-------
Table D1-5. Number of Residential Units that Contained (N) Individual Samples of Each
Component Type Based on First Site Visit Clearance Testing Data from the
Grantees with Higher Floor Dust-lead Clearance Standard in the HUD Grantee
Program.
inap^siiiifei
236
263
732
413
1068
727
334
501
170
343
133
46
258
41
13
288
20
142
8
36
9 +
42
Total
2092
2031
1697
Total with N 2:5
766
66
22
For each component type within a residential unit, the set of individual clearance sample
lead loading results for the first site visit were used to construct simulated composite samples for
the purpose of evaluating the three composite sample clearance criteria introduced hi Section
5.4.2 (Standard, Standard/n, and 2>
-------
Table D1-6. Numbers of Residential Units that Pass or Fail Clearance, Based on Individual
Sample Clearance Results vs. Simulated Composite Clearance Results, Using
Data from Grantees with Higher Floor Dust-lead Clearance Standard.
IndiviauaUSamplelClearanca
Pass
Standard
Inconclusive
Fail
1775
97
1854
72
63
157
98
1546
33
117
Pass
Standard/n
Inconclusive
Fail
1622
1797
78
75
317
55
177
1527
18
0
151
Pass
2xStandard/n
Inconclusive
Fail
1730
8
1829
8
28
62
17
247
25
167
1533
13
148
Values of the four performance characteristics defined in Section 5,4.3 (sensitivity,
specificity, positive predictive value, negative predictive value) are presented in Table Dl-7 for
each combination of component type and composite clearance criterion. By design, the
sensitivity for the Standard/n Rule is always 1.00 (as all sets of simulated composite samples
would fail if at least one individual sample result failed) while the specificity for this rule is
estimated at 0.91 for floors, 0.97 for window sills, and 0.99 for window troughs. In contrast, the
specificity of the Standard Rule is always 1.00 (as all sets of simulated composite samples would
pass if all individual samples passed) while the sensitivity for this rule is estimated at 0.50 for
floors, 0.55 for window sills, and 0.77 for window troughs. The 2>
-------
Table D1-7. Performance Characteristics of Composite Clearance Criteria Based on Data
from Grantees with Higher Floor Dust-lead Clearance Standard.
Floors
Sensitivity
Specificity
PPV
NPV
0.50
1.00
1.00
0.95
1.00
0.91
0.81
1.00
0.78
0.97
0.94
1.00
Window Silts
Sensitivity
Specificity
PPV
NPV
0.55
1.00
1.00
0.96
1.00
0.97
0.76
1.00
0.94
0.99
0.87
1.00
Sensitivity
Window Troughs
Specificity
PPV
NPV
0.77
1.00
1.00
0.98
1.00
0.99
0.89
1.00
0.98
0.99
0.92
1.00
The above estimates of the performance characteristics illustrate that the three composite
clearance testing criteria have different specificity and sensitivity rates. These rates correspond
to the consistency between clearance decisions and the true lead hazards present in the various
locations sampled (assuming the individual sample lead-loading results are representative of
these lead hazards). To further characterize the performance of each of the three composite
clearance criteria, the following logistic regression model was fitted to clearance data for each
combination of component type and composite clearance criterion:
where nijk is the estimated probability of clearance for component(/) in house(i) under composite
criterion (£), and Max,-, is the maximum individual sample lead loading result in house(z) for
component(/).
D1-19
-------
For each component type, the estimated relationship between the probability of passing
clearance and the maximum lead-loading is presented graphically in Figure Dl-9. In this figure,
the solid, long-dashed, and finely-dashed lines represent the estimated relationship for the
Standard, Standard/n, and 2>
-------
ProtebBy of Pasak^j Ct
1.0
S UshQ Composite Floor Sempte*
100
200 300 400 coo 000
Maximum kxMduaJ Sampto ROOT Pb Loading
TOO
BOO
Pretoobtitty of Porafcig
U*ig Comports Window SB Samptaa
2SO
500 78D 1000 12GO 1600
MEDdmum HMdUal Sarr^n. Wnttow SB F* baadUg
17GO
sooo
ProbabBy of PBSS*« CteennM lattng Udr« CompoBte WMow WM SenfUae
(2 x SuintKd / n) HUa
n) FU»
2000
2400
2800
3200
Maximum MMduaJ San** WMnv Wai Pb Laadng
Figure D1-9. Estimated Relationship Between the Probability of a Residential Unit Passing
Clearance Testing versus the Maximum Individual Lead-Loading Results by
Component Type Based on Simulated Composite Samples from the Grantees
with Higher Floor Dust-lead Clearance Standard in the HUD Grantee Program.
D1-21
-------
Dl-22
-------
APPENDIX D2
Grantees with Lower Floor Dust-Lead
Clearance Standard in the
HUD Grantee Program
D2-1
-------
APPENDIX D2
GRANTEES WITH LOWER FLOOR DUST-LEAD
CLEARANCE STANDARD IN THE
HUD GRANTEE PROGRAM
Appendix D2 presents analysis results on the clearance data collected from five grantees
that used a lower floor dust-lead clearance standard (i.e. 100 ug/ft2 or 80 ug/ft2). The clearance
standards still remained at 500 ug/ft2 and 800 ug/ft2 for window sills and window troughs,
respectively. The five grantees are Cleveland, Chicago, New Jersey, and New York City which
used 100 ug/ft2 as a floor dust-lead clearance standard, and Minnesota which used 80 ug/ft2.
Note that, as explained in Appendix D, 13 dwelling units did not have first site visit dust-
lead clearance data and were therefore not included in the first site visit analysis. However, since
these dwelling units had data for other site visits, they were included in those analyses. Fifty
(50) dwelling units did not have second site visit data but did have data for other site visits
(including first site visit data) and were included in those analyses.
To be consistent with other data sources presented in this report, a floor dust-lead
clearance sample in the analyses presented below was labeled as a "pass" if its loading was
below 200 ug/ft2 and as a "fail" if its lead loading was greater than or equal to 200 ug/ft2, despite
the lower clearance standard used by the grantee.
D2-1. Objective 1: Characterization of the Number of Individual Samples. Work Areas.
and Housing Units That Pass or Fail Clearance Testing Standards
Individual dust wipe samples were collected from floors, window sills and window
troughs as part of the HUD Grantee Program. A total of 7,664 samples were taken from 4,936
rooms in 1,038 residential units from five grantees using the lower floor dust-lead clearance
standard (either 100 or 80 ug/ft2). Table D2-1 presents the number of individual samples, work
areas and residential units that passed or failed clearance testing within each combination of
component type and site visit. Approximately 93% (7,135/7,664) of the dust samples were
collected during the first site visit to a residential unit. On the first site visit, 95% (6,782/7,135)
of the dust samples fell below the clearance standards of 200 ug/ft2 for floors, 500 ug/ft2 for
window sills and 800 ug/ft2 for window troughs while 93% (4,191/4,522) of the rooms and 78%
D2-2
-------
(796/1,025) of the residential units passed clearance. The increase in the failure rate from the
percentage of individual samples that fail to the percentage of rooms and residential units that
fail is attributable to the fact that if any individual sample exceeded the standard and failed
clearance, then both the room and residential unit also failed clearance.
Table D2-1. Clearance Testing Results by Individual Sample, Room, and Residential Unit
from the Grantees with Lower Floor Dust-lead Clearance Standard in the HUD
Grantee Program
W&r^
ilvisfe^
First
Second
Third
Fourth
Total
iXw^o^sssst
|Componenti
M^fce'stealfl
Floor
Sill
Trough
All
Floor
Sill
Trough
All
Floor
Sill
Trough
All
Floor
Sill
Trough
All
Floor
Sill
Trough
All
lISXvu*"**i'''l^i?"**ii*il*s"'!B!*
^fe?OT« SamRlesJf
Ipall
3486
2068
1228
6782
226
83
90
399
30
18
19
67
7
0
4
11
3749
2169
1341
7259
3sf&*"*l?ij!|l
156
79
118
353
19
8
18
45
3
0
4
7
0
0
0
0
178
87
140
405
SSSSfflSs*:
STotal?
3642
2147
1346
7135
245
91
108
444
33
18
23
74
7
0
4
11
3927
2256
1481
7664
's&BSf™™''''**? !'": •« • j'-'~ :.-'
-MJRoomslSampled.f^
••%t&l^*&-'-^'-&$#y&^r-^''*^'Z'''!:' •sftjitvs-
3410
2044
1221
4191
216
82
90
350
30
18
19
64
7
0
4
11
3650
2138
1333
4595
^Fail-T?
154
79
118
331
19
8
18
44
3
0
4
7
0
0
0
0
156
81
124
341
KS!
3564
2123
1339
4522
235
90
108
394
33
18
23
71
7
0
4
11
3806
2219
1457
4936
' K^Wi* * " Ta* Jgg*BliWj«!S3«»
•^Houses iSampJedjp
909
897
772
796
131
66
75
189
23
15
18
45
4
0
4
8
1005
951
849
971
2WMBI
112
71
101
229
16
8
17
39
3
0
4
7
0
0
0
0
25
22
32
67
^Totals
1021
968
873
1025
147
74
92
228
26
15
22
52
4
0
4
8
1030
973
881
1038
D2-3
-------
The failure rate for individual samples during the first site visit was at 4.3% (156/3,642)
for floors, at 3.7% (79/2,147) for window sills, and at 8.8% (118/1,346) for window trough
samples. The failure rate for individual samples showed an increase from the first site visit to the
second site visit and again from the second visit to the third site visit.
Of the 1,025 dwelling units in Table D2-1 for which initial post-intervention clearance
sampling (first site visit) data were available, 77.7% (796/1,025) passed on the first attempt. A
total of 11.0% (112/1,021) had at least one floor location with a dust-lead loading above the
clearance level. This failure rate was higher than that reported for window sills (7.3%, 71/968)
and lower than that reported for window troughs (11.6%, 101/873).
D2-2. Objective 2: Characterization of the Distribution of the Dust-lead Loadings.
Geometric Mean Dust-lead Loadings. Variability Within a Housing Unit, and
Variability Between Housing Units
Preliminary assessment of the data indicated that the distributions of dust-lead loading
clearance sample results were highly skewed. Therefore, a natural logarithm transformation was
applied to the data.
For the clearance data collected from five grantees with the lower floor dust-lead
clearance standard, Table D2-2 lists the geometric mean dust-lead loading, variance components
(within-home variability and between-home variability) associated with these loadings, and a
95% confidence interval for the geometric mean, calculated for each site visit and component
type. No sill dust sample results were available for the fourth site visit. Notice from this table
that all of the variance components are within a single order of magnitude of each other. Also,
the lengths of the confidence intervals increase from the first site visit to the third site visit for
any given component. The increase in the length of the confidence intervals is due primarily to
the decrease in the number of samples used to estimate the variance components.
The geometric mean lead loadings and their 95% confidence intervals were used to
compare trends between site visits for a given component type and to compare average results
between component types within a site visit. If the 95% confidence intervals on two geometric
means do not overlap, then these two geometric means are statistically significantly different at a
level less than 0.05. A comparison across site visits shows that the geometric means, from the
first site visit to the third site visit, increase gradually but were not statistically significant:
D2-4
-------
o
o
I
CB
O o
"I
if
II
« o
o —
« n
E c
sl
O (A
CN
CN
O
in
to
o>
o>
O)
CO
CO
CO
o
CM
CO
CM
CM
CM
IO
CO
O>
in
o>
o
CO
10
10
O)
CO
CO
CM
LO
O
LO
CO
0)
CO
IO
6
o
O)
CO
IO
§
O)
O)
10
LO
CO
CO
CN
00
CM
CM
10
CM
CM
CN
CO
q
CM
CO
CO
CM
O>
it
CM
IO
CO
(O
CO
CO
IO
o
CO
q
o>
O)
PX
O>
to
CO
CO
CO
o
o
CO
CM
o
CO
CO
0)
CO
r-
CO
CM
O)
CO
CM
LO
CM
CM
-------
12.3 fig/ft2,13.8 ng/ft2,19.7 ug/ft2 for floors and 24.9 ug/ft2, 30.5 ug/ft2, 38.6 ng/ft2 for window
sills. The geometric means for window troughs increase from 66.5 ug/ft2 to 99.0 ug/ft2 from the
first site visit to the second site visit but decrease to 97.1 ng/ft2 on the third site visit. A
comparison of the components tested for the first site visit shows that the floor dust-lead loadings
were significantly less than the window sill dust-lead loadings, which were in turn significantly
less than the window trough dust-lead loadings. Note that estimates from the fourth site visit
were associated with very small sample sizes and may not be reliable for establishing trends.
Figures D2-1 and D2-2 contain box and whisker plots that present the distribution of
dust-lead loadings from the first and passed clearance visits by component type. Figures D2-3 to
D2-5 contain box and whisker plots that present the distribution of dust-lead loadings from the
first visit for floor, window sill, and window trough, respectively; Figures D2-6 to D2-8 contain
box and whisker plots that present the distribution of dust-lead loadings from the passed
clearance visits.
D2-3. Objective 3: Characterization of the Correlation Between Components Sampled in
the Same Work Area
The relationships in dust-lead loadings among floor, window sill, and window trough
wipe samples collected from within the same room are another important aspect of examining a
clearance testing program. By estimating linear correlation coefficients, the strength of these
relationships can be assessed. Table D2-3 displays the Pearson product-moment correlation
coefficients and associated sample sizes for the log-transformed dust-lead loading measurements
of individual floor, window sill, and window trough samples taken within the same room. For
the first site visit, dust-lead loadings for floor samples were significantly positively correlated at
the 0.01 level with loadings for both window sill and window trough samples, as was the
correlation between window sills and window troughs. The correlations from the second site
visit were again significantly positive for the relationships between floor and sill dust-lead
loadings (at the 0.01 level) and between floor and trough dust-lead loadings (at the 0.05 level).
Data was not sufficient to estimate the correlation between window sill and window trough dust-
lead loadings from the second site visit.
D2-6
-------
100.0
CM
< 10.0
I 1JH
0.1
Roots WMowSBs Wndw Trojghs Bus WMtwSfc WHowtxtfs
Rnst Visit Passed Clearance
Surface
Figure D2-1. Comparison of In-Transformed Dust-Lead Loadings from the First Site Visit
vs. Passed Clearance Results on an Expanded Scale.
D2-7
-------
TOO
CM
< 500
3
g 400
300
200
100
0
Floors
Window Sills
Surface
Window Troughs
Figure D2-2. Box and Whisker Plots of the Distribution of Dust-Lead Loadings from the
Passed Clearance Data.
D2-8
-------
10000
1000
100
10
Wood Vnyl Alumiun
Substrate
Other
Unknown
Figure D2-3. Box and Whisker Plots of the Distribution of Floor Dust-Lead Loadings by
Substrate for the First Site Visit.
D2-9
-------
100000
10000
1000
D
C
.Q
a.
100
10
Ai
Wood
Vnyl Aljnjnum Other Unknown
Figure D2-4. Box and Whisker Plots of the Distribution of Window Sill Dust-Lead Loadings
by Substrate for the First Site Visit.
D2-10
-------
1000000
100000
CM 10000
1000
S. 100
10
Wood
Vinyl
Aluminum
Other
Unknown
Figure D2-5. Box and Whisker Plots of the Distribution of Window Trough Dust-Lead
Loadings by Substrate for the First Site Visit.
D2-11
-------
1000
CM 100
o>
1
Q. 10
M
Wbod Vnyl Aluminum Other
Substrate
Unknown
Figure D2-6. Box and Whisker Plots of the Distribution of Floor Dust-Lead Loadings by
Substrate for the Passed Clearance Visits.
D2-12
-------
1000
CM 100
10
Al
Wood
Vinyl
Aluminum
Other
Unknown
Figure D2-7. Box and Whisker Plots of the Distribution of Window Sill Dust-Lead Loadings
by Substrate for the Passed Clearance Visits.
D2-13
-------
1000
CM 100
10
Al
Wood Vhyl Aluminum Other
Substrate
Unknown
Figure D2-8. Box and Whisker Plots of the Distribution of Window Trough Dust-Lead
Loadings by Substrate for the Passed Clearance Visits.
D2-14
-------
Table D2-3. Observed Within-Room Correlation Coefficients Between Dust-Lead Loading
Measurements Collected From Floors, Window Sills and Window Troughs
from the Grantees with Lower Floor Dust-lead Clearance Standard in the HUD
Grantee Program.
733S;vfl*SLBlft;
First
0.48'
1687
0.40'
809
0.59a
20
Second
0.20"
21
0.53"
16
' Statistically significantly different from zero at the 0.01 level.
" Statistically significantly different from zero at the 0.05 level.
The conditional probabilities of a sample passing or failing a standard are given in Table D2-4.
These analyses were conducted on two different sets of data, the first set using all possible paired
observations from within the same room. The second subset of data restricted the
analyses to rooms in which floor, window sill, and window trough lead loadings were
simultaneously observed.
