United States
hi	Environmental Protection Agency

Office of Chemical Safety and
Pollution Prevention

Final Risk Evaluation for
1,4-Dioxane

Systematic Review Supplemental File:

Data Quality Evaluation of Epidemiological

Studies

CASRN: 123-91-1

December 2020


-------
Table Listing

1	Young 1977: Evaluation of ADME/PBPK Outcomes 		2

2	Young 1977: Evaluation of Irritation Outcomes		5

3	Union Carbide 1989: Evaluation of Cancer Outcomes		8

4	Garcia et al. 2015: Evaluation of Cancer Outcomes 		12

This document presents data quality evaluation results for epidemiological studies evaluated
for the Risk Evaluation for 1,4-Dioxane.

EPA's Office of Pollution Prevention and Toxics (OPPT) developed data quality criteria for
epidemiological studies. The first version of the criteria was documented in the Application of
Systematic Review in TSCA Risk Evaluations document (EPA Document #740-Pl-8001). The
initial criteria were updated as described in the supplemental file Final Risk Evaluation for
1,4-Dioxane Systematic Review Supplemental File: Updates to the Data Quality Criteria for
Epidemiological Studies.

1


-------
Table 1: Young 1977: Evaluation of ADME/PBPK Outcomes

Study Citation: Young, JD; Braun, WH; Rampy, LW; Chenoweth, MB; Blau, GE (1977).

Toxicology and Environmental Health, 3(3,3), 507-520
Data Type:	Dow_volunteers_14D_TK_Half-life_Urine- ADME/PBPK

HERO ID:	62956

Pharmacokinetics of 1,4-dioxane in humans Journal of

Domain

Metric

Rating^ MWF* Score

Comments^

Domain 1: Study Participation

Metric 1: Participant selection

Medium x 0.5

Metric 2: Attrition

Metric 3: Comparison Group

Medium x 0.5

Not Rated NA

NA

Some key elements of the study design were not
present and a limited number of subjects were se-
lected for the study raises the potential for selec-
tion bias. Specifically, the study was conducted on
4 Caucasian male volunteers comprised of healthy
scientists and business men ranging in age from 40-
49. Due to the low number of participants it is un-
clear whether the study population is likely to be
representative of the exposure-outcome distribution
of the population of persons eligible for inclusion.

No attrition. Metabolite used for TK model
(HEAA) was not determined in the plasma of 2-3
participants due to poor ability to separate from an-
other chemical.

Comparison group not relevant for TK model.
Subjects provided history and underwent exten-
sive physical examination, chest x-ray, electrocardio-
gram, blood chemistry panel, and urine analysis. All
tests were repeated 24 hrs and 2 weeks after expo-
sure. Results were not presented, but qualitatively
stated to be healthy.

Domain 2: Exposure Characterization

Metric 4: Measurement of Exposure

High

x 0.4

Metric 5: Exposure levels

Medium x 0.2

0.4 Controlled dosage study. Subjects exposed in a con-
trolled airflow chamber with 1,4-dioxane concentra-
tion of 48-52 ppm. Concentration in 3 breathing
zones confirmed using a Wilks Miram I IR analyzer
(8.75 um wavelength, standard curve). Exposure
lasted for 6 hrs. Plasma concentrations indicated
a dosage of 5.4 +/- 1.1 mg/kg.

0.4 Blood plasma reached a plateau concentration dur-
ing the study (~4hrs into exposure). Plasma con-
centrations indicated a dosage of 5.4 +/- 1.1 mg/kg.
Multiple levels of exposure not relevant for this
study, but exposure sufficiently high to determine
TK parameters.

Continued on next page . ..

2


-------
. continued from previous page

Study Citation: Young, JD; Braun, WH; Rampy, LW; Chenoweth, MB; Blau, GE (1977).

Toxicology and Environmental Health, 3(3,3), 507-520
Data Type:	Dow_volunteers_14D_TK_Half-life_Urine- ADME/PBPK

HERO ID:	62956

Pharmacokinetics of 1,4-dioxane in humans Journal of

Domain

Metric

Rating^ MWF* Score

Comments^

Metric 6: Temporality

High	X 0.4	0.4 Plasma collection started 30 minutes after exposure

began and continued for another 6 hrs. Urine collec-
tion throughout exposure and for the following. Eye
irritation and smell sensitization evaluated through-
out exposure.

