UNITED STATES ENVIRONMENTAL PROTECTION AGENCY
WASHINGTON. D,C, 20460
EPA-SAB-RAC-92-012
January 9, 1992
The Honorable William K Reilly
Administrator THE ADMINISTRATOR
U.S. Environmental Protection Agency
401 M Street, S.W.
Washington, DC 20460
Re: A National Survey for Radon in Schools
Dear Mr. Reilly;
On June 14-15, 1990, the National Radon Survey Review Subcommittee of
the Science Advisory Board's Radiation Advisory Committee reviewed the Office of
Radiation Programs' design options for the National School Radon Survey. These
design options attempt to respond to the requirements and constraints set by
Congress in the Indoor Radon Abatement Act and Section 118 (K) of the Super-
fund Amendments and Reauthorization Act (SARA).
Since the survey was scheduled to be undertaken during the 1990-91 school
year heating season, the Office of Radiation Programs utilized the preliminary
findings of the Subcommittee to revise parts of its study plan. The data collection
was completed on schedule and this report, therefore, constitutes an after-the-fact
report. Some of the Office of Radiation Program's changes to the original study
plan are noted throughout the report.
Both the Radiation Advisory Committee and the National Radon Survey
Design Subcommittee conclude that because this study is important from .a
national health standpoint, all efforts must be made to ensure a survey of high
quality. The Committees also recognize that the Office of Radiation Programs
(ORP) has strived to design an adequate school survey for radon within the
budget constraints imposed,
The document presents a statistically valid sampling design for selecting
school districts for the national survey for radon in schools. The Subcommittee
considers Design Option II to be preferable because, while it is not much more
complex or costly than Design Option I, it provides for 25 probability sampling
units within which alpha track detector measurements for both residences and
schools will be available for comparison. However, the Subcommittee raised
concerns about the primary radon measurement method chosen for a study of this
importance. As a result of the original proposal to use short-term charcoal
canister measurements, to be conducted over the weekend when no children are
present, the Subcommittee expected that the reliability of the results produced in
such a constrained study would be tenuous, because they could be defended only
"/." Pr'".*sd en Recycled Paper
-------
-------
as screening, or "litmus" test data, and not as valid measurements from which
realistic exposures may be derived. The potential importance of exposure to radon
in educational institutions warrants a study that would allow good determinations
of this exposure to radon, its distribution, and the key variables that influence it.
The constrained study plans originally presented would only have produced
screening data for ground-level rooms in schools; other likely interpretations would
be scientifically inappropriate. In response to these concerns, Q§P decided to
extend the charcoal canister deployment period from 2 to 7 days. The SAB is
pleased by this revision, but still wishes to stress that the formulation of conclu-
sions from screening data that cannot be wholly supported scientifically and the
preparation of technical documents using such data to support national programs
could compromise the positive aspects of the school survey,
More specific observations, findings, and recommendations can be found in
the attached report; it should be emphasized, however, that short of a continuous
measurement with alpha track detectors or repeated short-term measurement
during the entire school year, no single change in the survey design will result in
as much improvement as will the use of measurements made over longer periods
that would include the school week instead of the weekend measurements original-
ly planned.
The Subcommittee also notes that consideration of how the results are to be
used, prior to initiating the survey, will greatly enhance both data collection and
analysis and thereby strengthen the quality and defensibility of the study,
The Subcommittee appreciates the opportunity to conduct this review and
would be pleased to discuss it further with you. We also wish to acknowledge the
cooperation of the Office of Radiation Programs. The Subcommittee is- looking
forward to a formal response to the conclusions and recommendations presented in
the attached report.
Sincerely,
Raymond C, Loehr, Chairman
Executive Committee
Science Advisory Board •
Oddvar F. Nygaard, CKairman
Radiation Advisory Committee,
and National Radon Survey Review
Subcommittee
-------
-------
United State* Sci*nc* Advisory EPA^SAE-RAC 92-012
Environmental Board (A-101) January 1992
Protection Agency
&EPA AN SAB REPORT:
NATIONAL SURVEY FOR
RADON IN SCHOOLS
REVIEW OF THE ORP'S DESIGN FOR
THE NATIONAL SURVEY FOR
RADON IN SCHOOLS BY THE
RADIATION ADVISORY COMMITTEE
-------
NOTICE
This report has been written as part of the activities of the Science Adviso-
ry Board, a public advisory group providing extramural scientific information and
advice to the Administrator and other officials of the U.S. Environmental Protec-
tion Agency. The Board is structured to provide a balanced expert assessment of
scientific matters related to problems facing the Agency, This report has not been
reviewed for approval by the Agency, and, hence, the contents of this report do
not necessarily represent the views and policies of the Environmental Protection
Agency or other agencies in the Federal Government. Mention of trade names or
commercial products does not constitute a recommendation for use.
-------
ABSTRACT
Two designs for a national survey of radon in schools were developed by
Research Triangle Institute under Sandy Cohen and Associates contract #68D9-
0170 with the Environmental Protection Agency's Office of Radiation Programs
(ORP). The survey designs were submitted to the National Radon Survey Review
Subcommittee of the Science Advisory Board's Radiation Advisory Committee for
review.
