Mesothelioma among Motor Vehicle Mechanics: An Updated Review and Meta-analysis

Abstract

Background:
We published a meta-analysis of the association between work as a motor vehicle mechanic and mesothelioma in 2004. Since then, several relevant studies on this topic have been published. Thus, to update the state-of-the-science on this issue, we conducted a new systematic review and meta-analysis.

Methods:
A comprehensive PubMed literature search through May 2014 was conducted to identify studies that reported relative risk estimates for mesothelioma among motor vehicle mechanics (in general), and those who were engaged in brake repair (specifically). Studies were scored and classified based on study characteristics. Random-effects meta-analyses generated summary relative risk estimates (SRREs) and corresponding 95% confidence intervals (CI). Heterogeneity of results was examined by calculating Q-test P-values (P-H) and I 2 estimates. Sub-group and sensitivity analyses were conducted for relevant study characteristics and quality measures.

Results:
Ten case-control studies, one cohort study, and five proportionate mortality ratio (PMR)/standardized mortality odds ratio (SMOR) studies were identified and included in the quantitative assessment. Most meta-analysis models produced SRREs below 1.0, and no statistically significant increases in mesothelioma were observed. The SRRE for all studies was 0.80 (95% CI: 0.61–1.05) with significant heterogeneity (P-H

Full text

© e Author 2015. Published by Oxford University Press on behalf of the British Occupational Hygiene Society.
• 1
RE VIEW
Mesothelioma among Motor Vehicle
Mechanics: An Updated Review and
Meta-analysis
David H.Garabrant
1,2
*, Dominik D.Alexander
1
, Paula E.Miller
1
,
Jon P.Fryzek
1
, PaoloBoea
3
, M. J.Teta
4
, Patrick A.Hessel
5
, Valerie
A.Craven
6
, Michael A.Kelsh
7
and MichaelGoodman
8
1.EpidStat Institute, 2100 Commonwealth Blvd, Suite 203, Ann Arbor, MI 48105, USA;
2.University of Michigan School of Public Health, 1415 Washington Heights, Ann Arbor, MI 48109, USA;
3.e Tisch Cancer Institute and Institute for Translational Epidemiology, Icahn School of Medicine at Mount Sinai,
One Gustave L. Levy Place, Box 1133, New York, NY 10029, USA;
4.Department of Health Sciences, Exponent, Inc., 420 Lexington Avenue, New York, NY 10170, USA;
5.EpiLung Consulting, #25, 26204 Township Road 512, Spruce Grove, Alberta, Canada T7Y 1C5;
6.Exponent, Inc., 500 12th Street, Suite 220, Oakland, CA 94607, USA;
7.Amgen, Inc One Amgen Center Drive, ousand Oaks, CA 91320, USA;
8.Emory University School of Public Health, 1518 Clion Road, Atlanta, GA 30322, USA
*Author to whom correspondence should be addressed. Tel:734-929-9150; e-mail: david@epidstat.com
Submied 10 February 2015; revised 8 May 2015; revised version accepted 21 July 2015.
ABSTRACT
Background: We published a meta-analysis of the association between work as a motor vehicle mechanic
and mesothelioma in 2004. Since then, several relevant studies on this topic have been published. us,
to update the state-of-the-science on this issue, we conducted a new systematic review and meta-analysis.
Methods: A comprehensive PubMed literature search through May 2014 was conducted to identify
studies that reported relative risk estimates for mesothelioma among motor vehicle mechanics (in gen-
eral), and those who were engaged in brake repair (specically). Studies were scored and classied
based on study characteristics. Random-eects meta-analyses generated summary relative risk esti-
mates (SRREs) and corresponding 95% condence intervals (CI). Heterogeneity of results was exam-
ined by calculating Q-test P-values (P-H) and I
2
estimates. Sub-group and sensitivity analyses were
conducted for relevant study characteristics and quality measures.
Results: Ten case-control studies, one cohort study, and ve proportionate mortality ratio (PMR)/
standardized mortality odds ratio (SMOR) studies were identied and included in the quantitative
assessment. Most meta-analysis models produced SRREs below 1.0, and no statistically signicant
increases in mesothelioma were observed. e SRRE for all studies was 0.80 (95% CI: 0.61–1.05) with
signicant heterogeneity (P-H <0.001, I
2
=62.90). Asimilar SRRE was observed among the ve Tier
1 studies with the highest quality ratings (SRRE=0.76, 95% CI: 0.46–1.25), with no heterogeneity
Ann. Occup. Hyg., 2015, 1–19
doi:10.1093/annhyg/mev060
by guest on November 5, 2015http://annhyg.oxfordjournals.org/Downloaded from

