession No.
4. Title and Subtitle
Lead Exposure Associated with Renovation and Remodeling Activities: Environmental Field
Sampling Study Volume I: Technical Report
5. Report Date
May 1997
6.
7. Author(s)
Menkedick, J.R., Menton, R.G., Constant, P., Lordo, R.A., Strauss, W.J.
8. Performing Organization Rept. No.
9. Performing Organization Name and Address
10. Project/Task/Work Unit No.
Battelle Memorial Institute
505 King Avenue
Columbus, Ohio 43201-2693
and
Midwest Research Institute
425 Volker Boulevard
Kansas City, Missouri 641 10
1 1. Contract(C) or Grant(G) No.
(C) 68-D5-0008
(G)
12. Sponsoring Organization Name and Address
U.S. Environmental Protection Agency
Office of Pollution Prevention and Toxics
401 M Street, S.W.
Washington, D.C. 20460
1 3. Type of Report & Period Covered
Technical Report
14.
15. Supplementary Notes
In addition to the authors listed above, the following key staff members were major contributors to the study: Halsey Boyd, David
Burgeon, Beth Burkhart, Paul Feder, Pam Hartford, Mary Kayser, Steve Naber, Nick Sasso, and Shawn Shumaker of Battelle; and Jack
Balsinger, Derrick Bradley, John Jones, and Gary Wester of MRI.
16. Abstract (Limit 200 words)
The U.S. Environmental Protection Agency, in response to the Residential Lead Based Paint Hazard Reduction Act of 1992 (Title X),
conducted a study of lead exposure associated with renovation and remodeling (R&R) activities. This report presents the results of a
literature review and one of the principle data collection efforts of the study: the Environmental Field Sampling Study (EFSS). The EFSS
collected 90 personal air samples and 556 settled dust samples to assess potential exposure to workers and occupants from selected R&R
activities. Task length average exposures measured by personal air samplers on R&R workers were greater than 100 fjg/m3 for paint
removal, interior demolition, and sawing, and greater than 49 fjg/m3 for interior surface preparation and central heating system
maintenance/repair. Lead loadings from stainless steel dust collectors were measured as indicators of the amount of lead disturbed and
made available by the R&R activity for exposure to occupants. With the exception of carpet removal and drilling into plaster, all activities
monitored in the EFSS deposited significant amounts of lead, ranging from 218 fjg/ft2 for sawing lead-painted plaster to 42,900 fjg/ft2 for
paint removal. Other exposure modifiers, as well as sampling methodology issues, are discussed in the report.
17. Document Analysis
Descriptors
Lead-based paint, lead hazards, renovation and remodeling, field study, wipe and vacuum dust-lead sampling, dust-lead, personal
exposure samples, worker certification, dustfall sampling
Identifiers/open-ended Terms
Lead, renovation and remodeling, worker exposure, Title X, dustfall
COSATI Field/Group
18. Availability Statement
Release Unlimited
19. Security Class (This Report)
Unclassified
20. Security Class (This Page)
Unclassified
21 . No. of Pages
203
22. Price
(See ANSI-239.18)
OPTIONAL FORM 272 (4-77)
(Formerly NTIS-35)
Department of Commerce
------- |