Domain 3: Outcome Assessment

Metric 7: Outcome measurement or characterization High

Metric 8: Reporting Bias

High

X 0.667 0.67 Venous blood was drawn every hour beginning 30
minutes after initial exposure. Blood samples were
collected for 12.5 hrs after initial exposure, yielding
13 time points. Urine was collected for the 6 hrs
(during exposure), in 2 hr intervals for the next 10
hrs, then from 16-24, 24-36, and 36-48 hrs. 14D
levels in each were determined using GC/MS.

X 0.333 0.33 Plasma 14D presented as means/standard devia-
tions, and plasma presented as means alone for
HEAA metabolite. Urine concentrations of 14D
and HEAA presented for each individual and with
mean/standard deviation. All parameters in the TK
model and half-lives fully presented.

Domain 4: Potential Counfounding/Variable Control
Metric 9: Covariate Adjustment

Metric 10: Covariate Characterization
Metric 11: Co-exposure Confounding

Not Rated NA

Not Rated
Medium

NA
x 1

NA No covariates were adjusted for in the TK mod-
els, which is appropriate when trying to represent
a larger population. Minimal variation in SES ex-
pected (based on job titles). All Caucasian males
ages 40-49.

NA Covariates determined from interviews and physi-
cals.

2	No co-exposures expected. Participants experienced

identical exposure scenario, but previous history not
detailed. As some participants were scientists work-
ing at DOW, previous co-exposures are likely. How-
ever, not relevant to the current TK analysis.

Domain 5: Analysis

Metric 12: Study Design and Methods

Metric 13: Statistical power

Medium	X 0.4	0.8 Study exposed 4 volunteers to 14D and monitored

concentrations of 14D and its primary metabolite
in blood plasma and urine over the course of 2 days
to create a one-compartment toxicokinetic model for
14D . Study design appropriate for TK models, but
not for health outcomes (eye irritation).

Medium X 0.2	0.4 Only 4 participants. Statistical power not stated,

but able to establish TK parameters with moderate
standard deviations.

Continued on next page

3


-------
. continued from previous page

Study Citation: Young, JD; Braun, WH; Rampy, LW; Chenoweth, MB; Blau, GE (1977). Pharmacokinetics of 1,4-dioxane in humans Journal of

Toxicology and Environmental Health, 3(3,3), 507-520
Data Type:	Dow_volunteers_14D_TK_Half-life_Urine- ADME/PBPK

HERO ID:	62956

Domain

Metric

Rating^ MWF* Score

Comments^

Metric 14:

Reproducibility of analyses

Medium

X

0.2

0.4

Calculations used for the models are clear and fully
presented in tables/figures. All data needed to re-
created provided.

Metric 15:

Statistical models

Medium

X

0.2

0.4

One-compartment toxicokinetic model developed for
14D using nonlinear parameter estimates. Model pa-
rameters obtained per subject, such that standard
deviations of individuals would reflect wider popu-
lation..

Domain 6: Other Considerations for Biomarker Selection and Measurement











Metric 16:

Use of Biomarker of Exposure

Low

X

0.167

0.5

14D and primary metabolite b-hydroxyethoxyacetic
acid (HEAA) were determined. HEAA was only de-
termined in 3/4 of the participants (due to inter-
ference - not further explained). Study served as
a means of determining a quantitative relationship
between 14D dose and plasma/urine concentrations.
Precision and accuracy of measurement technique
not reported.

Metric 17:

Effect biomarker

Not Rated

NA

NA



Metric 18:

Method Sensitivity

Medium

X

0.167

0.33

14D detected in all samples. HEAA had some inter-











ferences for plasma. LOD 0.1-0.2 ug/ml for 14D in
plasma and urine. LOD for HEAA 1 ug/ml in urine
and 2-10 ug/ml in plasma.

Metric 19:

Biomarker stability

Low

X

0.167

0.5

Storage history and stability not stated.

Metric 20:

Sample contamination

Low

X

0.167

0.5

Contamination not discussed, but not anticipated.