among studies (P-H=0.912, I
2
=0.00). Meta-analysis of the Tier 2 (n=5) and Tier 3 (n=6) studies
resulted in SRREs of 1.09 (95% CI: 0.76–1.58) and 0.73 (95% CI: 0.49–1.08), respectively. Restricting
the analysis to Tiers 1 and 2 combined resulted in an SRRE of 0.92 (95% CI: 0.72–1.29). e SRRE
specic to brake work (n=4) was 0.64 (95% CI: 0.38–1.09).
Conclusions: is meta-analysis of the epidemiologic studies provides evidence that motor vehicle
mechanics, including workers who were engaged in brake repair, are not at an increased risk of mesothelioma.
KEYWORDS: asbestos epidemiology; auto repair; brake dust; cancer epidemiology; chrysotile;
mesothelioma
INTRODUCTION
Studies of motor vehicle repair activities and brake
repair activities provide an opportunity to examine
the risks of mesothelioma in seings in which chry-
sotile exposure is not accompanied by exposure to
commercial amphibole products. Exposure to short
chrysotile bers can occur during installation and
repair of asbestos-containing brakes, clutches, and
gaskets. In the past, automobile brakes typically con-
tained chrysotile asbestos embedded in a solid binder
(Jacko et al., 1975). In the USA, EPA banned the
manufacture, importation, and processing of brake
linings and pads containing asbestos in 1993 (EPA,
1989). Although the phase-out of asbestos-containing
brakes likely began in the 1980s, brakes that were in
place in 1993 continued to be replaced aer the ban
took eect. e process of brake replacement involves
three potential opportunities for release of asbestos
bers: (i) small amounts of chrysotile asbestos (usu-
ally <1%) may be present in the brake wear debris
(Lynch, 1968; Hickish and Knight, 1970; Anderson
et al., 1973; Jacko and DuCharme, 1973; Rowson,
1978; Williams and Muhlbaier, 1982; Cha etal., 1983;
NIOSH etal., 1989), (ii) asbestos can be released dur-
ing sanding, grinding, and beveling of new asbestos
brake linings or pads, and (iii) asbestos can be released
during handling of new brakes and packaging materi-
als (Madl etal., 2008).
Some authors (Freeman and Kohles, 2012; Lorimer
etal., 1976), regulatory agencies (EPA, 1986b), and
trade organizations (World Trade Organization,
2000) have opined in the past that motor vehicle
mechanics are likely to be at increased risk of asbes-
tos-related disease, most notably mesothelioma.
ese opinions were based primarily on the assumed
opportunity for asbestos exposure during brake work
(Paustenbach et al., 2004; Madl et al., 2008; IARC,
2012) and on case reports of mesothelioma among
brake repair workers (EPA, 1986a,b, 1989; Lemen,
2004). However, case reports and case series cannot
quantify associations and are not adequate for assess-
ing causality (Hennekens and Buring, 1987). Case
reports and case series become even less informative
in the presence of analytic studies.
When the EPA conducted its evaluation ~30years
ago (EPA, 1986a,b), the epidemiologic information
on mesothelioma among vehicle mechanics was lim-
ited to only three studies (McDonald and McDonald,
1980; Teta etal., 1983; Spirtas etal., 1985). Our pre-
vious meta-analysis (Goodman etal., 2004) included
11 studies. In the past 10years, ve additional epide-
miologic studies (Rake et al., 2009; Aguilar-Madrid
etal., 2010; Rolland etal., 2010; Merlo etal., 2010;
Roelofs etal., 2013) and four updates of previous stud-
ies (McElvenny etal., 2005; Milham, 2011; NIOSH,
2011; Health and Safety Executive, 2013) have exam-
ined the risk of mesothelioma among motor vehicle
mechanics or brake workers. erefore, an updated
systematic review is warranted.
Our objective was to conduct an updated system-
atic review and quantitative meta-analysis of the epi-
demiologic literature examining the relative risk (RR)
of mesothelioma among workers engaged in motor
vehicle repair and, when possible, among workers
occupationally exposed to brake dust. e specic
aims were to: (i) estimate summary relative risk esti-
mates (SRREs) between motor vehicle mechanics and
mesothelioma; (ii) conduct sub-group and sensitivity
analyses by relevant study characteristics (e.g. study
design and type of population) to identify potential
sources of heterogeneity; (iii) estimate the inuence
of each study on the overall eect size; (iv) examine
potential trends of associations for certain study fac-
tors (study quality, study period) using meta-regres-
sion; and (v) evaluate the potential for publication
bias.
Page 2 of 19 • Mesothelioma among motor vehicle mechanics
by guest on November 5, 2015http://annhyg.oxfordjournals.org/Downloaded from