As shown in Table D2-4, results from contingency table estimates and normal theory
estimates are roughly consistent. For both types of estimates and both data sets included in the
analysis, the probability that one component passes clearance given that another component
passes clearance ranges from 80% to 97%. The probability that a floor sample is less than
200 ug/ft2, given the samples from the window sills are less than 500 ng/ft2, is the highest at
97.4%.
For both types of estimates and both data sets included in the analysis, the probability that
one component fails clearance testing given that another component in the same room fails
clearance testing ranged from 13% to 67%.
D2-4. Objective 4: Demonstration of the Impact of Composite Sampling on Pass/Fail
Rates of Houses
Table D2-5 provides, for each component type, the number of residential units having a
given number of samples collected from a given component type. Across component types, most
of the residential units had fewer than four clearance samples collected. Data for residential units
having more than four samples from a given component type were used to construct multiple
D2-15
-------
~
i
i
IS
o
z
•o
(0
0
S
o
§
o>
•*3
O
0
X
CM
O)
"55
CO
IU
O)
I
8
c s
S .2
IS
o S
CA ^
e IE
1 a)
o
O
Q
JB
JO
D2-16
-------
Table D2-5. Number of Residential Units that Contained (N) Individual Samples of Each
Component Type Based on First Site Visit Clearance Testing Data from the
Grantees with Lower Floor Dust-lead Clearance Standard in the HUD Grantee
Program.
^IridividuaHSampIes?
"•-1fft1Pjf-*£&-Jt>~,*Ui!:'*fl-'~
•v--i^T--ii»:»w»T<"-^:Tri^' •- ^i^*5 I^Vi'llKf**-.":'™
|tornpj^n^T^ej;$r~
129
120
495
124
587
302
159
214
60
346
40
13
201
6
50
8
9 +
Total
1021
968
873
Total with N
263
simulated composite samples. When there were four or fewer individual samples from a
component type within a housing unit, the simulated composite sample included all samples.
From Table D2-5, the number of homes with five or more samples was 263 for floors, 7 homes
for window sills, and 3 homes for window troughs. Therefore, data for approximately 26%
(263/1,021), 1% (7/968), and less than 1% (3/873) of the residential units were used when
constructing multiple simulated composite samples from floors, window sills, and window
troughs, respectively.
For each component type within a residential unit, the set of individual clearance sample
lead loading results for the first site visit were used to construct simulated composite samples for
the purpose of evaluating the three composite sample clearance criteria introduced in Section
5.4.2 (Standard, Standard/n, and 2*Standard/n). The construction of simulated composite
samples is discussed in Section 5.4.1. For each combination of component type and composite
clearance criterion, each residential unit either passed clearance, failed clearance, or yielded
inconclusive results, according to whether the sets of simulated composite samples for the unit
D2-17
-------
either all resulted in a pass decision, all resulted in a fail decision, or had some combination of
pass and fail decisions, respectively.
Table D2-6. Numbers of Residential Units that Pass or Fail Clearance, Based on Individual
Sample Clearance Results vs. Simulated Composite Clearance Results, Using
Data from Grantees with Lower Floor Dust-lead Clearance Standard.
lndividual;Sampl9iClearance;Resu
Pass
Standard
Inconclusive
Fail
909
39
897
34
18
55
34
772
20
81
Pass
Standard/n
Inconclusive
Fail
858
883
14
37
112
14
70
757
14
101
Pass
2xStandard/n
Inconclusive
Fail
896
892
13
92
65
762
10
0
101
Values of the four performance characteristics defined in Section 5.4.3 (sensitivity,
specificity, positive predictive value, negative predictive value) are presented in Table D2-7 for
each combination of component type and composite clearance criterion. By design, the
sensitivity for the Standard/n Rule is always 1.00 (as all sets of simulated composite samples
would fail if at least one individual sample result failed) while the specificity for this rule is
estimated at 0.94 for floors and 0.98 for both window sills and window troughs. In contrast, the
specificity of the Standard Rule is always 1.00 (as all sets of simulated composite samples would
pass if all individual samples passed) while the sensitivity for this rule is estimated at 0.49 for
both floors and window sills and 0.80 for window troughs. The 2 * Standard/n Rule attempts to
maximize both sensitivity and specificity. For the 2* Standard/n Rule, the values of sensitivity
are higher than those calculated for the Standard Rule, while the values of specificity are higher
than those calculated for the Standard/n Rule. Estimates of sensitivity and specificity in these
D2-18
-------
•examples are always conservative, because inconclusive composite test results factor into the
denominator for each estimate, but never factor into the numerator.
Table D2-7. Performance Characteristics of Composite Clearance Criteria Based on Data
from Grantees with Lower Floor Dust-lead Clearance Standard.
Sensitivity
Floors
Specificity
PPV
NPV
0.49
1.00
1.00
0.96
1.00
0.94
0.75
1.00
0.82
0.99
0.91
0.99
Sensitivity
Window Sills
Specificity
PPV
NPV
0.49
1.00
1.00
0.96
1.00
0.98
0.83
1.00
0.93
0.99
0.93
1.00
Sensitivity
Window Troughs
Specificity
PPV
NPV
0.80
1.00
1.00
0.97
1.00
0.98
0.88
1.00
1.00
0.99
0.91
1.00
The above performance characteristics estimates illustrate that the three composite
clearance testing criteria have different specificity and sensitivity rates. These rates correspond
to the consistency between clearance decisions and the true lead hazards present in the various
locations sampled (assuming the individual sample lead-loading results are representative of
these lead hazards). To further characterize the performance of each of the three composite
clearance criteria, the following logistic regression model was fitted to clearance data for each
combination of component type and composite clearance criterion:
D2-19
-------
where n^ is the estimated probability of clearance for component(/) in house(f) under composite
criterion (k), and Maxjy is the maximum individual sample lead loading result in house(i) for
component(/).
For each component type, the estimated relationship between the probability of passing
clearance and the maximum lead-loading is presented graphically in Figure D2-9. In this figure,
the solid, long-dashed, and finely-dashed lines represent the estimated relationship for the
Standard, Standard/n, and 2* Standard/n Rules, respectively.
For each combination of component type and composite clearance criterion, Table D2-8
provides parameter estimates and associated standard errors from fitting the above logistic
regression model to data. Table D2-8 also presents estimates of the probability of passing
clearance (using composite samples) when the maximum lead loading among all locations
included in the composite sampling scheme is greater than or equal to Vi, 1,2 and 4 times the
associated HUD interim standard for individual samples. Results from these logistic regression
model fits show that there is a better than 55% chance that a residential unit will pass clearance
for floors and window sills under the Standard Rule for composite sampling, when there is an
individual sample location which has a lead-loading level that is equal to twice the HUD
clearance standard. Estimates for the Standard/n Rule demonstrate that this rule's high sensitivity
(probability of passing is below 19% when the maximum individual sample lead loading is
greater than or equal to the HUD Standard) along with a loss in specificity (probability of passing
is 85% for floors, 96% for window sills, and 99% for window trough when the maximum
individual sample lead loading is equal to V* HUD Standard). Once again, the 2xStandard/n
Rule is shown to be a compromise between the Standard and Standard/n Rules. At 1A the HUD
Standard, the estimated probability of passing clearance testing under the 2xStandard/n Rule is
one or nearly one, while at 2>
-------
ProbobBV of Passha Clearance Ten**) LMng Composte Floor Samples
10O
20O 300 400 600 60D
Maxknum kvMdual Sample ROOT Pb Loading
TOO
BOO
U*Q CQRfXMte Widow SB Samptee
250
500 750 1000 1250 1500
Maxknum MMdue) Swnpto WMow sn Pb Loadk«
1750
aooo
CtBoronov ^stkio UvhQ CornpovftB Wtidow VMsM
Approach
men) I
/ n) R
Ruto
400 flOO lann TGDO 2000 24OO
MBxknum MMcLiel Sample WMow WM Pb
2800
3200
Figure D2-9. Estimated Relationship Between the Probability of a Residential Unit Passing
Clearance Testing versus the Maximum Individual Lead-Loading Results by
Component Type Based on Simulated Composite Samples from the Grantees
with Lower Floor Dust-lead Clearance Standard in the HUD Grantee Program.
D2-21
-------
= j? 5>
w O OT
« ^ o
* £ «
0.3= J>
Z Q O
c g-°
o £ g
•-c f. £
£ »- *•
® i- 2
"* §
£ o as
«««
2 a S
g E 2
_ (0 c
s > *
-ll E
8-2
2 e «^
••s « ro
.2 g c
'
c e .S o>
I s«l
I III
tt o Q a
CO
CNJ
§
o
o
00
(O
CO
o
o
CD
0)
d
(D
CO
00
d
00
CN
8
oi
03
«™
en
CO
CO
en
cd
CM
o>
CD
•<*
in
CO
Ps
in
IX
a>
CO
in
cd
o>
in
in
o
o
CO
CO
in
CO
CO
CO
o
o>
oo
in
oo
ro
0)
(D
en
03
to
CO
CO
CN
CO
r-
in
ID
T3
re
•a
I
CO
•s
CD
(0
(A
nj
T)
X
CM
•o
CO
CO
X
CN
^
•S
co
•a
I
c/)
CA
O
•a
D2-22
o
•o
-------
APPENDIX E
Atlantic City
Housing Authority
E-1
-------
APPENDIX E
ATLANTIC CITY
HOUSING AUTHORITY
The Atlantic City Housing Authority provided data from a comprehensive rehabilitation
project performed on public housing buildings containing 10 to 23 multi-family housing units
[18]. These two- to three-story brick buildings had lead-based paint on the doors, windows,
radiators, trim, and stairwells. The doors and windows, along with most of the trim, were
removed and replaced during abatement. Walls were enclosed by drywall, then painted, and steel
panels were placed on the walls in the stairwells. Atlantic City Housing Authority employed a
two-phased lead clearance process: worker entry clearance and re-occupancy clearance. The first
set of lead clearance samples was collected after protected workers had finished the abatement
job (worker lead clearance testing). The second set of lead clearance samples was taken after the
renovation job was completed, but before the unit was occupied (re-occupancy clearance).
Worker lead clearance testing results were used to ensure that the housing unit was safe for
renovation workers (e.g. carpenters) to enter to complete their work. Re-occupancy clearance
testing results were collected after all work on the housing unit was completed (including
abatement and renovation and remodeling) to ensure that the unit was not contaminated by
leaded dust and could be re-occupied by residents. This report only analyzed data collected in
the re-occupancy clearance testing phase from the Atlantic City Housing Authority project.
Wipe samples were collected in accordance with HUD protocol. Within each completed unit,
one dust sample was collected from each room or area where abatement occurred. Sample
locations were randomly distributed between floors and window troughs. Since the abatement
process included removal of many windows, few window sill samples were tested. The standard
interim HUD-established thresholds of 200 ug/ft2 for floors, 500 jig/ft2 for window sills, and
800 ug/ft2 for window troughs were utilized in determining whether dust wipe samples passed or
failed. In all cases where results exceeded the clearance thresholds, the areas were re-cleaned and
re-tested until acceptable results were obtained.
E-2
-------
E-1. Objective1: Characterization of the Number of Individual Samples. Work Areas,
and Housing Units That Pass or Fail Clearance Testing Standards
Individual dust wipe samples were collected in Atlantic City from June 1994 through
May 1995 as part of their clearance testing program. In all, 923 individual dust wipe samples
were collected from floors, window sills, and window troughs within 779 rooms in 160
residential units. Table E-1 presents the number of individual samples, work areas, and
Table E-1. Clearance Testing Results by Individual Sample, Room, and Residential Unit for
Atlantic City.
s&w
-
Sam
feTotal
First
Floor
Sill
Trough
All
516
38
554
479
37
516
127
46
51
46
51
252
258
248
254
114
814
49
863
714
48
762
121
32
39
159
20
119
160
Second
Floor
Trough
All
29
13
42
25
12
37
12
14
17
14
17
43
16
59
31
15
46
13
16
11
20
Third
Trough
All
Total
Floor
Sill
Trough
All
545
46
267
858
51
5
9
65
596
51
276
923
502
46
260
733
37
5
6
46
539
51
266
779
136
16
118
133
23
4
4
27
159
20
122
160
residential units that passed or failed clearance testing within each combination of component
type and site visit. Approximately 93% (863/923) of the dust samples were collected during the
first site visit to a residential unit. While 94% (814/863) of the dust samples fell below the
clearance standards of 200 ng/ft2 for floors, 500 ug/ft2 for window sills, and 800 n.g/ft2 for
window troughs, and 94% (714/762) of the rooms met clearance standards, only 76% (121/160)
of the residential units passed clearance on the first site visit. This increase in the failure rate
from the percentage of individual samples and rooms that fail to the percentage of residential
units that fail is attributable to the fact that all individual samples must pass for a unit to pass.
E-3
-------
That is, if any individual sample exceeds the standard and fails clearance, then the entire
residential unit also fails clearance. Of the 39 residential units that failed clearance on the first
site visit, 20 (51%) of these residential units are known to have been revisited for a second
clearance testing site visit. Eventually, 133 of the 160 residential units (83%) are known to have
passed clearance testing.
Floor samples accounted for 64% (554/863) of the first site visit samples while window
troughs made up 30% (258/863) of the samples and window sills comprised the final 6%
(51/863) of the first site visit samples. The failure rate for individual samples during the first site
visit was highest for window sill samples (10% (5/51)) followed by floors (7% (38/554)) and
window troughs (2% (6/258)).
E-2. Objective 2: Characterization of the Distribution of the Dust-Lead Loadings.
Geometric Mean Dust-Lead Loadings. Variability Within a Housing Unit, and
Variability Between Housing Units
As seen in Table E-2, the geometric means for floor and window trough samples were
18.5 ug/ft2 and 19.4 ug/ft2, respectively. The within-house variance components and the
between-house variance components are similar to those estimated for the other data sources.
Figures E-l and E-2 contain box and whisker plots that present the distribution of dust-
lead loadings from the first and passed clearance visits by component type. Figures E-3 to E-5
contain box and whisker plots that present the distribution of dust-lead loadings from the first
visit for floor, window sill, and window trough, respectively; Figures E-6 to E-8 contain box and
whisker plots that present the distribution of dust-lead loadings from the passed clearance visits.
E-3. Objective 3: Characterization of the Correlation Between Components Sampled in
the Same Work Area
The relationships among floor, window sill, and window trough wipe samples are
another important aspect of examining a clearance testing program. One method to assess the
relationships among individual floor, window sill, and window trough wipe samples collected
from the same room is to estimate linear correlation coefficients. Table E-3 displays the Pearson
product-moment correlation coefficients and the associated sample size for the log lead loading
E-4
-------
c
o
o
Q.
E
o
o
•a
a
**
w
2
35
JO
o
**
0
UJ
0
o
a
I
i
ja
a
c **
m V
O W
oO
8*
O
Ul
O
2
03
ID
O
o
U)
If)
01
oo
cq
ai
o>
CM
ao
CM
CM
00
C3
CO
CO
CO
03
CM
O
tf)
CO
CM
CO
ID
O3
Ul
CO
in
10
CM
ro
u>
10
It)
00
in
cs
o
o
a
§
o
o
o
CO
•g
£
E-5
-------
100.0
< 10.0
?
D
&
1.0
0.1
Ftas
WndwSfc
RtstVtsft
Whto* Troupe
Rons Window Sfe Wndw loupes
Passed Qearance
Figure E-1. Comparison of In-Transformed Dust-Lead Loadings from the First Site Visit vs.
Passed Clearance Results on an Expanded Scale.
E-6
-------
800
700
600
N
<. 500
3
0400
1
^300
CL
200
100
0
I
i i i
Boors Window Sills Window Troughs
Surface
Figure E-2. Box and Whisker Plots of the Distribution of Dust-Lead Loadings from the
Passed Clearance Data.
E-7
-------
10000
1000
100
10
All
Wood
Vryl
Aluminum
Other
Unknown
Figure E-3. Box and Whisker Plots of the Distribution of Floor Dust-Lead Loadings by
Substrate for the First Site Visit.
E-8
-------
100000
10000
CM
<
1000
0)
1
e
100
10
All
Wood Vinyl Aluminum Other Unknown
Substrate
Figure E-4. Box and Whisker Plots of the Distribution of Window Sill Dust-Lead Loadings
by Substrate for the First Site Visit.
E-9
-------
1000000
100000
w 10000
3
I 1000
£ 100
10
All
Wood
Vinyl
ALrmim
Other
Unknown
Figure E-5. Box and Whisker Plots of the Distribution of Window Trough Dust-Lead
Loadings by Substrate for the First Site Visit.
E-10
-------
1000
CM 100
10
Al
Wood Vhyl Aluntun
Substrate
Otter
Unknown
Figure E-6. Box and Whisker Plots of the Distribution of Floor Dust-Lead Loadings by
Substrate for the Passed Clearance Visits.
E-11
-------
1000
N 100
c
•o
10
Wood Vinyl Aluminum Other Unknown
Substrate
Figure E-7. Box and Whisker Plots of the Distribution of Window Sill Dust-Lead Loadings
by Substrate for the Passed Clearance Visits.