Metric 21:

Method requirements

High

X

0.167

0.17

Instrumentation that provides unambiguous identi-
fication and quantitation of the biomarker at the re-
quired sensitivity (GC-MS).

Metric 22:

Matrix adjustment

Low

X

0.167

0.5

Creatinine levels determined in blood plasma and











urine, but not clear if adjustments were made. Study
only provides results using one method.

Overall Quality Determination1



Medium





1.8



Extracted



Yes









MWF = Metric Weighting Factor

High = 1; Medium = 2; Low = 3; Unacceptable = 4; N/A has no value.

The overall rating is calculated as necessary. EPA may not always provide a comment for a metric that has been categorized as High.

Overall rating =

]T\ (Metric Score; X MWF;) / J] . MWFj

if any metric is Unacceptable
(round to the nearest tenth) otherwise

where High => 1 to < 1.7; Medium => 1.7 to < 2.3; Low => 2.3 to < 3.0. If the reviewer determines that the overall rating needs adjustment, the original rating is
crossed out and an arrow points to the new rating.

^ This metric met the criteria for high confidence as expected for this type of study^


-------
Table 2: Young 1977: Evaluation of Irritation Outcomes

Study Citation: Young, JD; Braun, WH; Rampy, LW; Chenoweth, MB; Blau, GE (1977). Pharmacokinetics of 1,4-dioxane in humans Journal of
Toxicology and Environmental Health, 3(3,3), 507-520

Data Type:	Dow	volunteers	14D	Eyelrritation	SmellSensitization-Irritation

HERO ID:	62956

Domain

Metric



Rating^

MWF*

Score

Comments^

Domain 1: Study Participation















Metric 1:

Participant selection



Medium

X

0.4

0.8

Some key elements of the study design were not
present and a limited number of subjects were se-
lected for the study raises the potential for selec-
tion bias. Specifically, the study was conducted on
4 Caucasian male volunteers comprised of healthy
scientists and business men ranging in age from 40-
49. Due to the low number of participants it is un-
clear whether the study population is likely to be
representative of the exposure-outcome distribution
of the population of persons eligible for inclusion.

Metric 2:

Attrition



High

X

0.4

0.4

No attrition.

Metric 3:

Comparison Group



High

X

0.2

0.2

Table 1 indicates characteristics of the 4 subjects
were generally similar, although there were varia-
tions in urine flow rate (range: 1.14 - 2.74 ml/min)
and weight (range: 74.5 - 100.75 kg)

Domain 2: Exposure Characterization













Metric 4:

Measurement of Exposure



High

X

0.4

0.4

Controlled dosage study. Subjects exposed in a con-
trolled airflow chamber with 1,4-dioxane concentra-
tion of 48-52 ppm. Concentration in 3 breathing
zones confirmed using a Wilks Miram I IR analyzer
(8.75 um wavelength, standard curve). Exposure
lasted for 6 hrs. Plasma concentrations indicated
a dosage of 5.4 +/- 1.1 mg/kg.

Metric 5:

Exposure levels



Low

X

0.2

0.6

Same individuals served as unexposed and exposed
group (physical before/after exposure).

Metric 6:

Temporality



High

X

0.4

0.4

Plasma collection started 30 minutes after exposure
began and continued for another 6 hrs. Urine collec-
tion throughout exposure and for the following. Eye
irritation and smell sensitization evaluated through-
out exposure.

Domain 3: Outcome Assessment













Metric 7:

Outcome measurement or

characterization

Low

X

0.667

2

The outcome assessment method is an insensitive
measure: eye irritation and the loss of sensitivity to
the smell of dioxane were self-reported.





Continued

on next page ..









5


-------
.. . continued from previous page

Study Citation: Young, JD; Braun, WH; Rampy, LW; Chenoweth, MB; Blau, GE (1977). Pharmacokinetics of 1,4-dioxane in humans Journal of
Toxicology and Environmental Health, 3(3,3), 507-520

Data Type:	Dow	volunteers	14D	Eyelrritation	SmellSensitization-Irritation

HERO ID:	62956

Domain

Metric

Rating^

MWF*

Score

Comments^

Metric 8:

Reporting Bias

Medium

x 0.333

0.67

No specific result (e.g., frequency) presented on eye
irritation other than the comment 'Eye irritation
was a frequent complaint throughout exposure'.