METHODS
Literature search and study selection
A comprehensive literature search was conducted
using PubMed and Embase to identify studies evaluat-
ing mesothelioma risk among motor vehicle mechan-
ics. e literature search was conducted through May
2014, with no lower date truncation. Keyword searches
were conducted in PubMed and Embase for terms
including ‘mesothelioma’, ‘pleural neoplasms’, ‘pleura
tumor’, ‘occupational exposure’, ‘occupational dis-
eases’, ‘asbestos’, ‘motor vehicles’, ‘automobiles’, ‘auto-
mobile industry’, ‘mechanics’, ‘case control studies’,
‘odds ratio’, ‘risk factors’, ‘mortality/trends’, ‘registries’,
‘incidence’, ‘epidemiology’, and ‘population surveil-
lance’. References from relevant studies, review arti-
cles, and previous meta-analyses were also screened.
Additional electronic searches were conducted to
identify relevant studies that were not published in
the peer-reviewed literature, including government
documents. When information was missing from
published reports, aempts were made to contact the
authors to obtain it. Search results were screened by
two individuals to determine relevancy, with no dis-
crepancies between reviewers (Fig.1). To be included
in the meta-analysis, studies were required to meet the
following criteria: (i) cohort or case-control design
reporting incidence or mortality, or studies based on
analyses of proportions [proportionate mortality ratio
(PMR)/proportional incidence ratio (PIR)/stand-
ardized mortality odds ratio (SMOR) analyses]; (ii)
analysis of an adult human population; (iii) meso-
thelioma risk analyzed among subjects involved in
motor vehicle repair (if a study reported results only
on the broad category of general mechanics, it was not
included in the meta-analysis); and (iv) mesothelioma
RR estimates and variance measures either reported in
the original publication or calculated based on the data
obtained from the authors or reported in the papers.
Study quality scoring methodology
All studies retained for quantitative assessment under-
went formal evaluation and were assigned a quality
score according to their methodological strengths and
limitations, based on the methods used in our previous
Figure1 Flow diagram of study selection.
Mesothelioma among motor vehicle mechanics • Page 3 of 19
by guest on November 5, 2015http://annhyg.oxfordjournals.org/Downloaded from