E-12
-------
1000
CM 100
10
All
Wood Vinyl Akmnurn Other Unknown
Substrate
Figure E-8. Box and Whisker Plots of the Distribution of Window Trough Dust-Lead
Loadings by Substrate for the Passed Clearance Visits.
E-13
-------
Table E-3. Observed Within-Room Correlation Coefficients between Pb Loading
Measurements Collected from Floors, Window Sills and Troughs for the
Atlantic City Data.
First
Second
'sw)fStsSS<^^se^fni'>t>^itafa9lS>»t9a)ia/KS»*---t^ ••' ".-.-v.c.v.f;. ---.;,•..;.• a*<**K!yitfaiQ:iini*iiivv*iiHttt
0.32
26
0.65"
0.10
33
8 Statistically significantly different from zero at the 0.01 level.
measurements of individual floor, window sill, and window trough samples taken within the
same room. The correlation between window sill samples and window trough samples was not
able to be estimated due to insufficient data. In fact, for this data set, there were few cases where
multiple samples were taken per room. The data from the first site visit show that correlation
between floors and troughs is positive and significant at the 0.01 level. The correlation between
floors and sills was estimated to be positive but was not statistically significantly different from
zero. The correlation from the second site visit is difficult to interpret because of the small
number of samples from which it was estimated.
The conditional probabilities of a sample passing or failing a standard are given in
Table E-4. These analyses were intended to conduct on two different sets of data, the first set
using all possible paired observations from within the same room. The second subset of data
restricted the analyses to rooms in which floor, window sill, and window trough lead loadings
were simultaneously observed. However, no such results were observed in the Atlantic City data
for the second subset of data. Window sill and window trough samples were not simultaneously
present in rooms in this data set.
As shown in Table E-4, results from contingency table estimates and normal theory
estimates are consistent for the left side of the table, but less so for the right side. For both types
of estimates, the probability that one component passes clearance given that another component
passes clearance is quite high (over 88%). The probability that one component fails clearance
testing given that another component in the same room fails clearance testing ranged from 0% to
51%.
E-14
-------
JQ
10
JQ
I
•o
CO
W
o
O>
V
o
o
CM
X
eg
f
(0
T3
0
+*
a
I
ILI
O>
V
O
O
c *r
2 .2
II
O JC
O I-
t
UJ
Q)
..I
f
.
§
8
o
CO
V
>-_
8
o
o
rg
V
-
a.
E-15
-------
E-4. Objective 4: Demonstration of the Impact of Composite Sampling on Pass/Fail Rates
of Houses
Simulated composite sample results were constructed from individual sample lead
loading results from each component type within a residential unit. These simulated composite
samples were then used to evaluate the three composite sample clearance criteria (Standard,
Standard/n, and 2xStandard/n).
Table E-5 displays the number of residential units that were tested by the number of
individual samples that were collected and each component type. For example, there were 57
residential units for which two dust-wipe samples were collected from floors on the first site
visit. Residential units containing more than four samples from a particular component type
resulted in the estimation of multiple simulated composite sample results in this analysis.
Therefore, summing all the units that had five or more individual samples, approximately 25%
(40/159), 5% (1/20), and 1% (1/119) of the residential units resulted in the estimation of
multiple simulated composite samples from floors, window sills, and window troughs,
respectively. When there were four or fewer individual samples from a component type within a
housing unit, the simulated composite sample included all samples.
For each component type within a residential unit, the set of individual clearance sample
lead loading results were used to construct simulated composite samples for the purpose of
evaluating the three composite sample clearance criteria For each combination of component
type and composite clearance criterion, Table E-6 gives the number of residential units that either
passed clearance, failed clearance, or yielded inconclusive results based on the composite
clearance samples. Inconclusive results could only occur in a residential unit if more than four
individual samples were collected from that unit. They occur due to the many possible ways of
combining five (or more) individual samples into composite samples. For floor samples, the
percentages of residential units with greater than four individual clearance samples that actually
resulted in inconclusive results were 15% (6/40), 7.5% (3/40), and 17.5% (7/40) for the Standard,
Standard/n, and 2*Standard/n Rules, respectively. No window sill or window trough composite
sample yielded inconclusive results for any of the three criteria, and, of course, for each of these .
components, there was only on unit with more than four individual clearance samples.
E-16
-------
Table E-5. Number of Residential Units that Contained (N) Individual Samples of Each
Component Type Based on First Site Visit Clearance Testing Data From the
Atlantic City Housing Authority.
9 +
Total
Total with Ma 5
57
33
26
23
10
159
40
20
^Window
•- .....
ughs
14
79
19
119
Table E-6. Individual Sample Clearance Results Versus Simulated Composite Clearance
Results Based on Data from the Atlantic City Housing Authority.
Composites^.
^^?¥^&^^s^*^SR«as
Individual Sample ^Clearancif
esu
Pass
Standard
Inconclusive
Fail
127
16
16
10
114
Pass
Standard/n
Inconclusive
Fail
119
15
32
109
Pass
2 x Standard/n
Inconclusive
Fail
126
15
22
111
E-17
-------
The performance characteristics of each combination of component type and composite
clearance criteria are presented in Table E-7 in terms of sensitivity, specificity, positive
Table E-7. Performance Characteristics of Composite Clearance Criteria Based on Data
from the Atlantic City Housing Authority.
Floors
Sensitivity
Specificity
PPV
NPV
0.31
1.00
1.00
0.89
1.00
0.94
0.87
1.00
0.69
0.99
1.00
0.97
Window
Sills
Sensitivity
Specificity
PPV
NPV
0.25
1.00
1.00
0.84
1.00
0.94
0.80
1.00
0.50
0.94
0.67
0.88
Window
Troughs
Sensitivity
Specificity
PPV
NPV
0.80
1.00
1.00
0.99
1.00
0.96
0.50
1.00
1.00
0.97
0.63
1.00
predictive value (PPV), and negative predictive value (NPV). By design, the specificity of the
Standard Rule is always 1.00 and the sensitivity of the Standard/n Rule is 1.00. So, in some
sense, the Standard Rule sacrifices sensitivity for specificity while the Standard/n Rule sacrifices
specificity for sensitivity. The 2 * Standard/n Rule attempts to find a balance between these two
criteria. For this data, the Standard Rule's sensitivity is estimated to be 0.31 for floors, 0.25 for
window sills, and 0.80 for window troughs. The 2*Standard/n Rule seems to perform much
better in terms of sensitivity with estimates of 0.69, 0.50, and 1.00 for floors, window sills
and window troughs, respectively. The estimates of sensitivities for window sills and window
troughs, however, are based on very few units which failed individual sample clearance testing.
The 2*Standard/n Rule seems to perform quite well in terms of specificity with estimates of 0.99
for floors, 0.94 for window sills, and 0.97 for window troughs. When comparing the overall
performance of the three criteria in this case, the Standard/n Rule seems to perform quite well
both in terms of sensitivity and specificity.
E-18
-------
It is evident that all three composite clearance testing criteria have different sensitivities
and specificities associated with their application to the simulated composite samples. In order
to further investigate the performance of each of the three composite clearance criteria, the
following logistic regression model was fitted for each combination of component type and
composite clearance criterion to describe the relationship between the probability of passing
clearance based on composite samples and the maximum lead loading present in the individual
samples collected:
where nijk is the estimated probability of clearance for component(/) in house(i) under composite
criterion (&), and Max,-,- is the maximum individual sample lead loading result in house(i') for
component(/).
For each component type, the estimated relationship between the probability of passing
clearance and the maximum lead loading is presented in Figure E-9. Note that the relationship
could not be estimated for window troughs using the Standard Rule due to the nature of the data.
Under this rule, the observed data were such that at a given level the unit passed, and at a given
level the sample failed. Estimation problems arise when trying to estimate the relationship
between these two levels. An infinite number of curves could be fit to estimate the relationship
between the two levels, but there is no criterion to choose the optimal one. Therefore, the
relationship cannot be estimated for this rule.
Table £-8 provides parameter estimates and associated standard errors from the logistic
regression models, as well as estimates of the probability of passing clearance (using composite
samples) when the maximum lead loading among all locations included in the composite
sampling scheme is greater than or equal to 1A, 1 , 2, and 4 times the associated interim HUD
standard for individual samples. The low sensitivity of the Standard Rule is illustrated by the
high (73%) estimated probability of a unit passing clearance testing based on floor composite
samples and a maximum individual sample lead-loading of 2*HUD standard. Conversely, the
estimated probability of passing based on floor composite samples using the 2>
-------
to probabilities of 0.98 and 1.00 for the other two decision rules. The estimates for window
troughs at one-half the HUD standard are similar to those for floors, but the estimates of the
probabilities of passing with a maximum individual sample lead loading at the HUD standard are
much lower for the 2>
-------
Probability of Passing Clearance Testing Using Composite Floor Samples
(2 x Standard /n)RJo
(Standard /n)Rule
100
200 300 400 500 600
Maximum Individual Sample Floor Pb Loading
800
Probability of Passing Clearance Testing Using Composite Window SiD Samples
Approach
StanttordRuto
(2 x Standard/n)Riilo
(Standard/n) Rule
250 500 750 1000 1250 1500 1750 2000
Maximum Individual Sample Window SRI Pb Loading
ProbabHHy of Passing Clearance Testing Using Composite Window WeO Samples
Approach
Standard Rule
(2 x Standard/n}Rute
(Standard/n) Rote
400
BOO
1200 1600 2000 2400 2BOO 3200
Maximum Individual Sample Window Well Pb Loading
Figure E-9. Estimated Relationship Between the Probability of a Residential Unit Passing
Clearance Testing versus the Maximum Individual Lead-Loading Result by
Component Type Based on Simulated Composite Samples from the Atlantic
City Housing Authority.
E-21
-------
>*• O)
o c
II
_Q (0
« w
•8 »
s «
a g
i!
3 c
>
«
C
o c
* .£
o a> O
IE "a o
3 £f
— * i
•SfS
O (D <
•-g E
S - 2
2 E *
a> E Q
o 3 «S
oc E s
o '5 Q
111
.
••s S
i S.S
i;?
CO C o
** ffi »™
3 ™ .2
S — 9
EC O Q
•
CO
111
a
O
O
6
o
o
o
o
o
o
o
o
o
6
o
o
o
o
6
8
0
o
q
d
oo
o>
in
q
d
in
a>
in
d
(D
^*
d
q
d
8
00
a>
oi
OD
(O
03
O)
d
0>
CO
S
o
q
d
01
o
o
n
o
o
PS.
o
o
S
q
d
o
q
d
q
9
00
PJ
o
o
u>
o
CO
o
o
(0
o
o
q
d
(O
q
d
§
CO
8
EE
=
to
I
5
It
E-22
-------
APPENDIX F
Cleveland Lead Hazard Abatement Center
F-1
-------
APPENDIX F
CLEVELAND LEAD HAZARD ABATEMENT CENTER
The lead abatement program managed by the Cleveland Lead Hazard Abatement Center
recruited houses where children with elevated blood lead levels lived [19]. Most of the units
were large single-family houses built before 1950. The level of intervention depended on the age
and number of children in the house. The homes had a mix of interim controls and abatement
treatments. Chipped paint was scraped and repainted with a primer and a coat of paint. Carpets
were removed and the floors were either repaired, covered with plywood, or refinished.
Windows were replaced in some units. Vinyl siding was applied to homes with lead-based
painted wood siding. Porches were repaired, deteriorated sections replaced, and other surfaces
scraped and repainted. Lead contaminated soil was generally covered with sod, wood chips, or
some other form of landscape cover; if lead contamination was extremely high, soil was removed
and replaced. This project is ongoing. Results for window trough, window sill, and floor wipe
samples collected from each room in the unit were analyzed.
F-1. Objective 1: Characterization of the Number of Individual Samples. Work Areas, and
Housing Units That Pass or Fail Clearance Testing Standards
As part of clearance testing within the Cleveland Lead Hazard Abatement Center's
intervention program, 312 individual dust wipe samples were collected from floors, window sills,
and window troughs within 196 rooms in 38 residential units from December 1993 through May
1995. Table F-1 presents the number of individual samples, work areas, and residential units that
passed or failed clearance testing within each combination of component type and site visit.
Approximately 93% (290/312) of the dust samples were collected during the first site visit to a
residential unit. Although 92% (267/290) of the dust samples fell below the clearance standards
of 200 ng/fi2 for floors, 500 ug/ft2 for window sills, and 800 ug/ft2 for window troughs, and 88%
(173/196) of the rooms passed clearance on the first site visit, only 61% (23/38) of the residential
units passed clearance on the first site visit. This increase in the failure rate from the percentage
of individual samples and rooms that fail to the percentage of residential units that fail is
F-2
-------
Table F-1. Clearance Testing Results by Individual Sample, Room, and Residential Unit for
the Cleveland Lead Hazard Abatement Center.
XBffi
First
Floor
Sill
Trough
All
138
20
158
137
20
157
24
92
93
92
93
32
37
39
37
39
22
267
23
290
173
23
196
23
14
15
38
33
24
38
Second
Floor
Sill
Trough
All
15
18
15
18
10
18
21
18
21
11
13
14
Third
Floor
All
Total
Floor
Sill
Trough
All
154
93
39
286
23
1
2
26
177
94
41
312
153
93
39
192
4
0
0
4
157
93
39
196
35
33
24
35
3
0
0
3
38
33
24
38
attributable to the fact that if any individual sample exceeds the standard and fails clearance, then
the residential unit also fails clearance. Of the 15 residential units that failed clearance on the
first site visit, 14 (93%) of these residential units were revisited for a second clearance testing.
Seventy-nine percent (11/14) of the units passed clearance on the second site visit. Of the three
units that did not pass clearance on the second visit, one passed clearance after a third visit.
Thus, at the time the data for this report were collected, 35 of the 38 residential units (92%) had
passed clearance testing.
The failure rate for individual samples during the first site visit was higher for floor
samples (13% (20/158)) than that for window troughs (5% (2/39)) or window sills (1% (1/93)).
This pattern is reflected in the failure rates for residential units based on results from individual
components: 37% (14/38) of the residential units failed based on the results of floor samples, 8%
F-3
-------
(2/24) of the residential units failed based on the results of window trough samples, and 3%
(1/33) of the residential units failed based on the results of window sill samples.
F-2. Objective 2: Characterization of the Distribution of the Dust-Lead Loadings.
Geometric Mean Dust-Lead Loadings, Variability Within a Housing Unit, and
Variability Between Housing Units
The Cleveland program had the fewest number of residential units tested for clearance.
The number of samples collected during the second site visit was very small. The geometric
mean for the floor clearance samples was less than the window sills which was less than the
window troughs for the first site visit. However, they were not statistically different from one
another. For the first site visit, the estimated within components of variances are greater than the
between components for floors and window sills; the opposite is true for window troughs.
Figures F-l and F-2 contain box and whisker plots that present the distribution of dust-
lead loadings from the first and passed clearance visits by component type. Figures F-3 to F-5
contain box and whisker plots that present the distribution of dust-lead loadings from the first
visit for floor, window sill, and window trough, respectively; Figures F-6 to F-8 contain box and
whisker plots that present the distribution of dust-lead loadings from the passed clearance visits.
F-3. Objective 3: Characterization of the Correlation Between Components Sampled in
the Same Work Area
The relationships among floor, window sill, and window trough wipe samples are another
important aspect of examining a clearance testing program. One method of assessing the
relationships among individual floor, window sill, and window trough wipe samples collected
from the same room is to estimate linear correlation coefficients. Table F-3 displays the Pearson
product-moment correlation coefficients and the associated sample size for the log lead loading
measurements of individual floor, window sill, and window trough samples taken within the
same room. The correlation between window sills and window troughs is the highest of any
observed correlation but it is not statistically significant. None of the observed correlations are
statistically significantly different from zero at the 0.05 level.
F-4
-------
o
a
c
u>
o
Q.
E
o
u
«
*<
w
S
UJ
o
o
Q.
O
U
8
(0
•o
g
i
O 0
Ew
O (0
0 +*
oi
H >
SO
i£
o **
O k.
rsl
uL
.••jJO' C'-vg*
- '
til
•pit
-•->'-'i;r!*'Vi..5'ls"«
CO
0>
CO
<0
CO
in
en
oq
ifi
in
CO
o
o
CO
CO
CO
o
to
CO
in
*t
O)
CM
<0
o
0)
CM
CD
CO
O)
01
CO
00
co
05
m
en
co
eo
CM
O
o
TJ
O
U
F-5
-------
100.0
sr
< 10.0
I to
0.1
Hans WndowSfe Window Troughs
Fust Visit
Rows Window Sfc Window Houghs
Passed Clearance
Surface
Figure F-1. Comparison of In-Transformed Dust-Lead Loadings from the First Site Visit
vs. Passed Clearance Results on an Expanded Scale.
F-6
-------
800
700
600
N
^ 500
3
"400
1
vu
^300
0.