Domain 4: Potential Counfounding/Variable Control









Metric 9:

Covariate Adjustment

Medium

x 0.5

1

Participants served as own controls for the eye irri-
tation. Minimal variation in SES expected (based
on job titles). All Caucasian males ages 40-49.

Metric 10:

Covariate Characterization

Medium

x 0.25

0.5

Covariates determined from interviews and physicals
with no method validation against well-established
methods.

Metric 11:

Co-exposure Confounding

Medium

x 0.25

0.5

No co-exposures expected. Participants experienced
identical exposure scenario, but previous history not
detailed. As some participants were scientists work-
ing at DOW, previous co-exposures are likely. How-
ever, not relevant to the current TK analysis.

Domain 5: Analysis











Metric 12:

Study Design and Methods

Medium

x 0.5

1

Study exposed 4 volunteers to 14D in an controlled
experiment. Monitored irritations and smell sensiti-
zation during the experiment. Conducted full phys-
icals before and after. Smell sensitization results
were descriptive.

Metric 13:

Statistical power

Unacceptable

x 0.25

0.06

Only 4 participants.

Metric 14:

Reproducibility of analyses

Medium

x 0.25

0.5

The study did not use a statistical model.

Metric 15:

Statistical models

Not Rated

NA

NA

No statistical models were used in the study.

Domain 6: Other Considerations for Biomarker Selection and Measurement









Metric 16:

Use of Biomarker of Exposure

Low

x 0.167

0.5

14D and primary metabolite b-hydroxyethoxyacetic
acid (HEAA) were determined. HEAA was only de-
termined in 3/4 of the participants (due to inter-
ference - not further explained). Study served as
a means of determining a quantitative relationship
between 14D dose and plasma/urine concentrations.
Precision and accuracy of measurement technique
not reported.

Metric 17:

Effect biomarker

Not Rated

NA

NA



Metric 18:

Method Sensitivity

Medium

x 0.167

0.33

14D detected in all samples. HEAA had some inter-









ferences for plasma. LOD 0.1-0.2 ug/ml for 14D in
plasma and urine. LOD for HEAA 1 ug/ml in urine
and 2-10 ug/ml in plasma.

Metric 19:

Biomarker stability

Low

x 0.167

0.5

Storage history and stability not stated.

Metric 20:

Sample contamination

Low

x 0.167

0.5

Contamination not discussed, but not anticipated.

Continued on next page .. .

6


-------
... continued from previous page

Study Citation: Young, JD; Braun, WH; Rampy, LW; Chenoweth, MB; Blau, GE (1977). Pharmacokinetics of 1,4-dioxane in humans Journal of

Toxicology and Environmental Health, 3(3,3), 507-520
Data Type:	Dow_volunteers_14D_EyeIrritation_SmellSensitization-Irritation

HERO ID:	62956

Domain

Metric

Rating^

MWF*

Score

Comments^



Metric 21: Method requirements
Metric 22: Matrix adjustment

High
Low

x 0.167
x 0.167

0.17
0.5

Instrumentation that provides unambiguous identi-
fication and quantitation of the biomarker at the re-
quired sensitivity (GC-MS).

Creatinine levels determined in blood plasma and
urine, but not clear if adjustments were made. Study
only provides results using one method.

Overall Quality Determination1"

Unacceptable**



1.9



Extracted



No









* Consistent with our Application of Systematic Review in TSCARisk Evaluations document, if a metric for a data source receives a score of Unacceptable (score = 4), EPA
will determine the study to be unacceptable. In this case, one or more of the metrics were rated as unacceptable. As such, the study is considered unacceptable and the score
is presented solely to increase transparency.

MWF = Metric Weighting Factor

High = 1; Medium = 2; Low = 3; Unacceptable = 4; N/A has no value.

The overall rating is calculated as necessary. EPA may not always provide a comment for a metric that has been categorized as High.