paper (Goodman etal., 2004) and those recommended
in the Meta-analysis Of Observational Studies in
Epidemiology (MOOSE) (Stroup et al., 2000) and
Preferred Reporting Items for Systematic Reviews and
Meta-Analyses (PRISMA) (Moher etal., 2009) guide-
lines. Each study was awarded a point for fullling each
methodological criterion and a score of zero for failing
to do so. e scoring used the following 12 criteria:
• Overall study design: PMR/PIR studies or
death certicate-based SMOR studies=0;
else (cohort or case–control studies)=1.
• Asbestos exposure: Job title based=0 (e.g.
‘car mechanic’); task specic=1 [e.g. ‘brake
repairmen’ or industrial hygiene (IH) based].
• Was age taken into account? no=0; yes=1.
• Lifetime asbestos exposure history obtained:
no=0; yes=1.
• Did study methods or analysis address
confounding in any way? no=0; yes=1.
• Exposure-response analysis performed?
no=0; yes=1.
• Was latency taken into account? no=0; yes=1.
• For cohort studies, duration of follow-up:
<30years=0; ≥30years=1.
• For case-control studies, response rate <80%
or not reported:=0; ≥80%=1.
• Information bias: possible=0; unlikely/
addressed=1 (e.g. in case–control studies
using recorded occupational histories).
• Selection bias: possible=0 (e.g. in hospital-
based case-control studies); unlikely/
addressed=1 (e.g. in cohort studies or
population-based case-control studies).
• Cases conrmed by pathologic review:
no=0; yes=1.
Scoring was used as a formal approach to classify
studies into three equal sized tiers. Studies with scores
of seven or above were included in Tier 1 and consid-
ered to have the strongest study design and analysis
and less opportunity for bias. Tier 2 included studies
with a score of ve or six, which were considered less
informative due to methodological limitations. Tier 3
included studies that scored four or less. It should be
noted that many of the scoring criteria related to inher-
ent strengths or limitations of the study design and the
underlying data, and not necessarily to the methodo-
logical decisions of the original investigators.
Data extraction and statistical analysis
Information extracted from each study using a stand-
ard form included rst author, publication year, study
design, nature of cohort, geographic location of study,
exposure denition, source of cases, comparison
group, sample size, number of cases, years of follow-
up, population demographic characteristics, outcome
classication, measures RR and 95% condence inter-
vals (CIs) for each relevant exposure group and worker
classication, statistical adjustments, and the method-
ological factors on which scoring was based. If more
than one article from the same study population was
published, data from the publication with the longest
follow-up and/or the most completely adjusted risk
estimate wasused.
Random-eects models were used to calculate
SRREs and 95% CIs. e study weights were equal to
the inverse of the variance of each study’s eect estimate
(DerSimonian and Laird, 1986). Measures of RR [e.g.
odds ratios (ORs), rate ratios, standardized mortality
ratios(SMRs)] and associated measures of variance (e.g.
95% CIs) were used as input parameters in the meta-
analysis models. Four case-control studies (McDonald
and McDonald, 1980; Woitowitz and Rodelsperger,
1994; Agudo etal., 2000; Aguilar-Madrid etal., 2010)
did not report the results in terms of ORs but provided
information from which ORs and 95% CIs could be cal-
culated using SAS 9.3 (Cary, NC, USA). Two propor-
tional mortality studies (Milham and Ossiander, 2001;
NIOSH, 2011) did not report 95% CIs, but provided
numbers of observed and expected cases, from which
95% CIs were calculated based on the Poisson distribu-
tion (Breslow and Day, 1987).
Primary meta-analysis models were used to gen-
erate SRREs for case-control and cohort studies of
motor vehicle mechanics and mesothelioma overall
and by study tiers. e base model included all eligi-
ble studies, and separate models were created for each
study quality tier. Sub-group and sensitivity analyses
were conducted to examine paerns of associations
and to identify potential sources of heterogeneity
overall and for groups of studies. ese included analy-
ses by study design, geographic location, and asbestos
exposure history ascertainment. One-study-removed
Page 4 of 19 • Mesothelioma among motor vehicle mechanics
by guest on November 5, 2015http://annhyg.oxfordjournals.org/Downloaded from