The Subcommittee found either design to be statistically valid, but consid-
ered Design option II to be the better protocol. However, the Subcommittee
strongly urges the EPA to consider long-term radon measurements during occu-
pancy more representative of actual exposure than the two-day weekend charcoal
canister measurements planned by the Office of Radiation Programs, [Note; ORP
has subsequently decided to extend the deployment period from 2 to 7 days.]
Key words: radon, schools, survey, exposure
n
-------
U.S. Environmental Protection Agency
Science Advisory Board
Radiation Advisory Committee
National Radon Survey Review Subcommittee
Chairman
Dr. Qddvar F. Nygaard, Department of Radiology, Case Western Reserve
University, Cleveland, Ohio
Members
Dr. Graham W. Kalton, * Institute for Social Research
University of Michigan, Ann Arbor, Michigan
Dr. Jacqueline Michel, * Research Planning Institute
Columbia, South* Carolina
Dr. Leonard A. LoSciuto, Institute for Survey Research
Temple University, Philadelphia, Pennsylvania
Dr, James E, Martin,** University of Michigan
Ann Arbor, Michigan
Dr. Donald F. Sehutz, Teledyne Isotopes, Westwood, New Jersey
Dr. Richard G. Sextro, Lawrence Berkeley Laboratory
Berkeley, CA
Science Advisory Board Staff
Mrs. Kathleen W. Conway, Designated Federal Official
Radiation Advisory Committee, Science Advisory Board
Washington* DC
Mrs. Dorothy M. Clark, Secretary, Radiation Advisory Committee
Washington, DC
Dr. Donald G, Barnes, Director, Science Advisory Board
Washington, DC
*
NOTE: Dr. Kalton and Dr. Michel, who participated in 1986 NRRS review,
were unable to participate in 1990 NRSR review
**NOTE: Dr. Martin has been a member of the Radiation Advisory Commit-
tee since FY90 and is new to the Subcommittee
in
-------
TABLE OF CONTENTS
1. EXECUTIVE SUMMARY . , 1
2. INTRODUCTION . . 2
2.1 Background .............. 2
2.2 Charge to the Subcommittee 3
2.3 Review Process for this Report 3
3. DESIGN 5
3,1 Objectives ,....• , , , . 5
3,2 Design :............. 5
3.2.1 General Design , . . 5
3.2.2 Specific Alternative Designs 6
3.2.3 Design Recommendations 8
3.2.4 The Supplementary Sample 8
3.3 Limitations of Data '. . 8
3.3,1 Exposure . 9
3.3.2 States 9
3.3.3 Uncertainty , 9
3.3.4 ATD Added Sub-Sample 11
3.3.5 Acceptability • of Lower Confidence Limits 11
3.4 Measurement and Correlations 11
3.5 Upper Floors 12
4. ADDITIONAL SUBCOMMITTEE COMMENTS 13
4.1 Continuous Radon Measurements 13
4.2 School District Cooperation , 13
4.2.1 Training .............'... 13
4.2.2 Verification 14
4.3 Basements and Ground Floors 13
4.4 Measurement at Low Concentrations 14
4.5 Presentation of the Data , 14
4.6 Recommended Survey Design .'. 14
5, CONCLUSIONS AND RECOMMENDATIONS 15
5.1 Conclusions ...,..,.., 15
5.2 Recommendations -16
6. REFERENCES . 18
APPENDIX A: June 15, 1990 Memorandum from the
Office of Radiation Programs
iv
-------
1. EXECUTIVE SUMMARY
The proposed survey designs meet a limited objective: they will establish
a national distribution of radon screening measurements in schools and provide
an estimate of the frequency distribution of long-term measurements for public
schools nationwide using the alpha-track detector readings in the subsample. In
addition, Design Option II, which was preferred by the Subcommittee, will be
able to establish, on a limited basis, the relationship of the school study to the
National Residential Radon Survey because it provides for 25 probability
sampling units in common with the NRRS, in which alpha track detectors are
also used. Through this subsample a correlation to the radon levels in residenc-
es should be possible; however, the Subcommittee would have liked to see the
size of this subsample increased to enhance the statistical significance of the
results.
Whether these (short-term) survey measurements can be interpreted as
representative of radon levels is open to some question. Although the study
addresses the problem of making short-term and long-term radon measurements
in a sub-sample of the schools to determine whether they are well correlated,
the Subcommittee has misgivings about assuming the validity of such correla-
tions and encourages the Office of Radiation Programs (ORP) to clearly define,
in its findings, the basis for any related suppositions and conclusions,
As part of its report, the Subcommittee has attempted to provide sugges-
tions which could ensure a more useful (albeit more expensive) survey. While
recognizing the existence of budgetary constraints, the Subcommittee finds it
regrettable that the characterization of the quantitatively greatest source of
ionizing radiation exposure to the general population does not receive a higher
funding priority.
The Agency should bear in mind that an inadequate national radon
survey would be a disservice because it might well preempt the execution of any
future study of more significant scientific value.