Overall rating =

]T\ (Metric Score; X MWF;) / J] . MWFj

if any metric is Unacceptable
(round to the nearest tenth) otherwise

where High => 1 to < 1.7; Medium => 1.7 to < 2.3; Low => 2.3 to < 3.0. If the reviewer determines that the overall rating needs adjustment, the original rating is crossed
out and an arrow points to the new rating.

^ This metric met the criteria for high confidence as expected for this type of study

7


-------
Table 3: Union Carbide 1989: Evaluation of Cancer Outcomes

Study Citation: Union Carbide (1989). Lymphatic and hematopoietic tissue cancer in a chemical manufacturing environment with attached tables and

cover letter dated 02/21/89
Data Type:	occupational 1,4-D, lymphatic & hematopoietic cancer-Cancer

HERO ID:	597727

Domain

Metric

Rating^

MWF* Score

Comments^

Domain 1: Study Participation

Metric 1: Participant selection

High

Metric 2: Attrition

Low

Metric 3: Comparison Group

Low

X 0.4	0.4 Subjects were part of a large cohort mortality study

in two Union Carbide Corporation chemical manu-
facturing facilities and a research and development
center. This case-control study selected cases of four
distinct subcategories of lymphatic and hematopoi-
etic tissue cancers. Study was restricted to men be-
cause only 4 cases were identified in women. Con-
trols were selected from the total employee cohort.
Participation rates are not a concern because all in-
formation was obtained via records. Controls were
randomly selected and all cases (follow-up was avail-
able for 96%) were included indicating that selection
into or out of the study was not likely to be biased.

X 0.4	1.2 Vital status at follow- up was complete for 96% of

the 29,139 men in the cohort. It was noted that 5
controls were selected per case. Based on the 129
cases identified this would suggest 645 controls se-
lected. However, the study report does not indicate
how many controls were included in the study nor
does it report the numbers of controls in the different
work areas or chemical exposures. There is insuffi-
cient information provided on the control numbers
during important stages of the study to determine if
there was any attrition.

x 0.2	0.6 It is unclear that the controls were free of the out-

comes. The study authors did not provide baseline
characteristics for the subjects to determine if the
cases and controls were similar. Analysis only ad-
dressed age (and only males were used), but no other
potential differences were addressed. Controls were
selected from the total employee cohort according
to an unmatched incidence density sampling design.
It was noted that there were 5 controls selected per
case, but other than that the number of controls
was not mentioned. Time of hire to death for cases
was categorized into five year increments of survival.
Controls were selected such that they were first em-
ployed in the same decade and survived to the same
five year survival period as the case.

Continued on next page

8


-------
. continued from previous page

Study Citation: Union Carbide (1989). Lymphatic and hematopoietic tissue cancer in a chemical manufacturing environment with attached tables and

cover letter dated 02/21/89
Data Type:	occupational 1,4-D, lymphatic & hematopoietic cancer-Cancer

HERO ID:	597727

Domain

Metric

Rating^

MWF* Score

Comments^

Domain 2: Exposure Characterization

Metric 4: Measurement of Exposure

Metric 5:
Metric 6:

Exposure levels
Temporality

Medium	X 0.4	0.8 Exposure was based on job assignment and poten-

tial exposure, and classified on an ever/never basis.
Ever exposure was based on working 1 or more days
with the chemical. Details were stated to be avail-
able in Ott et al., 1989 (HERO ID 104202), which
provides more details on the definition of work areas
and environmental agents. All workplace exposures
were subdivided into six major categories. Using
departmental and job assignment records and his-
torical information regarding process dates and de-
scriptions from 1925-1978, all work activities were
further subdivided into 111 distinct and mutually
exclusive work areas. Exposure to each work area or
activity was based on the work history information
for each subject. 1,020 substances were identified
as having been used or produce in one or more of
the production units over the 54 years. Potential
employee contact was based on assignment to a de-
partmental unit, which implied potential exposure
to any chemical in use during the time period of the
employee's assignment to that unit. 21 substances
were selected for analysis. Because 1,4-dioxane did
not have more than 4 cases, it was not evaluated by
duration of exposure.