200
100
0
i
i
ROOTS
Window Sills
Surface
Window Troughs
Figure F-2. Box and Whisker Plots of the Distribution of Dust-Lead Loadings from the
Passed Clearance Data.
F-7
-------
10000
1000
CM
<
100
£1
D.
10
AS Wood
Vty Aluminum Other Unknown
Substrate
Figure F-3. Box and Whisker Plots of the Distribution of Floor Dust-Lead Loadings by
Substrate for the First Site Visit.
F-8
-------
100000
10000
CM
<
0>
1000
100
a
CL
10
1
fl Vtad Vty Ahmhum Other Unknown
Substrate
Figure F-4. Box and Whisker Plots of the Distribution of Window Sill Dust-Lead Loadings
by Substrate for the First Site Visit.
F-9
U.S. EPA Headquarters Library
Mail code 3201
1200 Pennsylvania Avenue NW
Washington DC 20460
-------
1000000
100000
cvi 10000
1000
S. 100
10
Wood Vnyl ALmwi
Substrate
Other Unknown
Figure F-5. Box and Whisker Plots of the Distribution of Window Trough Dust-Lead
Loadings by Substrate for the First Site Visit.
F-10
-------
1000
w 100
CL io
Wood
Vhyl
Aluminum
Other
Unknown
Figure F-6. Box and Whisker Plots of the Distribution of Floor Dust-Lead Loadings by
Substrate for the Passed Clearance Visits.
F-11
-------
1000
CM 100
^
10
All
Wood
Aluminum Other Unknown
Substrate
Figure F-7. Box and Whisker Plots of the Distribution of Window Sill Dust-Lead Loadings
by Substrate for the Passed Clearance Visits.
F-12
-------
1000
CM 100
S. 10
Alt
Wood Vryl Alumirun Other
Substrate
Unknown
Figure F-8. Box and Whisker Plots of the Distribution of Window Trough Dust-Lead
Loadings by Substrate for the Passed Clearance Visits.
F-13
-------
Table F-3. Observed Within-Room Correlation Coefficients between Pb Loading
Measurements Collected from Floors, Window Sills and Troughs for the
Cleveland Data.
Site
•-.- >bbservedWithiii
Floors and Window
Pnoor.SiU»
First
0.14
64
0.30
24
0.47
13
The conditional probabilities of a sample passing or failing a standard are given in
Table F-4. These analyses were conducted on two different sets of data, the first set using all
possible paired observations from within the same room. The second subset of data restricted the
analyses to rooms in which floor, window sill, and window trough lead loadings were
simultaneously observed.
As shown in Table F-4, results from contingency table estimates and normal theory
estimates are consistent, although more so on the left side of the table than on the right. For both
types of estimates and both data sets included in the analysis, the probability that one component
passes clearance given that another component passes clearance is quite high (over 88%), while
the probability that one component fails clearance testing given that another component in the
same room fails clearance testing is low (mostly under 26%).
F-4. Objective 4: Demonstration of the impact of Composite Sampling on Pass/Fail Rates
of Houses
Individual sample lead loading results from each component type within a residential unit
were combined to construct simulated composite sample results.
Table F-5 provides the number of residential units that were investigated by the number
of individual samples that were collected for each component type. For example, there were 6
residential units which included four window sill dust-wipe samples on the first site visit.
Residential units containing more than four samples from a component type resulted in the
estimation of multiple simulated composite sample results in this analysis. Therefore,
approximately 37% (14/38) of the residential units had five or more samples and resulted in the
F-14
-------
JO
CO
.a
s
O.
<0
w
I
0
O)
o
u
CM
X
O)
*«
•o
I
'+:
(0
Ul
o>
.£
8
«>
.2 u.
£ o
I
g S
go
S >
12 °
c v
_£
.a
a
F-15
-------
estimation of multiple simulated composite samples from floors. Conversely, only 9% (3/33) of
the units which had window sill samples collected and none of the residential units which had
window trough samples collected resulted in the estimation of multiple simulated composite
samples. When there were four or fewer individual samples from a component type within a
housing unit, the simulated composite sample included all samples.
Table F-5. Number of Residential Units that Contained (N) Individual Samples of Each
Component Type Based on First Site Visit Clearance Testing Data From the
Cleveland Lead Hazard Abatement Center.
9 +
Total
Total with N;>5
13
10
10
38
14
'
13
33
•^ti'>"™&'?- *iw^ss.^"r
14
24
For each component type within a residential unit, the set of individual clearance sample
lead loading results was used to construct simulated composite samples for the purpose of
evaluating the three composite sample clearance criteria (Standard, Standard/n, and
2xStandard/n). For each combination of component type and composite clearance criterion,
Table F-6 indicates the number of residential units that either passed clearance, failed clearance,
or yielded inconclusive results. The percentage of residential units that had more than four
individual samples for a particular component which actually resulted in inconclusive results was
F-16
-------
low (no more than 5.3% for any given combination of component type and clearance criterion)
for this data set.
Table F-6. Individual Sample Clearance Results Versus Simulated Composite Clearance
Results Based on Data from the Cleveland Lead Hazard Abatement Center.
mposite Sample
ihliiv^
'
Standard
Pass
Inconclusive
Fail
27
32
22
Standard/n
Pass
Inconclusive
Fail
20
30
11
22
2xStandard/n
Pass
Inconclusive
Fail
26
32
22
The performance characteristics of each combination of component type and composite
clearance criteria are presented in Table F-7 in terms of sensitivity, specificity, positive
predictive value (PPV), and negative predictive value (NPV). These performance characteristics
attempt to estimate the false negative and false positive error rates for the different clearance
criteria.
The Standard/n Rule is formulated such that specificity is always sacrificed for
sensitivity. The sensitivity for the Standard/n Rule is always 1.00 while the specificity for this
rule is estimated to be 0.74 for floors, 0.94 for window sills, and 1.00 for window troughs. On
the other hand, the Standard Rule is designed so that it always sacrifices sensitivity for
specificity. The specificity for the Standard Rule is always 1.00 while the sensitivity for this rule
is estimated at 0.18 for floors, 1.00 for window sills, and LOO for window troughs. For the
2 x Standard/n Rule, the values of sensitivity are higher than or equal to those calculated for the
Standard Rule, while the values of specificity are higher than or equal to those calculated for the
Standard/n Rule. Estimates of sensitivity and specificity in these examples are always
F-17
-------
Table F-7. Performance Characteristics of Composite Clearance Criteria Based on Data
from the Cleveland Lead Hazard Abatement Center.
Floors
Window
Sills
Window
Troughs
Sensitivity
Specificity
PPV
NPV
Sensitivity
Specificity
PPV
NPV
Sensitivity
Specificity
PPV
NPV
0.18
1.00
1.00
0.79
1.00
1.00
1.00
1.00
1.00
1.00
1.00
1.00
1.00
0.74
0.69
1.00
1.00
0.94
0.50
1.00
1.00
1.00
1.00
1.00
S^2x^ln^^4|-fi
0.73
0.96
1.00
0.90
1.00
1.00
1.00
1.00
1.00
1.00
1.00
1.00
conservative, because inconclusive composite test results factor into the denominator for each
inconclusive composite test results factor into the denominator for each estimate, but never factor
into the numerator.
It is evident that all three composite clearance testing criteria have different sensitivities
and specificities associated with their application to the simulated composite samples. To
evaluate the performance of each of the three composite clearance criterion in another manner,
the following logistic regression model was fitted for each combination of component type and
composite clearance criterion. The goal was to describe the relationship between the probability
of passing clearance and the maximum lead loading present in all of the individual samples
tested in a residential unit:
where nijk is the estimated probability of clearance for component(/) in house(i) under composite
criterion (&), and Max,-, is the maximum individual sample lead loading result in house(i) for
component(/).
F-18
-------
The estimated relationship between the probability of passing clearance and the
maximum lead loading for floors and window sills is presented graphically in Figure F-9. The
relationship was not able to be estimated for floors using the 2xStandard/n Rule and for window
sills using either the 2xStandard/n Rule or the Standard Rule because of the nature of the data.
The data were such that, under the 2* Standard/n Rule for floors and under the 2>
-------
Probability of Passing Clearance Testing Using Composite ROOT Samples
t
i
t
i
i
i
i
i
i
i
i
i
i
i
t
Approach
Standard Bute
(Standard / n) Rute
100 200 300 400 500 600
Maximum individual Sample Floor Pb Loading
700
800
Probability of Passing Clearance Testing Using Composite Window SOI Samples
Approach
- (Standard / n) Bute
250 500 750 1000 1250 1500
Maximum Individual Sample Window Sill Pb Loading
1750
2000
Figure F-9. Estimated Relationship Between the Probability of a Residential Unit Passing
Clearance Testing versus the Maximum Individual Lead-Loading Result by
Component Type Based on Simulated Composite Samples from Cleveland Lead
Hazard Abatement Center.
F-20
-------
**•* Q)
0 .E
>. "5.
i I
"§ w
5 c
CO
M 01
» c
•D a
O 0) "O
1 Is
* 5 »
• w |
O « O
s = «
§i25
I-S E
10 £ O
£ c^
ro E (a
o 2 5
cc £ ro
o 'x Q
n "O 2
ESl
S ««r
^ « c
ill
111
CC O O
u
«
J5
n
CM
m
o
s
CO
8
(D
CM
o
^
u.
O
O
O
o
CO
C)
CD
co
d
en
o
o
in
o
9
in
IN
10
(N
es
tv
h.
ID
CM
CM
CM
CM
CM
CM
CM
CO
•o
ID
TO
•a
-------
-------
APPENDIX G
Dover Housing Authority,
New Hampshire
G-1
-------
APPENDIX G
DOVER HOUSING AUTHORITY,
NEW HAMPSHIRE
One-hundred and eighty-four units in 49 multi-family buildings were abated and
renovated [20]. Forty-three of the buildings had four units per building and the other six
buildings had two units per building. The exterior siding and window sashes of the buildings
contained lead-based paint, so siding and windows including the sashes were removed and
replaced. Interior lead-based paint was found only on water radiators. Thirty-one of the units
had cast iron radiators. The radiators were removed from the apartments, sand blasted and
repainted offsite, and reinstalled. Even though there was very limited abatement activity inside
the units, dust wipe samples from the window troughs and sills along with floor samples were
analyzed for lead clearance.
G-1. Objective 1: Characterization of the Number of Individual Samples. Work Areas, and
Housing Units That Pass or Fail Clearance Testing Standards
The Dover Housing Authority tested all 184 units for clearance. Clearance data were
made available for this analysis from 704 rooms in 158 of the 184 residential units that were
abated and renovated. Nine-hundred and sixteen individual dust wipe samples were collected
over a six month period of time, spanning from May 1992 through October 1992. Three
different components were sampled with 633 samples (69%) being collected from floors, 18
samples (2%) from window sills and 265 samples (29%) from window troughs. Table G-1
presents the number of individual samples, work areas and residential units that passed or failed
clearance testing within each combination of component type and site visit. All of the dust
samples were collected during the first site visit to a residential unit. On the first site visit, 97%
(891/916) of the individual dust samples fell below the clearance standards of 200 ug/ft2 for
floors, 500 ug/ft2 for window sills and 800 ug/ft2 for window troughs, 96% (679/704) of the
rooms passed clearance and 87% (137/158) of the residential units met clearance standards. This
increase in the failure rate from the percentage of individual samples and rooms that fail to the
percentage of residential units that fail is attributable to the fact that if any individual sample
exceeds the standard and fails clearance, then the residential unit from which it was taken also
G-2
-------
Table G-1. Clearance Testing Results by Individual Sample, Room, and Residential Unit for
the Dover Housing Authority.
visit, :>
^Component
.
-petal?
Floor
620
13
633
610
13
623
140
11
151
Sill
11
18
11
18
First
12
Trough
260
265
256
261
141
146
All
891
25
916
679
25
704
137
21
158
Total
Floor
Sill
Trough
All
620
11
260
891
13
7
5
25
633
18
265
916
610
11
256
679
13
7
5
25
623
18
261
704
140
7
141
137
11
5
5
21
151
12
146
158
fails clearance. The residential units that failed clearance on the first site visit
were revisited for a second clearance testing. The location of the failed test was recleaned and
retested. All of the samples collected during retesting passed the clearance standards. The
retesting data were not available. This report only analyzed the first site visit data.
The failure rate for individual samples was highest for window sill samples (39%),
followed by floors (2.1%) and window troughs (1.9%). Floors were also the most densely
sampled component with an average of 4.2 samples (633/151) taken per unit while 1.8 samples
(265/146) were taken from window troughs per unit. Among 158 housing units which were
included in the Dover data, only 12 housing units had window sill samples. While nine housing
units had only one window sill sample per house, 3 housing units had multiple window sill
samples collected within a house. Altogether there were 18 window sill samples collected in 12
housing units (an average of 1.5 samples per unit). Failure rates of residential units were low
based on floors or window troughs individually. Only 7% (11/151) of the residential units would
have failed based on the results of floor samples and 3% (5/146) of the residential units would
have failed based on the results of window trough samples. Among 12 housing units which had
window sill samples, 5 housing units had at least one window sill dust-lead loading exceeding
G-3
-------
the HUD Interim Guideline window sill standard, which resulted in a residential unit failure rate
of 42% (5/12) based on the results of window sill samples.
G-2. Objective 2; Characterization of the Distribution of the Dust-Lead Loadings.
Geometric Mean Dust-Lead Loadings. Variability Within a Housing Unit, and
Variability Between Housing Units
Table G-2 presents geometric means and estimated variance components for lead
loadings by component type. Dover data had a high window sill geometric mean, 462 ug/ft2, and
low floor and window trough clearance sample geometric means, 13.5 jig/ft2 and 15.8 ug/ft2,
respectively. The interior renovation in Dover consisted of 1) removing lead painted radiators
generally located under a window, and 2) window replacement. Window replacement did not
necessarily include replacing the window sill. Among 12 housing units which had window sill
samples, 4 housing units with 5 window sills had dust-lead loading results over 1,100 jig/ft2.
The fact that floor and window trough dust-lead loadings for those 4 housing units were all
below HUD Interim Guideline standards suggests that window sills might have failed to be
cleaned after renovation.
Figures G-l and G-2 contain box and whisker plots that present the distribution of dust-
lead loadings from the first and passed clearance visits by component type. Figures G-3 to G-5
contain box and whisker plots that present the distribution of dust-lead loadings from the first
visit for floor, window sill, and window trough, respectively; Figures G-6 to G-8 contain box and
whisker plots that present the distribution of dust-lead loadings from the passed clearance visits.
G-3. Objective 3: Characterization of the Correlation Between Components Sampled in
the Same Work Area
The relationships among floor, window sill, and window trough wipe samples are another
important aspect of examining a clearance testing program. One method of assessing the
relationships among individual floor, window sill, and window trough wipe samples collected
from the same room is to estimate linear correlation coefficients. Table G-3 displays the Pearson
product-moment correlation coefficients and the associated sample size for the log lead loading
measurements of individual floor, window sill, and window trough samples taken within the
same room. The data shows that floor samples are positively correlated with both window sill
G-4
-------
o
I
I
I
o
o
*=
(0
LU
o
a
i
a
•e
a
•o
a
I"
= a
.2 w
0>
v .£
H §
§i
at i-
(0 >
•S O
O O
CD
in
co
in
co
CM
CO
n
m
co
o
o
CO
in
ID
CM
to
CM
n
CD
CM
-------
100.0
sr
< 10.0
c
n
K 1.0
0.1
Boos WntawSfc Wnfaw Troughs Roos Window Sis
RrstM Passed Clearance
Surface
Figure G-l. Comparison of In-Transformed Dust-Lead Loadings from the First Site Visit
vs. Passed Clearance Results on an Expanded Scale.
G-6
-------
700
CM
< 500
400
300
200
100
0
ROOTS
Window Sills
Surface
Figure G-2. Box and Whisker Plots of the Distribution of Dust-Lead Loadings from the
Passed Clearance Data.
G-7
-------
10000
1000
g 100
10
All
Wood Vnyl Aluminum Ofter Unknown
Substrate
Figure G-3. Box and Whisker Plots of the Distribution of Floor Dust-Lead Loadings by
Substrate for the First Site Visit.
G-8
-------
100000
10000
CM
<
1000
0>
100
Wood
Vnyl
Aiuroinum
Other
Unknown
Figure 6-4. Box and Whisker Plots of the Distribution of Window Sill Dust-Lead Loadings
by Substrate for the First Site Visit.
G-9
-------
1000000
100000
oi 10000
g 1000
TJ
100
10
AS Wood Vnyl Aluminum
Substrate
Other
Unknown
Figure G-5. Box and Whisker Plots of the Distribution of Window Trough Dust-Lead
Loadings by Substrate for the First Site Visit.
G-10
-------
1000
100
£ io
All
Wood Vnyl Aluminum
Substrate
Other
Unknown
Figure G-6. Box and Whisker Plots of the Distribution of Floor Dust-Lead Loadings by
Substrate for the Passed Clearance Visits.