-------
2. INTRODUCTION
2.1 Background
The "National School Radon Survey " (NSRS) responds to the Indoor
Radon Abatement Act of 1988 (IRAA PL 100-551) in which Congress required
the Administrator of EPA to "design a survey which, when completed, allows
Congress to characterize the extent of radon contamination in schools in each
state." In addition, section 118(K) of the Superfond Amendments and Reautho-
rization Act (SARA) also directs EPA to conduct a national assessment of radon
"found in structures* where people normally live and work, including educational
institutions." This survey is a first step in compliance with these Acts. In this
survey, the Environmental Protection Agency "will measure levels in a repre-
sentative sample of schools nationwide. The resulting data will permit the
estimation of radon concentration distributions in the public schools of the
nation as well as in public schools located in EPA4dentified high-risk areas"
(Reference 1),
In its May 17, 1990 briefing to the Radiation Advisory Committee, the
EPA has identified the following "Data Quality Objectives" (DQOs) for the
NSRS (see Reference 8):
1. The survey should provide a scientifically-sound estimate of
the frequency distribution of radon levels in public schools nation-
wide.
Precision constraint: The national estimate of the percent of public
schools with a radon concentration greater than or equal to- 4 pCi/L
should have a relative standard error of no more than 0.5, if the
estimate is in the neighborhood of 7 percent.
2, The survey should provide a scientifically-sound estimate of
the frequency distribution of radon levels in public schools located
in high-risk and other areas of the nation.
Precision constraint; The estimate of the percentage of schools with
a radon concentration greater than or equal to 4 pCi/L in high-risk
arms.should have a relative standard error .of no more than 0,St if
the estimate is in the neighborhood of 7 percent,
8, The room-level estimates for the NSRS should have confi-
dence intervals comparable to those of the housing-unit estimates for
the NRRS.
4. The survey should determine the correlation between short-term
and long-term radon measurements in schools.
-------
As a result of preliminary discussions, ORP staff modified the first two
DQOs so that the objective of the NSRS is to obtain the frequency distributions
of screening measurements and of measurements utilizing eight-month alpha
track detectors exposed according to EPA sampling protocols, for 1)
ground-floor and upper floor radon levels in public schools nationwide, and 2)
ground-floor radon levels in public schools located in high risk and other areas
of the nation. Each of these objectives has the precision constraint stated above
in DQO #1.
2.2 Charge to tlie Subcommittee
In its June 15, 1990 written request for the present review (Appendix A)
the Office of Radiation Programs addressed the following two questions to the
Science Advisory Board;
"(1) Are the identified Data Quality Objectives (DQOs) reason-
able?
"(2) Will these design options achieve the DQOs?"
ORP's final charge to the Subcommittee did not include all of the
objectives presented to the Radiation Advisory Committee at the May 17, 1990
meeting (identified in Section 2.1). The Subcommittee did not find it necessary
or desirable to address two of the three additional issues. However, while the
Office of Radiation Programs did not ask the Science Advisory Board to identify
the design option of preference, it became clear from the briefing and
subsequent discussions that the ORP agreed that Design Option II was superior
to Option L Therefore, in this review, comments will relate to Option II and
will include discussion of the ancillary material submitted to the Subcommittee.
2,3 Review Process for this Report
The formal request to the Science Advisory Board's Radiation Advisory
Committee for the review of its "Design of a National School Radon Survey"
(Appendix A) was submitted by ORP on June 15, 1990. The need for this
review had been communicated earlier, and at the May 17-18, 1990 meeting of
the Radiation Advisory Committee, the Office of Radiation Programs presented
a background briefing on the planned survey (8). The Committee determined
that the National Radon Survey Review Subcommittee would undertake this
review; this Subcommittee had previously reviewed the ORP's "Survey Design
for the National Radon Survey"—subsequently referred to as the "National
Residential Radon Survey" (NRRS)--in 19864987. The Subcommittee met June
14-15, 1990 in Washington, DC to review the available material and initiate the
drafting of its report.
-------
The Office of Radiation Programs provided the Subcommittee with the
following material prior to the meeting:
a. Design Options I and II—Draft Reports for the "Design of a
National School Radon Survey" prepared for the U.S.
Environmental Protection Agency's Office of Radiation Programs
by Research Triangle Park Institute. (1)
b. "Radon Measurements in Schools-An Interim Report" (EPA
520/1-89010, March 1989) (2)
c, A folder of public information materials on radon in schools, (3)
At the June 14-15, 1990 meeting, the Office of Radiation Programs
presented an extensive briefing on the survey design plans and provided addi-
tional information including;,
a, A June 8, 1990 memorandum from Research Triangle Institute's
Jane Bergsten to Lisa Ratcliff of the Office of Radiation Programs
concerning the list of variables to be used for questionnaire
construction and forms development, (4)
b. A June 13, 1990 "Addendum to Option II Draft Design Report:
Measuring Upper Floor Rooms in a Subsample of Schools" (draft of
June 3, 1990).(5)
c. The Subcommittee's previous report on the National Residential
Radon Survey (SAB~RAC-88-OQ2).(6)
d. Exhibit 1. Expected precision of estimates based on ATD
subsample 95 percent confidence limits for estimated percentages
assuming a sample of 125 Schools in 25 primary sampling units. (7)
The following sections will provide detailed comments to the various
aspects of the proposed design of the "National School Radon Survey."