Low	X 0.2	0.6 Exposure was ever/never

Medium	X 0.4	0.8 Temporality is established, but it is unclear whether

exposures fall within relevant exposure windows for
the outcome of interest. In the event that exposures
which occurred close to the time of death were unre-
lated to outcome, the data were also analyzed with a
lagged dose. Crude odds ratios were recalculated ex-
cluding exposures that occurred 5 years or less from
the beginning of the case survival interval. The lag
period was an average of 7 years. Because mortality
was evaluated and not incidence it cannot be specif-
ically determined if exposure occurred prior to de-
velopment of the disease, just that it occurred prior
to death. Nor can it be determined if 7 years is a
sufficient window to be considered a critical window
for the mortality from these cancers.

Domain 3: Outcome Assessment

Continued on next page

9


-------
.. . continued from previous page

Study Citation: Union Carbide (1989). Lymphatic and hematopoietic tissue cancer in a chemical manufacturing environment with attached tables and

cover letter dated 02/21/89
Data Type:	occupational 1,4-D, lymphatic & hematopoietic cancer-Cancer

HERO ID:	597727

Domain

Metric

Rating^

MWF*

Score

Comments^

Metric 7:

Outcome measurement or characterization

Medium

X

0.667

1.33

Cases were identified from a review of both under-
lying and contributory causes of death among male
decedents (1940-1978) from the cohort. Based in in-
formation provided in HERO ID 1010430, this in-
formation was obtained from death certificate di-
agnosis. This misses cases that survived and cases
where there may have been another cause of death.
The study authors acknowledge that there may be
some misclassification of disease status, they also
note that there was agreement between death certifi-
cates and tumor registry diagnoses for these tumors.

Metric 8:

Reporting Bias

Medium

X

0.333

0.67

All outcomes were reported. Confidence intervals for











risk estimates are provided in the text, but not in
tables. Number of cases were reported, but number
of controls was not.

Domain 4: Potential Counfounding/Variable Control











Metric 9:

Covariate Adjustment

Low

X

0.5

1.5

Only age-adjusted stratified analyses were also con-
ducted. No other confounder was considered.

Metric 10:

Covariate Characterization

Medium

X

0.25

0.5

Work records were presumably the source of the
information, but it was not specifically identified.
Age and gender were the only covariates considered
and work records are likely a reliable source. For
cases, this information was also likely available on
the death records.

Metric 11:

Co-exposure Confounding

Low

X

0.25

0.75

Co-exposures were considered when discussing the
cases and their exposures. However, for dioxane this
information was not available nor was indicated if
this exposure occurred in a single work area or over
several areas where co-exposures would have varied.
Controls might have been subject to different co-

exposures than cases.

Domain 5: Analysis

Metric 12: Study Design and Methods	Medium	X 0.4	0.8 The case-control design and calculation of odds ra-

tios were appropriate for determining the associa-
tion between exposure to 1-4D and lymphatic and
hematopoietic tissue cancers.

Continued on next page .. .

10


-------
. continued from previous page

Study Citation: Union Carbide (1989). Lymphatic and hematopoietic tissue cancer in a chemical manufacturing environment with attached tables and

cover letter dated 02/21/89
Data Type:	occupational 1,4-D, lymphatic & hematopoietic cancer-Cancer

HERO ID:	597727

Domain

Metric

Rating^

MWF*

Score

Comments^

Metric 13

Statistical power

Unacceptable

x 0.2

0.04

The number of cases exposed to dioxane was too











low to detect an effect of exposure. The number of











controls was not reported. There were only 4 cases











total exposed to dioxane and the 4 outcomes were











evaluated separately so there was 1 case with non-











Hodgkins lymphoma and 3 cases of non-lymphatic











leukemia. This was likely insufficient to determine











the effects of exposure.

Metric 14

Reproducibility of analyses

Low

x 0.2

0.6

The description of the analysis is insufficient to un-











derstand precisely what has been done and to be











reproducible.

Metric 15

Statistical models

Low

x 0.2

0.6

No description of the model was provided.