G-11
U S. EPA Headquarters Library
Mail code 3201
1200 Pennsylvania Avenue NW
Washington DC 20460
-------
1000
CM 100
Q
I
3
S. 10
Al
Wood Vnyj Akminun Other Unknown
Substrate
Figure G-7. Box and Whisker Plots of the Distribution of Window Sill Dust-Lead Loadings
by Substrate for the Passed Clearance Visits.
G-12
-------
1000
CM 100
0)
10
K
CL
AJ
Wood Vinyl Akmnum Other
Substrate
Unknown
Figure G-8. Box and Whisker Plots of the Distribution of Window Trough Dust-Lead
Loadings by Substrate for the Passed Clearance Visits.
G-13
-------
Table G-3. Observed Within-Room Correlation Coefficients Between Pb Loading
Measurements Collected From Floors, Window Sills and Troughs for
the Dover Housing Authority Data.
i
- 1. f
<-.
•v - *
'>>
*"• s .. _^
>
"* « ,
...Site Visit *
First
-v
'f
#
* City.'
Dover
" * * 'r'.T' ' ' " '* *
Observed*Within*RoornPCorrelation Coefficients Between!
V- ' L irffcsT
Floors and Window Sills-
?.
" r> **"*
Pptoor* suit
0.16
* *3- S* « * IS* ;T I j
i;n- _Jt *•
13
i, . J£§£*^f|
" Floors and Window Troughs
iff
Pploars Tioujli* k
0.12
184
and window trough samples. The correlations between floor and window trough results and
between floor and window sill results were not significant at the 0.05 level. A lack of data
prevented a correlation between window sills and window troughs from being estimated.
The conditional probabilities of a sample passing or failing a standard are given in
Table G-4. These analyses were conducted on two different sets of data, the first set using all
possible paired observations from within the same room. The second subset of data restricted the
analyses to rooms in which floor, window sill, and window trough lead loadings were
simultaneously observed. Note that the conditional probabilities results between window sills
and floors or window troughs may not be reliable due to lack of data.
For both types of estimates and both data sets included in the analysis, between floors and
window sills and between floors and window troughs, the probability that one component passes
clearance given that another component passes clearance ranges from 60% to 100%, while the
probability that one component fails clearance testing given that another component in the same
room fails clearance testing ranges from 0% to 54%.
G-4. Objective 4: Demonstration of the Impact of Composite Sampling on Pass/Fail Rates
of Houses
Individual sample lead loading results from each component type within a residential unit
were combined to construct simulated composite sample results.
Table G-5 provides the number of residential units which were investigated by the
number of individual samples collected from each component type. For example, there were 17
residential units where five floor dust-wipe samples were taken. In this analysis, residential units
containing more than four samples from a component type resulted in the estimation of multiple
G-14
-------
JD
CO
12
O
o
o
Z
•o
(0
(A
I
O
U
CM
X
CM
in
iu
o>
"•i5
0>
0)
o
c *i
« to
a >
•
.
a> iu
I a>
•i ^
>o *-
-° n
2
"
O £
O t-
63
1
F
i
•s
S:
"5
g
S
i1;
i
s-
•*•'
• .
H.
>-
C
o
ri:
"-
s
•p
•»=
1
*5
£
I
i
i
K
i
"J"
I
• ."
ill
CA ""S C
IL* 5
•1
JB
5
B-.
;*- j E
'«J*g |
IL"? 5
. .
•^
II
,3
i*. '
1 " '
< i '
•« •
5 '
Condltt
if!
it»*-S
J
'' 8 8
•££
i- i 6
1|]
* ^ *
a •
§ "i
: «
•*• • =
2
I-
*• -
§
*g
t3
s
0
o
§
e
o
S
CO
| P(Fi200
in
in
en
0
o
8
o
o
o
V-
§
V
CO
o
o
CN
V
IL
oT
n
UD
e
.
o
o
CM
U.
| PIStBOO
en
en
to
o
e
o
9
-------
simulated composite sample results. Therefore, summing all the units that had five or more
individual samples, approximately 19% (29/151) of the residential units resulted in the
Table G-5. Number of Residential Units that Contained (N) Individual Samples of Each
Component Type Based on First Site Visit Clearance Testing Data From the
Dover Housing Authority.
Total
Total with N*5
111
17
151
29
12
32
110
146
estimation of multiple simulated composite samples from floors. None of the residential units
tested resulted in the estimation of multiple simulated composite samples from window sills or
window troughs. This occurred since all of the units investigated had four or less individual
samples taken from window sills and four or less samples taken from window troughs. When
there were four or fewer individual samples from a component type within a housing unit, the
simulated composite sample included all samples.
For each component type within a residential unit, the set of individual clearance sample
lead loading results were used to construct simulated composite samples for the purpose of
evaluating the three composite sample clearance criteria (Standard, Standard/n, and
2xStandard/n). For each combination of component type and composite clearance criterion, each
residential unit either passed clearance, failed clearance, or yielded inconclusive results based on
the simulated composite samples. Inconclusive results were only possible for those residential
units which contained multiple simulated composite samples. Uncertainty in the decision rule is
G-16
-------
created by the multitude of ways in which multiple composite samples can be formed within
residential units which contained five or more individual clearance samples. Some of the
possible multiple composite samples formed may pass while other possible composite samples
may fail. As seen in Table G-6, the incidence of this type of uncertainty was very low for this
data set compared to the other data sets. The maximum observed percentage of residential units
with greater than four individual clearance samples of a given component type that resulted in
inconclusive results was 2.0% for floor samples using the Standard Rule.
Table G-6. Individual Sample Clearance Results Versus Simulated Composite Clearance
Results Based on Data from the Dover Housing Authority.
IfiCbmpbsiteo*
VKejxfusitfi&rtpgs
iCearance^
Sample Clearance I
Fail.:;
SPass'-
Pass
Standard
Inconclusive
Fail
140
141
Pass
Standard/n
Inconclusive
Fail
124
14
11
141
Pass
2 x Standard/n
Inconclusive
Fail
138
141
The performance characteristics of each combination of component type and composite
clearance criteria are presented in Table G-7 in terms of sensitivity, specificity, positive
predictive value (PPV), and negative predictive value (NPV). By design, the Standard/n Rule
sacrifices specificity for sensitivity. The sensitivity for the Standard/n Rule is always 1.00 while
the specificity for this rule is estimated to be 0.89 for floors, 1.00 for window sills, and 1.00 for
window troughs. On the other hand, the Standard Rule sacrifices sensitivity for specificity. The
specificity for the Standard Rule is always 1.00 while the sensitivity for this rule is estimated to
be 0.27 for floors, 1.00 for window sills, and 1.00 for window troughs. The 2xStandard/n Rule
attempts to maximize both sensitivity and specificity. The values of sensitivity for the
2*Standard/n Rule are always at least as high as those calculated for the Standard Rule. For
floors, the 2xStandard/n Rule, with an estimated sensitivity of 0.82, seems to do quite a bit better
than the Standard Rule. Similarly, the values of specificity are at least as high as those calculated
G-17
-------
for the Standard/ii Rule. Estimates of sensitivity and specificity in these examples are always
conservative, because inconclusive composite test results (i.e., houses in which only a fraction of
the simulated composite samples passed or failed) are included in the denominator for each
estimate, but are never included in the numerator.
Table G-7. Performance Characteristics of Composite Clearance Criteria Based on Data
from the Dover Housing Authority.
Sensitivity
0.27
1.00
0.82
Floors
Specificity
1.00
0.89
0.99
PPV
1.00
0.44
0.82
NPV
0.97
1.00
0.99
Sensitivity
1.00
1.00
1.00
Window
Sills
Specificity
1.00
1.00
1.00
PPV
1.00
1.00
1.00
NPV
1.00
1.00
1.00
Sensitivity
0.60
1.00
1.00
Window
Troughs
Specificity
1.00
1.00
1.00
PPV
1.00
1.00
1.00
NPV
0.99
1.00
1.00
Table G-7 shows that each of the three composite clearance testing criteria can have
different specificity and sensitivity rates. These rates correspond to the consistency between
clearance decisions based on individual clearance samples and clearance decisions based on
simulated composite samples. To further characterize the performance of each of the three
composite clearance criterion, the following logistic regression model was fitted for each
combination of component type and composite clearance criteria to describe the relationship
between the probability of passing clearance and the maximum lead loading present in all of the
sampling locations tested in a residential unit:
where nijk is the estimated probability of clearance for component(/) in house(j) under composite
criterion (£), and Max,j is the maximum individual sample lead loading result in house(0 for
component(/).
G-18
-------
The estimated relationship between the probability of passing clearance based on
simulated composite samples and the maximum lead loading in individual floor samples is
presented graphically in Figure G-9. In this figure, the solid, long-dashed, and finely-dashed
lines represent the estimated relationship for the Standard, Standard/n, and 2xStandard/n Rules,
respectively. These same relationships for window sills and window troughs were inestimable
because of the nature of the data. Essentially, an infinite number of curves could be fit to
estimate the relationship, but there is no criterion for choosing the optimal one.
Table G-8 provides parameter estimates and associated standard errors from the logistic
regression models, as well as estimates of the probability of passing clearance (using composite
samples) when the maximum lead loading among all locations included in the composite
sampling scheme is greater than or equal to 1A, 1,2, and 4 times the associated interim HUD
standard for individual samples. The estimates for the probability of passing based on composite
samples formed from window sill samples or window trough samples do not appear in the table
because the related parameters are not estimable due to the reasons stated in the previous
paragraph. The low sensitivity of the Standard Rule is demonstrated by the high estimated
probability (0.83) of passing when the maximum individual floor sample lead loading is equal to
the 2 times the HUD Standard. The estimated relationship for the Standard/n Rule demonstrates
this rule's high sensitivity (probability of passing is very low when the maximum individual
sample lead loading is greater than or equal to the HUD Standard) along with the loss in
specificity for this rule (probability of passing is only 0.73 when the maximum individual sample
lead loading is equal to !/2 HUD Standard). Once again, the 2xStandard/n Rule is shown to be a
compromise between the Standard and Standard/n Rules. At 1A HUD Standard, the estimated
probability of passing clearance testing under the 2xStandard/n Rule is 0.99, and at 2xHUD
Standard, the estimated probability of passing clearance testing under the 2 x Standard/n Rule is
0.00.
G-19
-------
Probability of Passing Clearance Testing Using Composite ROOT Samples
Standard Rule
(2 x Standard / n) Rule
(Standard / n) Bute
200 300 400 500 600
Maximum Individual Sample Floor Pb Loading
Figure G-9. Estimated Relationship Between the Probability of a Residential Unit Passing
Clearance Testing versus the Maximum Individual Lead-Loading Result by
Component Type Based on Simulated Composite Samples from the Dover
Housing Authority.
G-20
-------
£ a
li
2 w
2 S
°- '«
Q) O
£ &
C E
g o
*
«
II
CC O
.
W =
e -1 <
I « ?
0) .=
TO Q
111
.2 .£ **
« ^ £
g = o
. x
0 ^ •»
2^1
•^ o
•S c §
3 S «
O Q
CO
6
•lStiLl-_?S *K*
SSJfrV.i
•'->?* 1-1
.lit
Blls;
-;-S*»5',««
^524jg;
y»:-c
^ll
5J«0.
S^i?;
:^T"\.J «;*
",;«:.4^:'.'-.-i;Si
iii
ilil
" ^p >«-^
::;*ib£
g
8
d
P)
CO
o
q
d
o>
a>
I";
d
CM
o
o
o
03
0}
CO
d
CD
O
O
q
d
q
d
CD
^
q
d
in
n
o
to
q
d
s
CN
O
5
ao
en
s
q
(D
U>
CM
a
?
Q
X
CM
I
ID
**
CO
I
CO
0
•o
(O
X
(M
-a
I
a
•o
c
a
a
?
a
X
CM
.C
s
<5
?
-------
-------
APPENDIX H
Additional Analysis Results on the Percentage of Housing Units That
Passed Clearance on the First Site Visit, by the Number of Individual Samples
Collected Within a Housing Unit
H-1
-------
APPENDIX H
Additional Analysis Results on the Percentage of Housing Units That
Passed Clearance on the First Site Visit, by the Number of Individual Samples
Collected Within a Housing Unit
Appendix H presents additional analysis results on the percentage of housing units that
passed clearance on the first site visit, by the number of individual samples collected within a
housing unit. Clearance standards used in this appendix were based on HUD Interim Guidelines
(200 ug/ft* for floors, 500 ng/ft2 for window sills, and 800 fig/ft2 for window troughs). Table H-
1 uses a format similar to that of Table 6-13 in the text, but Table H-l presents the percentage of
units that passed clearance on the first site visit for the applicable component in each part of the
table (floor, window sill, and window trough).
Tables H-2a, H-3a, and H-4a present the number of residential units that contained
individual samples for the first site visit clearance testing data from the HUD Grantee Program
(both High and Low data groups), by interior intervention strategy code, for floors, window sills,
and window troughs, respectively. Tables H-2b, H-3b, and H-4b, companion tables for Tables
H-2a, H-3a, and H-4a, present the percentage of residential units that passed clearance for the
first site visit clearance testing data from the HUD Grantee Program (both High and Low data
groups), by interior intervention strategy code, for floors, window sills, and window troughs,
respectively. HUD Grantees reported the intensity of the interior intervention as a strategy code
(level 01 to 07). Higher strategy levels reflect more intensive treatment. Each dwelling unit was
assigned only one interior intervention strategy. The interior intervention strategies were
summarized in Table 4-1.
Table H-2a, H-3a, and H-4a show that, regardless of component type, level 05 of the
interior intervention strategy was used in housing units more than any other intervention strategy
for both HUD Grantee High and Low data groups. Level 05 of the interior intervention strategy
includes window replacement and wall enclosure/encapsulation, and other lower levels of
intervention activities.
Tables H-2b, H-3b, and H-4b show that, generally, the percentage of housing units
passing clearance does not increase as the intensity of the interior intervention increases. This is
H-2
-------
true across all 3 components and for both HUD Grantee High and Low data groups. The most
popular interior intervention strategy, level 05, has residential unit passing rates of 86%, 95%,
and 94% based on floors, window sills, and window troughs, respectively, for the HUD Grantee
High data group. For the HUD Grantee Low data group, interior intervention strategy level 05
has residential unit passing rates of 83%, 91%, and 91% based on floors, window sills, and
window troughs, respectively.
H-3
-------
Table H-1. Percentage of Housing Units that Passed Clearance for Each Data Source That
Contained (N) Individual Clearance Samples of Each Component Type
Based on the First Site Visit.
-fiSJ^Data!i^rui!c?|S^f|
'^^^^^^^^^M^^^^^^^^^Utt^
M-SiSi
HlUt
•,-s.v. £"amt.
.tei^sss
?Klfe#
~!V>';L'Wi.^:i,M
^;w3-.;X'-r
••*;""-•**?#
••'. -.f^i-SS
*s-fci
... Floor Standard 200 t/g/ft2 •.-•_.' :
Maryland
HUD FHA
HUD PHA
HUD Grantee (High)'
HUD Grantee (Low)"
Atlantic City
Cleveland
Dover
89
50
100
87
93
100
67
87
71
92
86
92
86
100
100
80
33
75
89
82
70
62
83
81
63
69
86
92
85
50
96
73
58
71
82
90
70
60
82
70
53
63
85
78
80
100
88
67
47
76
79
57
100
100
74
65
38
81
50
50
.
67
64
41
80
60
67
100
100
•
79
54
74
85
89
80
63
93
"— ~. -, " " '-- •. . Window Sill Standard 500 j/g/ft2 I - ""•';--*•"•- "- -
Maryland
HUD FHA
HUD PHA
HUD Grantee (High)'
HUD Grantee (Low)"
Atlantic City
Cleveland
Dover
79
88
100
90
94
100
100
78
79
60
93
94
95
89
100
0
74
71
87
89
91
67
92
0
73
88
96
88
73
50
100
0
88
57
86
90
75
100
100
80
70
96
65
50
.
.
71
55
.
100
.
.
.
78
33
100
100
.
.
69
57
.
0
0
.
-
77
63
93
91
93
80
97
58
Window Trough Standard 800 //g/ft2 '• • • - ...-,, "•'
Maryland
HUD FHA
HUD PHA
HUD Grantee (High}'
HUD Grantee (Low)"
Atlantic City
Cleveland
Dover
72
92
100
92
91
100
93
94
65
50
87
89
86
95
83
97
62
26
100
94
85
100
100
100
65
38
100
96
62
83
100
100
63
50
91
92
67
100
.
68
21
100
71
36
50
100
.
.
46
0
100
100
.
.
.
47
0
.
,
.
.
63
44
95
91
88
96
92
97
* Grantees that used
* * Grantees that used
200 jug/ftz as clearance standard for floor.
100 //g/ft2 or 80 //g/ft2 as clearance standard for floor.