-------
3. DESIGN ISSUES
3,1 Objectives
The objective of the survey is to determine the national frequency
distribution of radon levels in schools. The attainment of this objective is
complicated by a number of factors, including the fact that building operating
conditions in many (perhaps most) schools are different during times of occu-
pancy than during non-occupied hours. The operation of the heating, ventila-
tion and air conditioning (HVAC) system, if it exists, and opening of doors and
windows are two such factors; both can affect indoor radon concentrations,
although the magnitude and direction of the effects cannot be known at present,
due to limited data in schools. Nevertheless, the implementation phase of the
school radon survey should attempt to minimize these uncertainties in order to
provide a reliable estimate of the concentration distribution. [Note: ORP's
amended approach, basing most measurements on 7-day short-term sampling is
an improvement over the initial 2 day deployment, and should provide a more
realistic assessment of screening measurements during occupancy,]
3,2 Design
3,2.1 General Design
The proposed Design Option II for the NSRS is to obtain short-term
radon screening measurements via charcoal canisters in a nationwide probability
sample of approximately 1,100 public schools during the 1990-91 heating season.
In addition, both short-term screening and long-term alpha track detector (ATD)
measurements will be made in a second probability sample of approximately 125
schools selected within the sampling units used in the National Residential
Radon Survey (NRRS).
Prior to measurement, the selected school districts will be contacted for
authorization to gather information. Once permission is obtained, necessary
information about the sample schools will be collected. This includes the
number of buildings in the school, the structure and ventilation characteristics
of the buildings, and the number, types and locations of the school rooms, A
radon testing period will be subsequently agreed upon and scheduled,
Detectors and instructions for their placement will be shipped directly to
the person appointed by the school or school district to deploy them in the
designated survey-eligible rooms.
Survey-eligible rooms for the canister-only sub-sample are occupied
"ground floor" rooms with direct ground contact, or those separated from the
ground by a crawl space. For the ATD sample, occupied rooms on all floors
will be eligible for canister and ATD placement. After deployment for the speci-
-------
fled time period, the detectors will be retrieved and immediately shipped to
either of the two EPA laboratories for analysis or, in the case of ATDs, to a
commercial laboratory for analysis.
The Subcommittee believes that this general design approach to aceonv
plishing the project's goals is straightforward, reasonable and efficient so long
as those goals are delimited as expressed above.
3.2.2 Specific Alternative Designs
The contractor originally delineated two basic possible designs for the
NSRS (a third option was prepared, but was excluded from further consider-
ation by ORP), Design options I and II vary considerably in detail but are
consistent in approach and conformity to the following guidelines;
a. Stratifying by EPA region and by designation of "high" vs. "low"
indoor radon level potential defined on the basis of available state
and local data,
b. Clustering of schools within school districts to facilitate solicitation
of participants. (Design I and II only)
c. Gathering of school building and room information from sample
schools by telephone.
d. Selecting survey-eligible rooms in- sampled schools.
e. Identifying a smaller sample of schools to receive both short-term
canister measurement and long-term alpha track detector (ATD)
measurement.
f. Drawing of a supplementary sample for sample size enhancement if
participation rates are unexpectedly low.
A summary of the commonalities and differences of the two designs (as
well as a third alternative rejected by ORP) is provided by Exhibit 2-2 of the
proposal (reproduced on the next page). As can be seen in Exhibit 2-2, Design
Options I and II employ school districts as Primary Sampling Units (PSUs),
and differ mainly in the method of selection of the combined ATD and charcoal
canister subsamples. The ATD sample consists of a 10 percent subsample of
those schools. The Radiation Advisory Committee (RAG) was not asked to
evaluate Design III which ORP had determined to be too costly for the available
budget, (Design Option III differs from I and II in that it uses the 125 county-
based PSUs selected for the NRRS, and schools are selected directly within the
PSUs.)
-------
2-2 Su»*ijry OescrlpHoci »F **«?ie C*e%i 1 1 i f JF ^ repart f on *1 E*
»«Bb*r of tludrflti I* dlvtrlet.
abml
J. Mfjiur* radon lm»ll In til
eligible rorMt ta **eh
jctMiol. yl*ldfnQ jfoawt
mom.
S*1* of lltr 1?5 WMS
i ithiwlt ftr K» far f
«l tht 75 KBf KUl^ yleNInf !
KhaoN fn ikoul 100 tthwil 4»
to r«ct1*( tath AID Md «*n lite
J. S-xw Jt far Deilgn 1, bwt
jit Wing (»Kl.WO}-*t.HN
SMK Jv fer Dcitqn t
Wuuirt riil
Cllfllklc tmwii, yFeMfnf 7, WO tttlplt
oom% for AID *e4furr*ent In
Tot*l napl* r«*lTl»t -Ctnliter im
it itoot 3/0 tciioal ^itlricll, thovl 1,70S
IchoeH *nif 2*, 100 icltool room.