Domain 6: Other Considerations for Biomarker Selection and Measurement









Metric 16

Use of Biomarker of Exposure



NA

NA



Metric 17

Effect biomarker



NA

NA



Metric 18

Method Sensitivity



NA

NA



Metric 19

Biomarker stability



NA

NA



Metric 20

Sample contamination



NA

NA



Metric 21

Method requirements



NA

NA



Metric 22

Matrix adjustment



NA

NA



Overall Quality Determination"1'

Unacceptable**



2.4



Extracted

No

* Consistent with our Application of Systematic Review in TSCARisk Evaluations document, if a metric for a data source receives a score of Unacceptable (score = 4), EPA
will determine the study to be unacceptable. In this case, one or more of the metrics were rated as unacceptable. As such, the study is considered unacceptable and the score
is presented solely to increase transparency.

MWF = Metric Weighting Factor

High = 1; Medium = 2; Low = 3; Unacceptable = 4; N/A has no value.

The overall rating is calculated as necessary. EPA may not always provide a comment for a metric that has been categorized as High.

Overall rating =

]T\ (Metric Score; X MWF;) / J] . MWFj

if any metric is Unacceptable
(round to the nearest tenth) otherwise

where High => 1 to < 1.7; Medium => 1.7 to < 2.3; Low => 2.3 to < 3.0. If the reviewer determines that the overall rating needs adjustment, the original rating is crossed
out and an arrow points to the new rating,
ft This metric met the criteria for high confidence as expected for this type of study

11


-------
Table 4: Garcia et al. 2015: Evaluation of Cancer Outcomes

Study Citation: Garcia, E; Hurley, S; Nelson, DO; Hertz, A; Reynolds, P (2015). Hazardous air pollutants and breast cancer risk in California teachers:
A cohort study Environmental Health: A Global Access Science Source, 14(1), 14

Data Type:	Cohort	1,4-D	CTS	BreastCancer	Ql-Cancer

HERO ID:	3014082

Domain

Metric

Rating^ MWF* Score

Comments^

Domain 1: Study Participation

Metric 1: Participant selection

High x 0.4

Metric 2: Attrition

High x 0.4

Metric 3: Comparison Group

High x 0.2

0.4 California Teachers Study including active and re-
tired female teachers and administrators were en-
rolled in the California State Teachers Retirement
System and completed a questionnaire. Study pop-
ulation was comprised on 5676 women. All partic-
ipants were included using the same inclusion and
exclusion criteria.

0.4 Large sample of study population excluded due to
women who were not residing in California at base-
line, had unknown history of prior cancer, had prior
history of invasive or in situ breast cancer, asked to
be removed from study after joining, or had an ad-
dress that couldn't be geocoded. This represents ad-
equate explanation of attrition and is not expected
to bias the results.

0.2 Cases and controls were stated to be similar. Covari-
ates that were different between groups were consid-
ered and included as covariates in the final model.,
including a term for grouped personal risk factors.

Domain 2: Exposure Characterization

Metric 4: Measurement of Exposure

Medium x 0.4

Metric 5: Exposure levels

Medium x 0.2

0.8 NATA identified and prioritized the air toxicants
with respect to their potential population health
risks. The first NATA was conducted based on 1996
emissions. EPA models annual ambient HAP con-
centrations using the Assessment System for Pop-
ulation Exposure Nationwide (ASPEN). This is a
well-established method of determining exposure.,
but may lead to some non-differential exposure mis-
classification.

0.4 By examining each compound individually, they cat-
egorized them into four quantiles of concentration
without including exposure from any other com-
pound in the model. Level of exposure adequate.
Included four quantiles of exposure, Q1 being no ex-
posure.

Continued on next page . ..

12


-------
. continued from previous page

Study Citation: Garcia, E; Hurley, S; Nelson, DO; Hertz, A; Reynolds, P (2015). Hazardous air pollutants and breast cancer risk in California teachers:
A cohort study Environmental Health: A Global Access Science Source, 14(1), 14

Data Type:	Cohort	1,4-D	CTS	BreastCancer	Ql-Cancer

HERO ID:	3014082

Domain

Metric

Rating^ MWF* Score

Comments^

Metric 6: Temporality

Medium x 0.4

0.8 Chose to use the 2002 ambient air concentration es-
timates for this study because that year was approx-
imately the mid-point for the follow up period. De-
cided against combining multiple years of estimate
due to inconsistent methodical approaches and tem-
poral variations in the level of agreement between
years of the assessments which could introduce ex-
posure misclassification.