H-4
-------
Table H-2a. Number of Residential Units that Contained (N) Individual Floor Samples for
the First Site Visit Clearance Testing Data from the HUD Grantee Program,
by Interior Intervention Strategy Code.
.i»Jnteiy«ntion,-jW:i?'
lsSt^egy;Co^l|!
prill
•- • f-
01
02
03
04
05
06
07
Missing
Total
14
30
24
29
83
1
0
55
236
10
70
31
80
113
0
2
107
413
• rPi.JSis&pigS-
^tttf.-'WtaElW
fe4W
$^~|||Ki
•^""SS^i^'r!;
'3f,»V>ivirji
^JJ6*rr»
••»*:S^K'ii^!<'t;
•^-i^T'-i-j,^
>'vr.v7?r,**..»v
, HUD Grantee (High)* -
7
28
15
64
123
1
2
94
334
4
24
9
61
128
2
1
114
343
2
15
7
45
108
5
0
76
258
2
9
5
38
161
4
0
69
288
0
1
4
11
93
2
0
31
142
0
0
0
19
14
0
0
3
36
- . ... - ' : : HUD Grantee (Low)"
01
02
03
04
05
06
07
Missing
Total
2
21
15
9
40
0
0
42
129
4
5
21
15
34
4
0
41
124
1
4
33
19
48
10
1
43
159
1
10
40
19
59
47
0
170
346
0
4
44
9
53
15
0
76
201
0
2
15
6
8
1
0
18
50
0
0
1
2
3
0
0
1
7
0
0
1
0
0
0
0
1
2
0
0
0
18
21
0
0
3
42
0
0
0
0
1
1
0
1
3
39
177
95
365
844
15
5
552
2092
' ."•*•• '
8
46
170
79
246
78
1
393
1021
Grantees that used
Grantees that used
200 /sg/ft2 as clearance standard for floor.
100 pg/ft2 or 80 jvg/ft2 as clearance standard for floor.
H-5
-------
Table H-2b. Percentage of Residential Units that Passed Clearance for Floors (Based on
Standard at 200 //g/ft2) for the First Site Visit Clearance Testing Data from
the HUD Grantee Program, by Interior Intervention Strategy Code.
^.liit«fv«*itioii.|fels;'l|i«^"'i-'»*-'."'-iv
Sstfet^c'odiftllisa'fe
^"H-^lIiT"' ,: *•£•"•'
i£i2' £••'
;^!p
'.sjsssf&sr
MS&
- " . HUD Grantee (High!" '/ • •-•••
01
02
03
04
05
06
07
Missing
Total
* - '*""»•
. . , ^ * , ' ~ ' '
01
02
03
04
05
06
07
Missing
Total
93
87
83
72
89
0
.
93
87
.
50
100
87
89
95
.
.
93
93
90
89
65
88
86
.
100
90
86
.'. ' .
100
80
90
87
91
100
.
95
92
71
86
80
94
89
100
100
89
89
100
88
78
82
85
100
0
89
86
50
87
43
84
89
60
76
82
100
89
100
82
86
75
.
81
85
100
75
73
80
100
.
77
79
84
71
.
100
81
50
67
.
.
67
60
87
88
74
83
86
73
80
86
85
HUD Grantee (Low)" ' ' .
100
100
94
74
77
60
100
86
82
100
80
93
74
81
96
97
92
.
75
91
78
81
93
.
96
90
.
50
93
67
63
100
.
78
78
.
0
50
100
.
.
0
57
.
0
.
.
.
.
100
50
.
0
100
.
100
67
88
89
91
77
83
91
100
94
89
* Grantees that used 200 pgltt? as clearance standard for floor.
** Grantees that used 100 /jg/ft2 or 80 //g/ft2 as clearance standard for floor.
H-6
-------
Table H-3a. Number of Residential Units that Contained (N) Individual Window Sill Samples
for the First Site Visit Clearance Testing Data from the HUD Grantee Program,
by Interior Intervention Strategy Code.
-^^iSsSPSiPil
j'i sv- s?; UtUnOfnatSlSlar
fyt> j=>.i-ir«.-S«s>p;^!tMS»fe
^ lntwyttilMn%&
•t'S'tratelyttCodell?-:'
I?,-:;*:- -3i?,!-i^isilsp
-S^l»^5|*-s«2^|'
01
02
03
04
05
06
07
Missing
Total
•• ,-"'•'*.'•' ''.'. • " -'
01
02
03
04
05
06
07
Missing
Total
9
36
24
61
80
1
0
52
263
r*S.3SSR~i«f
PS33&
•rwaspQji
..,1:^.4^^
'Sifs^illfilefe-
;- 7 :.•;.?•
HUD Grantee (High)*
10
76
22
123
520
8
1
308
1068
11
37
20
107
155
3
3
165
501
3
13
7
51
42
2
1
14
133
0
3
2
10
24
0
0
2
41
0
0
0
8
9
0
0
3
20
0
0
0
1
1
0
0
0
2
1
0
0
0
0
0
0
1
2
' ;.:-' HUD Grantee (Low)"
2
13
11
7
30
9
0
48
120
2
28
128
31
108
49
1
240
587
2
3
20
30
80
9
0
70
214
0
0
3
7
15
0
0
15
40
0
0
1
1
0
0
0
2
4
0
0
0
0
1
0
0
1
2
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
1
0
0
0
1
34
165
75
361
832
14
5
545
2031
0
0
0
0
1
0
0
0
1
6
44
163
76
235
67
1
376
968
Grantees that used
Grantees that used
200 i/g/ft2 as clearance standard for floor.
100 //g/ft2 or 80 pg/ft2 as clearance standard for floor.
H-7
-------
Table H-3b. Percentage of Residential Units that Passed Clearance for Window Sills (Based
on Standard at 500 //g/ft2) for the First Site Visit Clearance Testing Data from
the HUD Grantee Program, by Interior Intervention Strategy Code.
iiiSciiSiSysSi
•«»fSjS«(»«»".\«Hsp>
,-1*;£,liiieiv6!iti6ii?c<^^
*s'SaitwfCo" ...','?-. ••>:-• W..-T>>^-:. .r* :•»•-.,••!.? :'"W- s,v *»™i rr^iiiaetpfr'*-;', v-iis'j,-- siiii.^^wjsijT:^:'5.1 .I1 * ™''v,;«?4?f»HWi:**ip»s.M^j*^5^^
<$*|S*4l$S^
• .
67
78
96
87
93
100
98
90
90
80
86
93
97
88
100
93
94
100
73
90
94
93
100
100
85
89
^^^Mlifllt
HUD Grantee (High)* "
100
85
86
86
90
100
100
86
88
33
100
90
96
.
100
90
.
.
38
78
.
.
100
65
J.3.:~te~l.'$
&«57.>K$
.
.
100
100
100
'£M*i$eis IflWPSSS?
Ss^^li^rotaii!
100
.
.
.
.
100
100
.
.
.
.
.
.
,
.
88
78
91
90
95
93
100
91
91
-. •'.' .-!•'.-.•' •'•• "'•• . • -. '^ :,''<• ~'\ •• •-; .. 'HUD Grantee iUwj't; .:/!.;; , •••;••'• -..•.i."^'^ ••',../'• '--..'..;
01
02
03
04
OS
06
07
Missing
Total
50
100
91
86
97
89
.
96
94
100
96
95
87
93
98
100
95
95
100
33
95
80
90
100
.
97
91
67
29
80
.
87
73
.
.
0
100
.
.
100
75
.
.
0
100
50
.
.
.
.
.
-
.
.
.
.
.
.
.
.
.
.
.
.
.
-
83
93
94
79
91
97
100
95
93
Grantees that used
Grantees that used
200 pg/ft2 as clearance standard for floor.
100 //g/ft2 or 80 Arg/ft2 as clearance standard for floor.
H-8
-------
Table H-4a. Number of Residential Units that Contained (N) Individual Window Trough
Samples for the First Site Visit Clearance Testing Data from the HUD Grantee
Program, by Interior Intervention Strategy Code.
*|fe Interior;:*^
li'Strategy3 Codelif
^^^'i^^&^-i^^^^^^^^^^^^^*'y^^^^^^^^^l^^^^S^&
•iifciiif/jwi-ji"
•••MS*?-'— Ail- ;,'*
•sS??1;M*
••-."Vi'.'"-i ••**,'
'•^••.zS'^if
,£?!»! t"--'i'
'••^'^•'•i'-T"''' ••
•8*1*3 JX-;
-^!gSP|:|i»l;i^
lf:j'|4;35*j:|.^,!i|.5_ii^i
©^•gSpii'h^Si^MJ^irdtali
HUD Grantee (High)* "
01
02
03
04
05
06
07
Missing
Total
11
99
24
67
260
0
5
246
732
4
17
14
109
350
12
0
221
727
0
1
1
57
78
2
0
31
170
0
0
0
21
21
0
0
4
46
0
0
0
4
9
0
0
0
13
0
0
0
5
2
0
0
0
7
0
0
0
0
1
0
0
0
1
0
0
0
1
0
0
0
0
1
.. ••;: " '-
0
0
0
0
0
0
0
0
0
• •
15
117
39
284
721
14
5
502
1697
'L.:... . : '. . ^ . . - HUD Grantee (Low)" 5v : . ..•'>•.:-..••'
01
02
03
04
05
06
07
Missing
Total
4
13
59
32
148
51
1
187
495
2
27
72
27
52
14
0
108
302
0
2
21
10
9
0
0
18
60
0
0
0
4
3
1
0
5
13
0
0
1
0
2
0
0
0
3
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
6
42
153
73
214
66
1
318
873
* Grantees that used
* * Grantees that used
200 //g/ft2 as clearance standard for floor.
100 //g/ft2 or 80 //g/ft2 as clearance standard for floor.
H-9
-------
Table H-4b. Percentage of Residential Units that Passed Clearance for Window Troughs
(Based on Standard at 800 /ig/ft2} for the First Site Visit Clearance Testing
Data from the HUD Grantee Program, by Interior Intervention Strategy Code.
pSirP^Hi
^Strategy Code%
«t3f»? >v
•stSSJM
*£.*•'*(£*%:
&iS2-3;*a
lKs%.
g^'j-iO^.^.rr-
^ffs-^SiiS*
felit
•?.:'.•.•••-•.' "••I'.'i • : HUD Grantee (High)* :. l
01
02
03
04
.05
06
07
Missing
Total
100
88
96
95
96
.
100
88
92
.''-..-. • ' .'.'• " " ' r
01
02
03
04
05
06
07
Missing
Total
75
77
93
88
94
88
100
91
91
75
94
93
93
91
92
.
84
89
100
100
91
96
50
.
97
94
100
95
.
75
96
.
100
89
.
.
.
92
.
.
60
100
.
71
• ••;"•*.. HUD Grantee (Low)"
100
59
93
70
85
86
.
92
86
50
86
100
89
.
78
85
.
.
.
100
67
100
.
20
62
.
0
100
•
.
.
67
.
.
.
.
.
.
.
.
.
100
.
.
.
100
• -
.
.
.
.
.
.
.
.
.
100
.
.
.
100
.
.
.
.
.
93
89
95
93
94
86
100
86
91
.; ;- .. . '- - • ': '
,
.
.
.
.
.
.
.
.
.
83
64
92
84
91
88
100
90
88
Grantees that used 200 //g/ft2 as clearance standard for floor.
Grantees that used 100 jjg/tt1 or 80 fjg/ft* as clearance standard for floor.
H-10
-------
APPENDIX I
Additional Analysis Results
1-1
-------
APPENDIX I
ADDITIONAL ANALYSIS RESULTS
Appendix I presents additional analysis results on the distributions of the clearance data.
Figures 1-1 to 1-6 display box and whisker plots for the distributions of floor, window sill, and
window well dust-lead loadings (ug/ft2) and log-transformed dust-lead loadings
(log ug/ft2) from the first site visit by substrates for Dover and Maryland data. Detailed values
for different percentiles for the untransformed dust-lead loadings (ng/ft2) are presented in Table
1-1. Figures 1-7,1-8, and 1-9 display box and whisker plots for the distributions of dust-lead
loadings (fig/ft2) and log-transformed dust-lead loadings (log ug/ft2) from the first site visit by
data source, for floors, window sills, and window wells, respectively. Similar detailed values for
different percentiles for the untransformed dust-lead loadings (ug/ft2) are presented in Table 1-2.
Box and whisker plots of the percentiles by surface and substrate for Dover and Maryland
on the "Passed Clearance" data sets are provided in Figures 1-10 through 1-15. Specific upper tail
percentiles are given in Table 1-3. Figures 1-16 through 1-18 display the distribution of clearance
results for data sources that "Passed Clearance." Additional percentile information for these data
sources is presented in Table 1-4.
In Tables 1-2 and 1-4, Figures 1-7 to 1-9, and Figures 1-16 to 1-18, "Grantee 1" refers to
data from a group of nine grantees (Alameda County, Baltimore, Boston, California,
Massachusetts, Milwaukee, Rhode Island, Vermont, and Wisconsin) in the HUD Grantee
Program that used the HUD Interim Guidelines clearance standards, i.e., 200,500, and 800 ug/ft2
for floors, window sills, and window troughs, respectively; "Grantee 2" refers to data from the
other group of five grantees (Cleveland, Chicago, New Jersey, New York City, and Minnesota)
that used a lower floor dust-lead clearance standard (i.e., 100 ug/ft2 or 80 ug/ft2). These were
referred as "HUD Grantee (High)" and "HUD Grantee (Low)" in the main body of the report.
1-2
-------
10000
Of
k
e
woo
wo-
Vlnyt
*»
I'
Vinyl
Figure 1-1. Box and Whisker Plots of the Distributions of Floors Dust-Lead Loadings and
In-Transformed Dust-Lead Loadings on Dover Data by Substrate for the First
Site Visit
I-3
-------
fi
woo
100
Al
Vinyl
Vta*
Mumtoum
Figure I-2. Box and Whisker Plots of the Distributions of Window Sills Dust-Lead
Loadings and In-Transformed Dust-Lead Loadings on Dover Data by
Substrate for the First Site Visit
I-4
-------
K3000.0
1000.0
e
KXO
1.0
0.1-
Vinyl
Aluminum
CXhtr
ID-
S'
w 7
1 e
j:
I 3
2'
1
0
-1
1
A
1 Vfe
-
I
U 5
t
«l Mumtom Qttm IMmmn
SubetratB
Figure [-3. Box and Whisker Plots of the Distributions of Window Troughs Dust-Lead
Loadings and In-Transformed Dust-Lead Loadings on Dover Data by Substrate
for the First Site Visit
1-5
-------
£
10000'
1000-
ttO-
TOr
Other
10
f
Wood
Ottwr
Figure I-4. Box and Whisker Plots of the Distributions of Floors Dust-Lead Loadings and
In-Transformed Dust-Lead Loadings on Maryland Data by Substrate for the
First Site Visit
I-6
-------
£
100000'
10000'
woo
1-
wsx*
Ahvrinum
Otar
11
10
9'
a
tlMinauot
-------
1000000
fi
10000
1000
100
1-
Wood
Vinyl
Mun*wn
13-
12
tl
TO
W 9
% *'
f7-
6
fs
4-
•
2
1-
0
1
^
i !
_
'
M Waod Vinyl AUrtntm IMnewi
Figure 1-6. Box and Whisker Plots of the Distributions of Window Troughs Dust-Lead
Loadings and In-Transformed Dust-Lead Loadings on Maryland Data by
Substrate for the First Site Visit
1-8
-------
Table 1-1. Percentiles (//g/ftz) of the Clearance Data by Surface and Substrate for the First
Site Visit for Dover and Maryland.
' .,t'Data;~^
": Source "^
Dover
Maryland
• i i %
K \ * •
*• H
'Surface
Floors
Sitls
Troughs
Floors
Sills
Troughs
••* O ' -,
t \ •, *
Substrate "
All
Wood
Vinyl
Unknown
All
Wood
Vinyl
Aluminum
Unknown
All
Vinyl
Aluminum
Other
Unknown
All
Wood
Vinyl
Other
Unknown
All
Wood
Vinyl
Aluminum
Other
Unknown
All
Wood
Vinyl
Aluminum
Unknown
. Sample „
.' SiV'
633
3
625
5
18
9
1
4
4
265
185
11
2
67
27SO
1082
660
19
989
2465
1687
97
40
18
623
1985
458
551
286
690
l*50th .•
'Percentile •
5
31
5
9
443
718
212
443
993
7
7
10
136
7
19
18
21
84
17
28
24
12
10
12
48
92
283
43
51
133
80th ;
Percentile
44
61
44
72
1260
1260
212
467
2280
65
77
16
239
12
85
77
99
116
82
181
144
47
84
55
439
844
2160
239
337
1370
/•j-i-JBOth:^:^.