III
I. Utf Ihc l?S fS(t Hint «er* trlrcted fPA ref Fi
for the KRPS. A fill contains in I**tl. u
•vertift of o 5i:hool 4Hie(« maw.
jhtmt
N^ecEvii Indoor
ral !*?«. nwfib^r
di^ units.
«< €or Oet ign )
Select I 111 percml tiri»Mpl« sf
lo rrc^lv^ AID «VAtumnetit lit
En c-dtiitlrr »«*$«rr««ml ,
I?O schanU in !?0 uhaal
t Mj IK rStJV,
, rUNinf ?,
-------
3,2-3 Design Recommendations
In the Subcommittee's judgement, Design II is the better of the designs
for present purposes. The perceived advantages and disadvantages of each are
outlined below,
a. Design I's major advantages are simplicity in sample design,
implementation, and calculation of sampling errors. There also may
be some cost savings since fewer school districts will be needed as
a result of the straightforward method of drawing the subsample
for ATP vs, canister measurement comparison. However, this
scenario precludes the possibility of determining the relationship
between residential radon levels and school screening levels.
b. Design II is not much more complex or costly than Design I, but
provides for 25 PSUs (in common with the NRES) in which ATD
measurements for both residences and schools will be available for
comparisons. The ATD sample component of either design is
confined to a relatively small number of schools (about 125) so that
instructing school personnel in ATD placement and retrieval,
another quality control issue, will be feasible.
3.2.4 The Supplementary Sample
As noted above, the basic design includes a provision for supplementary
sampling in case selected districts/schools refuse to participate. The Subcommit-
tee wishes to emphasize, and ORP agrees, that the supplemental sample is to
be used only as a last resort, after all reasonable efforts to elicit cooperation
from the originally selected schools are exhausted. The potential nonresponse
bias inherent in a large number of refusals Is great and cannot be reduced
simply by adding other schools. It is possible, for example, that those admin-
istrators who refuse to participate may have reasons directly related to suspect-
ed potential radon contamination.
3.3 Limitations of Data
There are three major observable limitations to the data:
a. Short term (screening) measurements are inadequate for making
long-term (school year) exposure estimates. The Subcommittee
recognizes, however, that development of exposure estimates was
not one of the survey's stated objectives.
b. Individual short-term and to a lesser extent long-term'radon.
concentration measurements, are still surrounded by relatively high
uncertainty.
8
-------
c. Other, less significant, limits to the study and the resulting data
are that only public schools will be surveyed and that the survey
will not provide statistically-valid results for each individual state.
3.3.1 Exposure
In planning and implementing a national radon survey for schools, the
use and interpretations of the results should be a major consideration in the
study design. With the attention given radon as an issue and the frequent
misinterpretation in the press of the results of similar data previously obtained,
it is important to provide data which are as reliable as the current scientific
understanding allows. One key issue that will arise' is the potential, if unintend-
ed use of the data to infer individual exposures from the survey data. This
might be attempted in order to place the school results in context with average
exposures in houses, or to interpret the short-term radon level data in light of
the other variables affecting exposures (time and activity patterns, for example).
Screening measurements of the type proposed in this study are a poor basis for
making exposure estimates since it is difficult to relate the observed screening
levels to the long-term average radon concentration, Further, utilization of
short-term measurement data without a reliable basis for estimating exposures
(and hence risks) could result in inappropriate decisions on mitigation. To
reiterate, the Subcommittee recognizes that ORP did not intend to estimate
exposures; however, the possibility that others might use the data in this way is
so great that the Subcommittee feels compelled to stress the tenuous connection
between screening measurements and exposures in this review. Therefore, ORP
in releasing the results should be careful to emphasize the fact that these are
screening data and that using them as a basis for estimating long-term concen-
trations or exposures in individual cases is inappropriate,
3.3.2 States
The NSRS will not provide statistically-valid results for each individual
state, but will provide a reasonable representation of the shape of the national
distribution of radon in the schools.
3.3.3 Uncertainty
The precision constraints described in DQO 1 and 2 are reasonable in
using the survey data to infer the distribution of radon screening measurements
in public schools nationwide.
A significant limitation of the data is the uncertainty inherent in the
actual measurements. This will raise issues regarding the accuracy attached to
any single measurement, just as it has in interpreting results from short-term
residential measurements. The current study design does not address this
problem, although it is a necessary component in determining the statistical
validity of the study results. There are also precision errors encountered in
9
-------
using the seven-day "diffusion barrier" charcoal canisters to sample time-variable
radon concentrations. (Note: See, however,, the note added under Section
3.3,5.)