Domain 3: Outcome Assessment

Metric 7: Outcome measurement or characterization

Metric 8: Reporting Bias

High	X 0.667 0.67 CTS cohort is followed annually for cancer diagno-

sis, death, and change of address. Annual linkage
between CCR and cohort membership was used to
identify incident cancer rates. Defined a case as any
woman diagnosed with invasive breast cancer (ICD-
03 site codes C500-C509, excluding those with his-
tology codes for 9050-9055, 9140, and 9590-9992) af-
ter the date they completed their baseline question-
naire through Dec 31, 2011.

High	X 0.333 0.33 CCR maintains high standards for data quality and

completeness and is estimated to be 99% complete.
Ascertained date and cause of death from mortality
files as well as reports from relatives.

Domain 4: Potential Counfounding/Variable Control
Metric 9: Covariate Adjustment

Metric 10: Covariate Characterization

High

x 0.5

Metric 11: Co-exposure Confounding

Medium x 0.25

Medium x 0.25

0.5 All models were stratified by age and adjusted either
for race alone or for race and personal risk factors of
interest. For each compound, p-values no each non-
degenerative quantile HR were adjusted for multiple
testing across the ten subsets using False Discovery
Rates.

0.5 Covariates were obtained from the CTS baseline
questionnaire. This was self-reported information,
but there is no evidence to suggest that it is not a
valid method of obtaining covariate information.

0.5 No indication of unbalanced co exposures.

Domain 5: Analysis

Metric 12: Study Design and Methods

Metric 13: Statistical power

Medium x 0.4 0.8

Medium x 0.2 0.4

Cohort was appropriate study design. Examined the
relationship between risk of breast cancer and nu-
merous compounds of interest. Used two different
methods of parameterizing exposure in the models.
Number of subjects for estimated exposure was 5676
women. There were enough subjects to detect effects
for some chemicals and for some trends.

Continued on next page . ..

13


-------
. continued from previous page

Study Citation: Garcia, E; Hurley, S; Nelson, DO; Hertz, A; Reynolds, P (2015). Hazardous air pollutants and breast cancer risk in California teachers:

A cohort study Environmental Health: A Global Access Science Source, 14(1), 14
Data Type:	Cohort_l,4-D_CTS_BreastCancer_Ql-Cancer

HERO ID:	3014082

Domain



Metric

Rating^

MWF*

Score

Comments^



Metric 14

Reproducibility of analyses

Medium

x 0.2

0.4

Study design and methods can be reproducible with
information provided. Provided reasoning on how
categories were created for exposure quantiles, why
covariates were used. Covariates included in the
models are reported explicitly.



Metric 15

Statistical models

Medium

x 0.2

0.4

Used COX proportional hazard models to estimate
hazard rate ratios. Parameterized exposures into
quantiles, modeled exposure as a continuous vari-
able, and tested for non-zero slope using a likelihood
ratio test.

Domain 6:

Other Considerations for Biomarker Selection and Measurement











Metric 16

Use of Biomarker of Exposure



NA

NA





Metric 17

Effect biomarker



NA

NA





Metric 18

Method Sensitivity



NA

NA





Metric 19

Biomarker stability



NA

NA





Metric 20

Sample contamination



NA

NA





Metric 21

Method requirements



NA

NA





Metric 22

Matrix adjustment



NA

NA



Overall Quality Determination1"

High



1.5



Extracted





Yes







MWF = Metric Weighting Factor

High = 1; Medium = 2; Low = 3; Unacceptable = 4; N/A has no value.

The overall rating is calculated as necessary. EPA may not always provide a comment for a metric that has been categorized as High.

Overall rating =

]T\ (Metric Score; x MWF;) / J] . MWFj

if any metric is Unacceptable
(round to the nearest tenth) otherwise

where High => 1 to < 1.7; Medium => 1.7 to < 2.3; Low => 2.3 to < 3.0. If the reviewer determines that the overall rating needs adjustment, the original rating is
crossed out and an arrow points to the new rating.

^ This metric met the criteria for high confidence as expected for this type of study

14


-------