Percentile I
68
61
68
96
1620
1620
212
467
2280
137
152
16
239
68
181
159
200
725
182
511
376
116
186
327
1350
3120
7300
684
983
5700
|::|?J5thS|
fPercentiie*
107
61
107
96
2280
1620
212
467
2280
202
222
106
239
89
330
297
360
1110
347
1370
864
420
280
406
2790
9480
16600
1530
2580
15100
i!>>P.iereehtHe"v
: ?^::.-*- i™-.'.iaf.!.*iS-..rt •.
317
61
317
96
2280
1620
212
467
2280
1830
2040
106
239
643
1130
892
1400
1110
2580
5650
3590
6620
4080
406
9500
49800
59600
10200
38900
63800
I-9
-------
fi
100000.00
10000.00
1000.00
100.00
10.00
1.00
0.10
0.01
PHA Grantee 1 Grantee 2 FHA Atlantic City Cleveland
Data Source
11
10
9
8
* 7
< 6
H 5
E? 4
1 3
fi 2
* ;
-1
-2
-3
-4
,
j
4.
T
I
1
u
T
I
1
PHA Grantee 1 Grantee 2 FHA
Data Source
Atlantic City Cleveland
Figure I-7. Box and Whisker Plots of the Distributions of Floors Dust-Lead Loadings and In-
Transformed Dust-Lead Loadings by Data Source for the First Site Visit
1-10
-------
100000.00
10000.00
1000.00
100.00
10.00
1.00
0.10
0.01
PHA Grantee 1 Grantee 2 FHA AttenBc City Cleveland
Data Source
11
10
9
8
sr 7
k 6
I :
.§ 3
§2
1
0
-1
-2
-3
•
T
[
u
I
1
PHA Grantee 1 Grantee 2 FHA
Data Source
Atlantic City Cleveland
Figure 1-8. Box and Whisker Plots of the Distributions of Window Sills Dust-Lead Loadings
and In-Transformed Dust-Lead Loadings by Data Source for the First Site Visit
1-11
-------
1000000.0
100000.0
10000.0
1000.0
100.0:
10.0]
1.0
0.1]
PHA Grantee 1 Grantee 2 FHA Atlantic City Cleveland
Data Source
13
12
11
10
9
8
7-
6
PHA Grantee 1 Grantee 2 FHA
Data Source
Atlantic City Cleveland
Figure 1-9. Box and Whisker Plots of the Distributions of Window Troughs Dust-Lead
Loadings and In-Transformed Dust-Lead Loadings by Data Source for the First
Site Visit
1-12
-------
Table 1-2. Percentiles (j/g/ft2) of the Clearance Data by Surface and Data Sources for the
First Site Visit.
f' Surf ace'; ^.
Floors
Sills
Troughs
; Data'Source '
PHA
Grantee 1
Grantee 2
FHA
Atlantic City
Cleveland
PHA
Grantee 1
Grantee 2
FHA
Atlantic City
Cleveland
PHA
Grantee 1
Grantee 2
FHA
Atlantic City
Cleveland
Sample *
Size :-~ •
558
8179
3642
967
554
158
411
4798
2147
668
51
93
371
3002
1346
442
258
39
:""••'. - • •' ,, • •-
s^'SOth-i^
;Percentite:',
48
12
16
48
10
27
28
17
38
52
31
33
42
30
81
375
10
33.3
:^80th'i^
•iPercentileft
111
47
39
187
55
87
80
84
106
271
150
125
85
150
285
2569
70
170
; :•''•' ;9oth Kl:
Percentile .r
200
110
76
418
130
190
175
202
171
714
380
190
167
396
692
5810
190
530
^'PercehtOe^?
312
227
168
871
230
320
281
418
356
1678
660
250
268
937
1527
9150
450
850
1-PercentBeF
600
1030
664
3119
700
440
1175
1870
1568
9054
790
1700
1089
7600
7088
50534
2130
2900
1-13
-------
1000-
e
«0
1-
Al
Wootf
f
Al
Figure 1-10. Box and Whisker Plots of the Distributions of Floors Dust-Lead Loadings and In-
Transformed Dust-Lead Loadings on Dover Data by Substrate for the Passed
Clearance Visits
1-14
-------
1000
I
£
100
Vinyl
Mun*wn
Wood
Aluminum
Figure 1-11. Box and Whisker Plots of the Distributions of Window Sills Dust-Lead Loadings
and In-Transformed Dust-Lead Loadings on Dover Data by Substrate for the
Passed Clearance Visits
1-15
-------
CM
f
noao
101X0
100
MumkMn
Otttar
7'
6
5
C 4
S 3
ffc 8'
1-
0
-1-
4
•
i
•
*
I
•
Al Vinyl Abmhun OOMT Unknown
Figure 1-12. Box and Whisker Plots of the Distributions of Window Troughs Dust-Lead
Loadings and In-Transformed Dust-Lead Loadings on Dover Data by Substrate
for the Passed Clearance Visits
1-16
-------
1000-
£
100-
1-
Vh*
Ottwr
f 4*
Vlnyt
OOur
Figure 1-13. Box and Whisker Plots of the Distributions of Floors Dust-Lead Loadings and In-
Transformed Dust-Lead Loadings on Maryland Data by Substrate for the
Passed Clearance Visits
1-17
-------
£
10
Wfcod
Othsr
Muntinum
Otfw
Figure i-14. Box and Whisker Plots of the Distributions of Window Sills Dust-Lead Loadings
and In-Transformed Dust-Lead Loadings on Maryland Data by Substrate for the
Passed Clearance Visits
1-18
-------
1000
100
£
Wood
Vinyl
Akmfemm
Aunvnum
Figure 1-15. Box and Whisker Plots of the Distributions of Window Troughs Dust-Lead
Loadings and In-Transformed Dust-Lead Loadings on Maryland Data by
Substrate for the Passed Clearance Visits
1-19
-------
Table 1-3.
Percentiles (//g/ft2) of the Passed Clearance Data by Surface and Substrate for
Lead Loading Results for Dover and Maryland.
. Source •
Dover
Maryland
Surface
Floors
Sills
Troughs
Floors
Sills
Troughs
Substrate
All
Wood
Vinyl
Unknown
Ail
Wood
Vinyl
Aluminum
Unknown
All
Vinyl
Aluminum
Other
Unknown
All
Wood
Vinyl
Other
Unknown
All
Wood
Vinyl
Aluminum
Other
Unknown
All
Wood
Vinyl
Aluminum
Unknown
>'.- Sample >'s.
- ; Size' "•'•''
620
3
612
5
11
4
1
4
2
260
180
11
2
67
2712
1052
651
20
989
2411
1651
106
44
18
592
1827
373
575
282
597
" ' ,',/• - *•'•'* •„"•! '•%";
.^(Vj-hSOthv-iv- '.
• Percentile "\
5
31
5
9
361
262
212
443
398
7
7
10'
136
7
16
15
18
67
15
21
20
12
11
12
32
50
88
38
38
53
•r.";", ->:-'-'vv-:., .*%:.•
-' '•'•'• 80th ,.'.•.::: /.
rPercentile '•".
43
61
43
72
443
377
212
467
434
59
67
16
239
12
60
58
69
105
56
101
99
37
83
55
119
231
338
163
168
257
; Percentfl*%
61
61
60
96
443
377
212
467
434
111
140
16
239
68
98
88
106
116
98
201
193
66
149
327
238
400
530
308
322
412
f Percentile y
82
61
81
96
467
377
212
467
434
157
169
106
239
89
128
121
135
118
134
295
295
167
234
406
315
571
655
515
456
604
1-20
-------
£
1000.00
100.00
10.00
1.00
0.10
0.01
PHA Grantee 1 Grantee 2 FHA
Data Source
Atlantic City Cleveland
5 -1
-2
-3
-4
PHA Grantee 1 Grantee 2 FHA
Data Source
Atlantic Cily Cleveland
Figure 1-16. Box and Whisker Plots of the Distributions of Floors Dust-Lead Loadings and
In-Transformed Dust-Lead Loadings by Data Source for the Passed Clearance
Visits
1-21
-------
1000.00
100.00
CM
<
10.00
£
1.00
0.10
0.01
PHA Grantee 1 Grantee 2 FHA
Data Source
Atlantic City Cleveland
7
6
5
ST 4
§ 3
•5 2
n 1
0
-1
-2
—3-
*»•
I
PHA Grantee 1 Grantee 2 FHA
Data Source
Atlantic City Cleveland
Figure 1-17. Box and Whisker Plots of the Distributions of Window Sills Dust-Lead
Loadings and In-Transformed Dust-Lead Loadings by Data Source for the
Passed Clearance Visits
I-22
-------
CM
<
£
1000.0
100.0
10.0
1.0
0.1
PHA Grantee 1 Grantee 2 FHA
Data Source
Aflantte City Cleveland
7
6
51
sr 4
<
I* 2
£ 1
-1
-2-
-3-
PHA Grantee 1 Grantee 2 FHA
Data Source
Atlantic CIV Cleveland
Figure 1-18. Box and Whisker Plots of the Distributions of Window Troughs Dust-Lead
Loadings and in-Transformed Dust-Lead Loadings by Data Source for the
Passed Clearance Visits
I-23
-------
Table 1-4. Percentiles (/ig/ft2) of the Passed Clearance Visits by Surface and Data Sources
for Lead Loading Results.
Surface
Floors
Sills
Troughs
Data Source
PHA
Grantee 1
Grantee 2
FHA
Atlantic City
Cleveland
PHA
Grantee 1
Grantee 2
FHA
Atlantic City
Cleveland
PHA
Grantee 1
Grantee 2
FHA
Atlantic City
Cleveland
Sample
Size
554
8345
3749
997
545
161
412
4906
2170
689
46
93
371
3036
1341
423
267
39
50th
Percentile
43
12
15
40
10
22
27
17
36
47
30
32
41
27
68
169
10
33
80th
Percentile
93
39
34
102
46
64
68
72
99
164
90
120
81
116
198
431
66
170
90th
Percentile
118
70
57
141
80
110
150
141
145
263
180
170
138
237
346
598
170
490
95th
Percentile
148
108
90
165
120
134
222
236
215
358
270
230
213
389
512
678
320
540
99th
. Percentile
185
173
163
191
170
182
378
410
372
481
380
388
399
691
725
760
650
648
(-24
-------
APPENDIX J
Comparison of Clearance Guidance
J-1
-------
APPENDIX J
COMPARISON OF CLEARANCE GUIDANCE
The following is a comparison of the guidance on clearance in the 1990 HUD Interim
Guidelines [2], the 1995 EPA 403 Guidance [5], the 1995 HUD Guidelines [3], the 1996 EPA
402/404 Rule [4], the 1999 HUD 1012/1013 Final Rule [22], and the 2001 EPA 403 Final Rule
[1]. Table J-l presents a comparison of the number and location of samples for single sample
and composite sample testing in contained and non-contained areas. Clearance testing
procedures for visual inspections are shown in Table J-2, procedures for sealants are in Table J-3,
and waiting times for conducting clearance testing are compared in Table J-4. Clearance
standards are outlined in Table J-5, failure procedures in Table J-6, a comparison of the units in
which to report samples is presented in Table J-7, compositing rules given in Table J-8, and
conflict of interest issues are compared in Table J-9.
Below are some of the major changes that have taken place in clearance standards over
time.
• 1990 HUD Interim Guidelines [2] clearance standards are 200 ng/ft2 for bare floors,
500 ug/ft2 for window sills, and 800 ug/ft2 for window wells (troughs).
• 1994 EPA 403 Guidance [5] clearance standards are 100 jig/ft2 for bare floors, 500
ug/ft2 for interior window sills, and 800 ng/ft2 for exterior window sills (troughs)-and
exterior horizontal surfaces.
• 1995 EPA 403 Guidance was the same as 1994 EPA 403 Guidance, but was
disseminated as a Federal Register Notice [5].
• 1995 HUD Guidelines [3] clearance standards are 100 ug/ft2 for floors (including
carpeted and uncarpeted floors), 500 ug/ft2 for window sills, and 800 ug/ft2 for
window wells (troughs) and exterior concrete or other rough surfaces.
• 1996 EPA 402/404 Rule [4] refers to the clearance levels in the EPA Guidance on
Residential Lead-Based Paint, Lead-Contaminated Dust and Lead Contaminated Soil
or other equivalent guidelines.
• 1999 HUD 1012/1013 Final Rule [22] dust-lead clearance standards are 40 ^ig/ft2 for
floors (including carpeted and uncarpeted interior floors), 250 ^ig/ft2 for interior
window sills, and 800 ug/ft2 for window troughs.
J-2
-------
• 2001 EPA 403 Final Rule [1] dust-lead clearance standards are 40 ug/ft2 for floors
(including carpeted and uncarpeted floors), 250 ug/ft2 for interior window sills, and
400 ug/ft2 for window troughs.
J-3
-------
£
Q.
E
(9
V)
(0
£
o
O
•o
CO
0)
J»
Q.
CO
_
o>
c
CO
**
CO
•o
CO
CO m
5 «
o
flu
o
«=
CO
•^ 2
0 c
c o
1 =
i§
Is
CO
3
u
it
«s-S-
I 2 S £
E
-------
rlt
;-"-!V5^?(
-$$$!
Jilt
~ '^ySfeifl
vj'j-IS^IB
ijyploa
>fl
•i«j'<*jjc-.'.
18
i.jili;
'$£1 ip^
Mm
s, aj
Its
i*6 is
'If «S
•Ui
Hi
ft;*
I
If
IS
:« 3o
w ;:z
i S-
•'*<, '?£•'
11
if
5 St
III
J-> *-i- f"
Hi
'^ft;fe
t^i^-^"
H
c
a
•& S
a> I
3 jg
» a.
°> c _
— C T3
Q. CO 03
£ W £
Sol
0 .E -2
Z W -D
^
c
"a «
15
e ~
'» s S
lil
li|
0 =6 -D
J£
O
(0 o
°T3 1
CAO a>
|| |
ia »
S-D "s
.£•3 i
= | |
viost of the clearance
sampling should be
:onducted within the
;ontainment area.
Sj <»
£§
Sf
ME
o .£ -i a ** o .S
SI IE « SI
s| 5* 1=1
*& I'l 5i5
2 ? Is -i S ?
I to •£ c.l =•-£ « ^
+-'55 w w S »*-'» w
c a? w o ,2 ccv
|l|Jii ||i|l
.Is »Il
o o £ .5 (- £ "K c
£ .ti 5 -— o CB B1 "C
"|| 2 llfS 5
S-8-o-g « a 8 | S 'g
= 2 § 5 I •§ ? ? 1 e
**
"5 >^ *
Q> ^ <-
t E <»»;•£
1 Si tllil
2 IS SjJ|*-s
m *2 ^ScSw
Q.^^ (Oq> "colOirjO
§ o fi * |« «£ •
i! 11 ?i!«|
|5 .08 o^S.f»
!!
"5.-o -c-c CBCBI
Ibl £ I e 8 S -8
I I!|i f! 11,1 ififl
^ -55^0 S° vs.2'0 £„»£«
^ o.*"^eo(B *- ^^=c n"*Offl
•i SiSJt'l as8'" l^t!
.2 '1§^|S1.£ .2.11.0 fi-S-l-i
fit 17! ill Ullillllii
g^g •«-2£5^'-'(B ^oa^S^cSSioi
O
Z
a
J-5
-------
g
to
o
U
•o
re
I
(0
w
35
0 •
E=B
3 S>
» 2
o .£
*., »
12 «
C 0
0
«
3 |
O §
£ o
=
-o
« §
c -
« g
Q o
ll
0
3
(B
.1:1*
9
£ a. .. -a
o> 3
^ O
o> _
o> *- o .E
J-6
-------
w
o
a.
E
o
O
a
° **
f -n ™ O> OS
i § £1.1.
g ° £E £
o S § S S
S > 3 ~ £•£ s 5 | «•=
*l«Sl-s . sS'i|£
.g ? .g — J •£ 10^ oc- |
o> < i
O> DL 3
•- ui tr
(O ^ >
o •& —
05 — co
c * CN
.£ o *-
g I i ? ?
E § o '•& =5.
g « g-E E
E
?
'•=>-«
9--Q 7i
111 I
o « w° "
o ag
oca
""S?^-8S
S€sfslls
•g g -a I -5 -S I a
_. S *- * an * ?t
•S w '«
W oj AV)0Ca>cO'Q
isT-cis w -"IE s>- =
w^^w^gWTSWWflj
5 5£ 9-®^ nia^J-S
Oe-— 'Cc"-Sco)i3c5
S E -^ •;: .wES«i--r-*m —
x:
««
c ~ g c S .o>c -og
1111 511| 1111 n
O (0 A "O ic x O CQ £ O CQ
O *-* O JJQ*^ r>*j Ort**
c wu*-%=o=.£:«i>uc »
^1
UJ|
o "«
o .S
CM u.
J-7
-------
Table J-2. Comparison of Procedures for Visual Inspections from HUD and EPA Clearance
Guidance and Rule Documents.