As the EPA Interim Report, "Radon Measurements in Schools" (March
1989) notes, there are significant uncertainties associated with short-term
measurements of radon in schools conducted according to EPA protocols
(Reference 2: Interim Report). Current research, as summarized in this interim
report, indicates that school heating, ventilation, and air conditioning (HVAC)
systems are important potential determinants of school radon concentrations,
though the effects may not be consistent: "It has been observed that having the
HVAC system operating normally, at a reduced rate, or completely shut down
can increase or decrease radon concentrations depending upon the type of
ventilation system and the construction of the school, (Reference 2, p,4).
The report also notes .that "radon concentrations vary significantly over
time. Changes in ventilation, occupancy patterns, weather conditions and other
variables may cause maximum screening concentrations in a room to vary by as
much as a factor of 10 or more. Average concentrations may vary by a factor
of two to three. The variability found in schools may be higher than that
found in houses" (Reference 2, p, 5).
Further, the Interim Ueport outlines the advantages and disadvantages of
short-term charcoal canister measurements vs those for alpha track detectors.
(2) Two of the disadvantages of using the short-term measurements are:
"Two-day measurements may be affected by ventilation
systems. Two-day measurements may reflect fluctuations in radon
concentrations caused by changes in the ventilation system
operation" Longer term measurements are less susceptible to these
types of changes.
"Two-day measurements vary over time. Radon
concentrations in schools can fluctuate dramatically over time. If
two measurements are made in the same school room on different
weekends, the radon concentration may differ by a factor of 2 to 3"
Based on these observations, the Subcommittee concludes that the
specified Data Quality Objectives 1 and 2 would not be met under the original
two-day measurement scheme. Further, the estimates of confidence limits for
the short-term data do not factor in the variability inherent in the measurement
protocols. Thus, the confidence intervals for these data would not be comparable
with those of the housing-unit estimates for the National Residential Radon
Survey i.e., DQO 3 will not be met, (Note: ORP has subsequently decided to
extend the deployment period from 2 to 7 days. Although this is a significant
improvement in that it reduces the inherent measurement uncertainties, it is
10
-------
not clear whether this reduction in uncertainty will allow the stated DQO to be
met,)
3.3.4 ATD Added Sub-Sample
The survey also proposes to make long-term measurements in a sample of
the schools being surveyed with charcoal canisters, in part to correlate long-
term and short-term measurements (Reference 7), Similar attempts to correlate
long and short-term measurements have been made on data from residences,
and a wide variation in the relationship between these two techniques has been
observed. The Subcommittee is concerned that the correlation for schools, given
expectations that ration concentrations may show greater temporal variability
than homes, could be very low. This would further confound attempts to use
short-term sampling data as a basis for making reliable estimates of longer-term
average radon concentrations,
3.3.5 Acceptability of Lower Confidence Limits
Variance components arising from measurement and sampling errors and
in day-to-day variation in radon concentration are not included in the calcula-
tion of confidence intervals for given measured radon level. Because these errors
may be large compared to errors associated with statistical design they should
be included in the calculation of confidence intervals, [Note: the SAB has been
informed that ORP will attempt to estimate and account for measurement error
through the use of blanks, spikes, and duplicates,]
3.4 Measurement and Correlations
Data Quality Objective No. 4 concerning the correlation between short-
term and long-term measurements in schools raises several issues regarding
analytical methods, radon sampling protocols and quality assurance related to
the radon sampling process. In general the Subcommittee recognizes the
importance of seeking to validate and improve the usefulness of data obtained
from the commonly employed short term (2-day or 7-day) charcoal canister (CC)
tests by including a controlled comparison of that method with 8-month ATD
measurements as part of the NSRS. The draft design did not detail how such a
correlation would be established, however, and the Subcommittee recommends
that the methodology be carefully analyzed. In its analysis of the correlation,
the Agency should consider which radon tests were used, the time period of
measurements, the influence of weather conditions and the operation of the
HVAC system on the measurements.
The modified seven-day charcoal canister tests, with day and night
exposure under "closed window" conditions but with normal HVAC operation,
still constitute a screening effort that will not provide measurements and data
directly usable as an indicator of exposure to the school population which
occupies the school only on weekdays during day-time hours. The 8-month
11
-------
ATD samples left in place during the school year also are exposed during
weekends and at night, but will provide a truer picture of the radon levels to
which the school children are exposed.
There are two aspects to the issue regarding correlations between short-
term and long-term measurements in homes. The first is whether the aggregate
data sets show correlations between the two types of measurements and
secondly, how large the coefficient of variation is for any one set of measure-
ments. The number of measurements should be adequate to determine how .
well the aggregate short-term and long-term data correlate, However, screening
measurements for a, single room may not correlate well. The residential data
sets obtained from studies throughout the U.S. show a considerable variation in
the correlations obtained, some of which appear to be related to seasonal
factors, or to local building design and construction practices. Data from
schools can also be expected to also depend on a number of complicating
factors,
The Subcommittee is concerned that the short-term/long-term
measurement correlation for schools could be very low, given expectations that
radon concentrations in schools may show greater temporal variability than in
homes. This will further confound any attempt to use short-term data as a
basis for reliably estimating long-term concentrations.