Clearance Guidance Document 'i
-. *7_--l~ .a* .. ^""V/JTr •>.'*'*' "-." '-"-'-I'- 'VJ" > •" ""
1 990 HUD Interim Guidelines [2]
1995 EPA 403 Guidance [5]
1995 HUD Guidelines [3]
1999 HUD 101 2/1 01 3 Final
Rule [22]
1996 EPA 402/404 Rule [4]
2001 EPA 403 Rnal Rule [1]
KMi^ftfC^
Perform visual inspection after abatement but before repainting.
Surface dust sampling should not be conducted if there is a visible
accumulation of dust or debris
All interior rooms or areas and exterior areas should be visually clean
before collecting dust samples. If this is not the case, clean the rooms
and areas before starting dust collection for clearance testing
Room by room while environmental samples taken.
Prior to repainting
No evidence of settled dust
The visual assessment shall be performed to determine if deteriorated
paint surfaces and/or visible amounts of dust, debris, paint chips or other
residue are still present. Both exterior and interior painted surfaces shall
be examined for the presence of deteriorated paint. If deteriorated paint
or visible dust, debris or residue are present in areas subject to dust
sampling, they must be eliminated prior to the continuation of the
clearance examination, except elimination of deteriorated paint is not
required if it has been determined, through paint testing or a lead-based
paint inspection, that the deteriorated paint is not lead-based paint. If
exterior painted surfaces have been disturbed by the hazard reduction,
maintenance or rehabilitation activity, the visual assessment shall include
an assessment of the ground and any outdoor living areas close to the
affected exterior painted surfaces. Visible dust or debris in living areas
shall be cleaned up and visible paint chips on the ground shall be
removed.
A visual inspection shall be performed to determine if deteriorated
painted surfaces and/or visible amounts of dust, debris or residue are still
present. If deteriorated painted surfaces or visible amounts of dust,
debris or residue are present, these conditions must be eliminated prior to
the continuation of the clearance procedures.
Addressed in the 1996 EPA 402/404 Rule [4], 40 CFR §745.227 (e)(8)(i).
J-8
-------
Table J-3. Comparison of Procedures for Sealants from HUD and EPA Clearance Guidance
and Rule Documents.
•:-Ji*W«i; J..-. * • "-. ••*V^!-3:.'."^I->>-'_V''V^^*ii"_v'. »•"• •«-• ; • -' 'Vi jV^-i^V"
|^C(eara^;-Gu^iweDpaimentjg^
1990 HUD Interim Guidelines [2]
1995 EPA 403 Guidance [5]
1995 HUD Guidelines [3]
1999 HUD 1012/1013 Rnal Rule [22]
1996 EPA 402/404 Rule [4]
2001 EPA 403 Final Rule [1]
Painting or otherwise sealing abated surfaces and all interior
floors is the next step of cleaning process.
All abated surfaces, including walls, ceilings, and woodwork,
should be primed with an appropriate primer. All applicable areas
may then be repainted. Wooden floors, vinyl tile, linoleum, and
concrete floors should be sealed.
Document does not address the use of sealants for clearance.
Seal floors before clearance testing
Document does not address the use of sealants for clearance.
Document does not address the use of sealants for clearance.
Document does not address the use of sealants for clearance.
Table J-4. Comparison of Procedures for Waiting Time from HUD and EPA Clearance
Guidance and Rule Documents.
1990 HUD Interim Guidelines 12}
Dust sampling should take place no sooner than 24 hr. after
completion of post-abatement cleanup activities.
1995 EPA 403 Guidance [5]
Sampling of dust should take place at least one hour after
completion of all abatement and interim control work, including
cleanup.
1995 HUD Guidelines [3]
Wait one hour after cleaning.
1999 HUD 1012/1013 Final Rule
[22]
In accordance with 40 CFR 745.227(e)(8). This is part of the 1996
EPA 402/404 Rule (41.
1996 EPA 402/404 Rule [4]
Dust samples for clearance purposes shall be taken a minimum of 1
hour after completion of final post-abatement cleanup activities.
2001 EPA 403 Final Ruled]
Addressed in the 1996 EPA 402/404 Rule [4], 40 CFR §745.227
J-9
-------
Table J-5. Comparison of Clearance Standards from HUD and EPA Clearance Guidance
and Rule Documents.
^-Clearance Guidance Document ^|e
1 990 HUD Interim Guidelines [2]
1995 EPA 403 Guidance [5]
1995 HUD Guidelines (3}
1996 EPA 402/404 Rule [4]«
1999 HUD 1012/1013 Final Rule
[221
2001 EPA 403 Rnal Rule [1]
200 //g/ft2 for bare floors
500 jt/g/ft2 for window sills
800 //g/ft* for window wells (troughs)
100 j/g/ft2 for bare floors
500 //g/ft2 for interior window sills
800 jug/ft2 for exterior window sills (troughs) and exterior horizontal
surfaces
1 00 //g/ft2 for floors (including carpeted and uncarpeted floors)
500 //g/ft2 for window sills,
800 j/g/ft* for window wells (troughs) and exterior concrete or
other rough surfaces.
Clearance levels which are appropriate for the purposes of this
section may be found in the EPA Guidance on Residential Lead-
Based Paint, Lead-Contaminated Dust and Lead Contaminated Soil
or other equivalent guidelines. (5, 23]
40 jug/ft2 for floors (carpeted or uncarpeted interior floors),
250 fjg/it2 for interior window sills,
800 #g/ft2 for window troughs
40 A/g/ft2 for floors (including carpeted and uncarpeted floors),
250 //g/ft2 for interior window sills,
400 //g/ft2 for window troughs.
1998 EPA 403 Proposed Rule [24] included proposed amendments to the final Section 402 rules for post-
abatement dust clearance standards: 50 i/g/ft2 for uncarpeted floors, 250 //g/ft2 for interior window sills,
and 800 /sg/ft2 for window troughs.
J-10
-------
Table J-6. Comparison of Failure Procedures from HUD and EPA Clearance Guidance and
Rule Documents.
1990 HUD Interim
Guidelines [2]
If any of the residual lead dust level results exceeds the clearance criteria, the area must be
cleaned again and re-tested until the criteria are met. All areas must pass clearance to be re-
occupied.
1995 EPA 403
Guidance [5]
Samples above or equal to the appropriate standard have failed clearance, and all rooms or areas
represented by those samples are said to have failed. For samples that have failed, the
components represented by those samples (floors, interior window sills, exterior window sills,
exterior horizontal surfaces, or interior areas outside a containment area) must be recleaned and
retested. The process continues until clearance is obtained for all components.
1995 HUD Guidelines
I3J
If the dust sample for any surface is above the standard, all similar surfaces in the dwelling that
sample represents (e.g., all interior sills or floor) should be re-cleaned and re-tested. Only the
similar component needs to be re-cleaned. If a surface fails twice additional LHC measures
and/or further sealing should be considered.
1999 HUD
101 2/1 01 3 Final Rule
[22]
All surfaces represented by a failed clearance sample shall be recleaned or treated by hazard
reduction, and retested, until the applicable clearance level is met.
1996 EPA 402/404
Rule (4]
If the residual lead levels in a dust sample exceed the clearance levels, all the components
represented by the failed sample shall be re-cleaned or treated by lead hazard reduction and re-
tested until clearance levels are met.
2001 EPA 403 Final
Ruled]
If a property fails clearance, it must be recleaned until it passes, although it is not automatically
necessary to reclean the entire property when clearance fails, such as when some of the visual
and dust-testing clearance results have indicated that portions of the property are already
clearned.
J-11
-------
Table J-7. Comparison of Concentrations versus Loadings from HUD and EPA Clearance
Guidance and Rule Documents.
nents.
£||fe£|||^
~ - • . ling method which is currently recommended is surface wipe sampling
iCtoarance Guidance
' '
1990 HUD Interim
Guidelines 1 2]
The surface dust sampling method which is currently recommended is surface wipe sampling
(i.e., loadings).
1995 EPA 403
Guidance [5]
The basic method for collecting dust clearance samples is the wipe method (i.e., loadings).
Other dust collection methods may be used provided the user establishes comparability to the
wipe method and is responsible for providing comparable standards for clearance.
Note that when assessing multiple sources of lead, dust lead concentration may be a more
appropriate measurement. The utility of concentration measurements for identifying hazardous
room dust will be further considered in the development of section 403 rule making
1995 HUD Guidelines
[3]
Until the EPA standards and protocols are established, wipe sampling (loadings) should be
performed on all surfaces. While vacuum samples (concentrations) can be collected, neither
HUD nor EPA can provide standards to interpret vacuum sampling results at this time. Until
vacuum sampling standards have been established, wipe sampling is the preferred method.
1999 HUD 1012/1013
Final Rule [22].
Standards are reported as a loading for interior samples.
1996 EPA 402/404
Rule [4]
The type of dust samples to be collected, wipe (loading) or vacuum (concentration) are not
explicitly defined in this document.
2001 EPA 403 Final
Ruled]
Dust-lead hazard and clearance standards are set as in loadings. In 40 CFR 5745.65, it states
"A dust-lead hazard is surface dust in a residential dwelling or child-occupied facility that
contains a rnass-per-area concentration of lead equal to or exceeding 40 //g/ft2 on floors or
250 /ig/ft2 on interior window sills based on wipe samples." Dust-lead clearance standards are
specified in 40 CFR §745.227 (e)(8)(viii) which were shown in Table J-5 [41.
J-12
-------
Table J-8. Comparison of Compositing Rules from HUD and EPA Clearance Guidance and
Rule Documents.
^Clearance Guidance a
-r*3*3ii-'£--'*!*--- • -'"'< - -•: •-, •**:*•
ijsyKDocument^; -:•-<**
1990 HUD Interim
Guidelines [2!
1994/1 995 EPA 403
Guidance [5]
1995 HUD Guidelines
[31
1999 HUD
101 2/1 01 3 Rnal Rule
[221
1996 EPA 402/404
Rule [41*
2001 EPA 403 Final
Rule [11
||l^^!^a^^
Compositing was not addressed in this document.
Compositing was not addressed in this document.
Samples are composited in the field. Separate samples are required from carpeted and hard
surfaces; and from component sampled; and from each dwelling floor surface areas sampled in
each room should be the same size. Interior sill and well sampling sizes are dependent on
window characteristics, but should be similar from room to room; all subsamples should be
inserted into the same tube. No more than four different wipes should be inserted into a single
container.
Dust samples shall be collected and analyzed in accordance with standards established either by
a State or Indian tribe under a program authorized by EPA in accordance with 40 CFR part 745,
subpart Q, or by the EPA in accordance with 40 CFR 745.227 (both are part of the 1996 EPA
402/404 Rule 14]).
Composites expressly permitted for clearance testing. Composite dust samples consist of at
least two subsamples. Every component that is being tested shall be included in the sampling.
Composite dust samples shall not consist of subsamples from more than one type of component.
A composite sample may contain from two to four subsamples of the same area as each other
and of each single surface sample in the composite [40 CFR §745.63]
1998 EPA 403 Proposed Rule [24] included proposed amendments to the final Section 402 rules includes requiring the
risk assessor to compare the composite sample to the clearance standard divided by the number of subsamples in the
composite.
J-13
-------
Table J-9. Comparison of Conflict of Interest Issues from HUD and EPA Clearance
Guidance and Rule Documents.
-[Clearance Guidance •:;;
f^i^*hte::f
1990 HUD Interim
Guidelines [2]
1994/1 995 EPA 403
Guidance [51
1995 HUD Guidelines
[3]
1999 HUD 101 2/1 01 3
Final Rule [22]
1996 EPA 402/404
Rule (41
2001 EPA 403 Final
Ruled)
. ffi.\ :r> V * ; - ^r'> ':,-*-. (•:• i^f. . .* ., '"'' .'•., ; fc; - „• • "„• •• — ;1: • ?5a.J^ *" ' "V'i *•-.' '-; •" - •.•>".,--' -•'~.:r .'V - '.', .•„•>•'•. -^!t' '>:• . ' * '-,--' ™%V?*<*jB2s:?V3'*'0"*'^!''t ;«*»''
To avoid potential conflict of interest, the abatement contractor should not conduct the final
inspection. This should be done by a qualified inspector, industrial hygienist, or local public
health official.
Clearance testing should be conducted by an organization that is independent of the
organization that completed the abatement or interim controls.
Inspectors should be independent of abatement contractor
Clearance examinations shall be performed by persons or entities independent of those
performing hazard reduction or maintenance activities, unless the designated party uses
qualified in-house employees to conduct clearance. An in-house employee shall not conduct
both a hazard reduction or maintenance activity and its clearance examination.
EPA requested comment on whether to preclude individuals or firms conducting abatement
activities from performing inspection and risk assessment activities, and from performing
clearance procedures following an abatement. Although many public commentators supported
a requirement that inspection, risk assessment and clearance procedures be conducted by
individuals and firms independent of the individuals and firms conducting abatements, the final
rule does not include such a requirement. Some of the reasons for not supporting a conflict-of-
interest requirement were that the potential convenience and cost savings of hiring one firm, as
opposed to two or three firms should not be denied a property owner. The Agency also noted
that there may be instances in which, due to a regional scarcity of lead-based paint
professionals, it may be cost prohibitive or logistically difficult for a building owner to hire two
different companies. Nonetheless, the Agency believes that parties involved in lead-based paint
activities should avoid situations of potential conflict of interest.
Addressed in the 1996 EPA 402/404 Rule [4].
J-14
-------
REPORT DOCUMENTATION PAGE
Form Approved
OMB No 0704-0188
Sources, gathering and maintaining die data needed, and completing and reviewing the collection of information. Send comments regarding this burden estimate or any other
aspect of this collection of information, including suggestions for reducing this burden, to Washington Headquarters Services, Directorate for information Operations and
Reports, 1215 Jefferson Davis Highway, Suite 1204, Arlington, VA 22202-4302, and to the Office of Management and Budget, Paperwork Reduction Project (07044188),
Washington, DC 20S03.
1. AGENCY USE ONLY (Leave blank)
2. REPORT.DATE
December 2001
3. REPORT TYPE AND DATES COVERED
Final Report
4. TITLE AND SUBTITLE
. Analysis of Lead Dust Clearance Testing
6. AUTHOR(s)
Bradley Skarpness, Ying-Liang Chou, and Warren Strauss
5. FUNDING NUMBERS
C: 68-D5-0008,68-W-99-033
7. PERFORMING ORGANIZATION NAME(s) AND ADDRESS(ES)
Battelle Memorial Institute
505 King Avenue
Columbus, Ohio 43201
8. PERFORMING ORGANIZATION
REPORT NUMBER
Not Applicable
9. SPONSORING/MONITORING AGENCY NAME(S) AND ADDRESSfES)
U.S. Environmental Protection Agency
Office of Pollution Prevention and Toxics
1200 Pennsylvania Avenue NW (7401)
Washington, D.C. 20460
10. SPONSORING/MONITORING AGENCY
REPORT NUMBER
EPA 747-R-01-005
11. SUPPLEMENTARY NOTES
Other Battelle staff involved in the production of this report included Jennifer Holdcraft, Pamela Hartford, and Matt
Palmgren.
12.a DISTRIBUTON/AVAILABILITY STATEMENT
12b. DISTRIBUTION CODE
13. ABSTRACT (Maximum 200 words)
This report presented results from lead clearance testing activities that occurred between 1989 and 1999 under the 1990
HUD Interim Guidelines, which set dust lead clearance standards at 200, 500, and 800 ng/ft2 for floors, window sills, and
window troughs, respectively. Dust lead clearance results were obtained from the HUD's FHA and PHA Demo Studies,
and Grantee Program; the Maryland Department of the Environment; the Atlantic City and Dover Housing Authorities; and
the Cleveland Lead Hazard Abatement Center. Over 90% of individual clearance samples passed clearance. Only 67% of
the housing units passed clearance on the first site visit. Eventually 87% of the housing units were known to have passed
clearance. Floor samples had a smaller geometric mean lead loadings than sills, which in turn had a smaller geometric
mean than troughs. Geometric mean lead loadings generally increased from the first site visit to the third site visit. The
pairwise correlations between components during the first site visit were positive and significant. The simulated composite
samples analysis results indicated mat composite sampling is associated with a decrease of sensitivity if compared directly
to the standards. Comparison of composite samples to two lower standards resulted in an increase in sensitivity.
14. SUBJECT TERMS
Clearance Testing, Clearance Standards, Lead Loadings, Floors, Window Sills, Window
Troughs, Composite Samples, Geometric Mean, Performance Characteristics (Sensitivity,
Specificity, PPV, NPV), Correlation, Conditional Probability, Logistic Regression.
15. NUMBER OF PAGES
374
16. PRICE CODE
17. SECURITY CLASSIFICATION
OF REPORT
Unclassified
18. SECURITY CLASSIFICATION
OF THIS PAGE
Unclassified
19. SECURITY
CLASSIFICATION
OF ABSTRACT
Unclassified
20. LIMITATION OF
ABSTRACT
NSN 7540-01-280-5500
Standard Form 298 (Rev 2-89)
Prescribed by ANSI Std. Z39-18
298-102
-------
-------