3,5 Upper Floors
It is impossible to gain a representative estimate of radon levels, in the
absence of data on the number of rooms there are on upper floors and whether
they differ from ground floor rooms with regard to radon levels. Where a
school building that is selected for sampling has upper floors at least one upper
floor should be sampled in its entirety. (Note: In the amended design all
schools in the ATD subsample had both ATD and charcoal canisters placed in
all occupied rooms, regardless of floor level)
12
-------
4. ADDITIONAL SUBCOMMITTEE COMMENTS
4.1 Continuous Radon Measurements
The only way to obtain a meaningful correlation between the two
sampling methods is to run a continuous radon monitor (CRM) with hourly
data recording during the entire period of sampling. In this way, the concen-
tration in the room while occupied ean be measured and the relationship
between the short term CC test and the 8-month 24 hour/day ATD test can be
assessed, Obtaining correlative CRM data is beyond the scope of the NSRS
but such data may be obtained from other research. If resources are not
available to provide continuous monitoring data, the Subcommittee strongly
recommends that several seasonal CC tests be made during the 8-month ATD
tests so that the variability of the CC tests can be determined.
4,2 School District Cooperation
4.2.1 Training
To ensure quality control, the personnel designated by the school for
canister or ATD emplacement should be properly trained, by attending a radon
testers course as offered at a regional training center. Special training could be
done at part of the course to assure, at a minimum, that designated personnel
had read, understood, and could follow instructions given in the sampling
protocol. (Note: As an alternative, ORP prepared a training video which was
distributed to school districts along with written instructions.)
4.2.2 Verification
The Quality Assurance Program must include some type of on-site
verification of canister/ATD placements and record keeping. The design
document mentions verification but gives no details or plan for implementation.
A need for verification exists because of the adverse impact on the validity of
the study if there is no QA on the sampling process. Of course, training
should be at no cost to the school districts, and there is also the possibility
that industry might volunteer to help with verification,
4.3 Basements and Ground Floors
Basement rooms should be measured if they can be occupied. While the
focus is on students, school employees must also be considered as in the recent
discussions of exposure of janitors to asbestos. (Note: All ground-contact,
occupied rooms were tested.)
13
-------
4*4 Measurement at Low Concentrations
The sensitivities of the CC and ATD measurements are generally suffi-
cient to measure radon at the expected concentrations. One would not antici-
pate severe impact if the protocol calls for measurements at a lower limit of
detection (LLD) of
-------
5. Conclusions and Recommendations
5.1 Conclusions
The current approach of basing most measurements on single-weekend
short-term CC sampling does not provide the most reliable method of obtaining
sound estimate of the radon concentration distribution, (3.1). (Note; ORP subse-
quently amended the design to use 7-day sampling periods for CC measure-
ments.)
It is important that the distinction between screening tests, average
concentrations over various time periods, continuous real-time measurements,
and estimates of human exposure not be blurred. Screening measurements of
the type proposed in this study are a poor basis for making exposure estimates,
since it is difficult to relate the observed screening levels to the long-term
average radon concentration, (3,3.1) Even if ORP does not intend to use the
data in this way, such use by others is almost inevitable.
As the EPA Interim Report, "Radon Measurements in Schools" (March
1989) notes, there are significant uncertainties associated with short term
measurements of radon in schools conducted in accordance with EPA protocols
(Reference 2),
The Interim Report, in discussing the advantages and disadvantages of
short-term CC measurements versus those for alpha track detectors notes that
two of the disadvantages of using the short-term measurements are that
"Two-day measurements may reflect fluctuations in radon concentrations caused
by changes in the ventilation system operation" and that, "two-day measure-
ments vary over time..,.. If two measurements are made in the same schoolroom
on different, weekends, the radon concentration may differ fay a factor 2 or 3."
Current research, as summarized in this interim report, indicates that school
heating, ventilation and air conditioning (HVAC) systems are important poten-
tial determinants of school radon concentrations, though the effects may not be
consistent (3.3.3).
The estimates of confidence limits for the short-term data do not factor
in the variability inherent in the radon measurement protocols. Thus the
confidence intervals for these data will not be comparable with those of the
housing-unit estimates for the National Residential Radon Survey and thus
DQO 3 may not be met, (3,4)
The ATD sample component is confined to a relatively small number of
schools (about 125) so that instructing school and personnel in ATD placement
and retrieval and other quality control measures will be feasible. (3.2.4)
15
-------
Based on these observations, the Subcommittee, in summary, concludes
that:
a. The correlation of two-day short-term weekend or seven-day
charcoal canister tests with long-term ATD tests for schools, could
be very low. Even if the correlations among the aggregate data
are acceptable, the interpretation of single measurements may still
contain substantial uncertainties, (3,4)
b. Variance components arising from measurement and sampling
errors ^are not included in the calculation of confidence intervals for
given measured radon level. (3.3.7)
c. One way to estimate whether a meaningful correlation between the
two sampling methods is to run a continuous radon monitor (CRM)
with hourly data during the entire period of sampling. (4.1)
5,2 Recommendations
a. The Subcommittee recommends the use of Design Option I! with a
longer than 2 day survey period and an Addendum for upper floor
sampling. It further recommends an added requirement for
training and for the use of Continuous Radiation Monitoring
(CRM) or multiple Charcoal Canister (CC) tests to allow
interpretation of exposure data.
b, Unless the radon measurement methodology is amended to satisfy
the concerns of the Subcommittee, the objectives of the survey need
to be restated to reflect the limitations of the information that can
be obtained using primarily screening data with a relatively small
subsample of long-term ATD added. (3.1)
c. Based on its observations (as delineated in "Conclusions"), the
Subcommittee is concerned that not all of the overall objectives of
the study will be met,
d. Provided the goals of the survey are delimited as expressed above,
the Subcommittee believes that this relatively straightforward
general design approach to accomplishing the project's goals is
reasonable and efficient (3.2.1).
e. In the Subcommittee's view, Design Option II is the better of the
two relevant designs for present purposes since it is not much
more complex or costly than Design Option I, and yet provides for
25 PSUs in which ATD measurements for both residences and
schools will be available for comparisons (3,2.3),
16
-------
f. The Subcommittee wishes to emphasize that'a supplementary
sample should be used only as a last resort, after all reasonable
efforts to elicit cooperation from the originally selected schools are
exhausted (3.2,4),
g. The Subcommittee recommends that the methodology be carefully
analyzed and explicitly stated with regard to what Is being
measured, the time period of measurements and the influence of
weather conditions and the operation of the HVAC system (3.4).
h. If resources are not available to provide continuous monitoring
data the Subcommittee suggests that several CC tests be made
during the 8-month ATD testing period so that the variability of
the CC tests can be determined (4.1),
The Subcommittee has additional concerns with regard to carrying out of
a limited radon survey, in schools. Specifically, these concerns are that:
a. The results, when available, might be overinterpreted by the press,
the public and even other governmental agencies, regardless of
what cautionary notes might be inserted in the final report, and
b. A limited national radon study in the schools might preempt the
execution of any future study of greater statistical value.
17
-------
6. References
1. Design Options I and II-Draft Reports for the "Design of a
National School Radon Survey" prepared for the U.S. Environmental Protection
Agency's Office of Radiation Programs by Research Triangle Park Institute.
2, "Radon Measurements in Sehools-An Interim Report" (EPA
520/1-89-010, March 1989).
3. A folder of public information materials on radon in schools,
4. A June1 8, 1990 memorandum from Research Triangle Institute's
Jane Bergsten to Lisa Ratcliff of the Office of Radiation Programs concerning
the list of variables to be used for questionnaire construction and forms devel-
opment
5, A June 13, 1990 "Addendum to Option II Draft Design Report;
Measuring Upper Floor Rooms in a Subsample of Schools" (draft of June 3,
1990),
6. The Subcommittee's previous report on the national residential
radon survey (SAB-RAC-88-002).
7. Exhibit 1, Expected precision of estimates based on ATD subsam-
ple 95 percent confidence limits for estimated percentages assuming a sample of
125 Schools in 25 primary sampling units
8. National Schools Radon Survey, Science Advisory Board Briefing,
prepared by the Office of Radiation Programs, May 17, 1990.
9. National Schools Radon Survey, Science Advisory Board Briefing,
prepared by the Office of Radiation Programs, June 14, 1990.
18
-------
DISTRIBUTION LIST
Assistant Administrators
Regional Administrators
Director, Office of Radiation Programs
Director, Office of Policy, Planning and Evaluation
Headquarters and Regional Librarians
19
-------
»
UNITED STATES ENVIRONMENTAL PROTECTION AGENCY
WASHINGTON, D.C. 20460
J|)AJ
<*>Lt.iy
Of
AIR AND RADIATION
MEMORANDUM
SUBJECT: Review of the Design Options
Schools Radon Survey
FROM: Richard J. Guimond, Directo
Office of Radiation Programs
TO: Donald G. Barnes, Director
Science Advisory Board
On May 17, 1990, members of my staff briefed the SAB's
Radiation Advisory Committee on the design options for the
National Schools Radon Survey (NSRS). At that time, the
Committee decided to reconvene the Subcommittee which previously
reviewed the design report for the National Residential Radon
Survey for the purpose of reviewing the NSRS design options. The
Committee also identified a number of preliminary concerns with
these design options, and we are currently modifying one of the
options to reflect the Committee's comments.
The Subcommittee will meet on June 14-15 to collectively
discuss the NSRS. Members of my staff will be present to answer
any questions the Subcommittee may have, I would like the*
Subcommittee to focus its review on the two questions listed
below:
o Are the identified Data Quality Objectives (DQOs)
reasonable?
o Will these design options achieve the DQOs?
-------
In order to stay on schedule and. initiate the data
collection phase of the survey this Fall, we need to submit the
design report to the office of Management and Budget early next
month, I therefore request that the Subcommittee forward its
comments to ORP as soon after the June 14-15 meeting as possible.
Thank you for your assistance in this matter.
cc: Kathleen Conway (A-101F)
Gordon Hurley (ANR-458)
Margo Oge (ANR-464)
------- |