sensitivity analyses were conducted to determine the
relative inuence of each study on the overallmodel.
Heterogeneity was assessed quantitatively using the
Cochran’s Q-test with the corresponding P-value (P-
H) and by calculating the I
2
statistic, which indicates
the percentage of variation aributable to between-
study heterogeneity (Higgins and ompson, 2002).
e presence of publication bias was assessed visually
by examining a funnel plot measuring the standard
error as a function of eect size, as well as perform-
ing Egger’s regression method (Rothstein etal., 2005).
We generated forest plots for models of motor vehicle
mechanics and mesothelioma by study quality tiers.
Meta-regression analyses were performed to evaluate
the paerns in summary eect sizes based on relevant
factors (modeled as continuous variables), such as
increasing study quality and study period. Statistical
analyses were conducted using Comprehensive
Meta-Analysis Soware (version 2.2.046; Biostat,
Englewood, NJ, USA), SAS 9.3 (SAS Institute, Cary,
NC, USA), and STATA statistical soware (Stata
Corporation, College Station, TX, USA).
RESULTS
Ten case-control studies, one cohort study, and ve
PMR/SMOR studies published between 1980 and
2013 were included in the meta-analysis (Table1). Since
RRs could not be calculated for three cohort studies of
motor vehicle mechanics (two from Sweden and one
from Denmark) that provided information on meso-
thelioma (Jarvholm and Brisman, 1988; Hansen, 1989;
Gustavsson etal., 1990), these studies were not included
in the meta-analysis. Five studies were of higher method-
ological quality (Tier 1), ve studies were of moderate
methodological quality (Tier 2), while six studies were
of lower methodological quality (Tier 3)(Table2).
Overview of studies included in meta-analysis
Tier1
McDonald (McDonald and McDonald, 1980) com-
pared histologically conrmed mesothelioma cases in
the USA and Canada to matched controls that had pul-
monary metastases from non-pulmonary malignancies.
Lifetime occupational histories were obtained through
interviews with relatives. Of the 156 cases and 156
controls who worked in occupations thought to entail
asbestos exposure but not previously recognized as
associated with mesothelioma risk, 11 cases and 12 con-
trols were ‘garage’ workers, from which we calculated an
OR of 0.91 (95% CI: 0.35–2.34). is study had a large
sample size and a high next-of-kin response rate for
both cases (95%) and controls (91%). Hospital-based
cancer cases may not have been representative of the
source population of cases (i.e. possible selection bias).
However, the choice of controls with other cancers may
have reduced recall bias. e use of the highest asbestos
exposure to characterize each participant’s occupational
history likely decreased potential confounding by other
asbestos exposures.
Teta (Teta et al., 1983) compared pathologi-
cally conrmed cases from the Connecticut Tumor
Registry and from a Veterans Administration hospi-
tal with controls from Connecticut death certicate
les. Occupational histories were obtained from death
certicates and city directories and did not include
information specic to brake work. ere were 147
total cases and 464 controls with 1 exposed case and 5
exposed controls included in the analysis. e OR for
subjects employed in ‘automobile repair and related
service’ was 0.65 (95% CI: 0.08–5.53). e choice of
population controls, the completeness of occupational
records (job titles were obtained for 98% of cases and
99% of controls) and the histological conrmation of
cases were methodological strengths. Shortcomings
were a lack of information on brake repair work and
lack of control for confounding by other potential
sources of asbestos exposure. e OR estimate was
based on relatively few observations and did not take
into account exposure-response or latency.
Hessel (Hessel et al., 2004) updated a previous
National Cancer Institute (NCI) case-control study
(Spirtas etal., 1985, 1994). Pathologically conrmed
denite or probable mesothelioma cases in the Los
Angeles County Cancer Surveillance Program, the
New York State Cancer Registry (excluding New York
City) and 39 Veterans Administration hospitals were
included (n=147). Controls included patients who
died of causes other than cancer, respiratory disease,
suicide, or violence (n= 358). Interviews requested
lifetime occupational history and information spe-
cically on brake lining installation or repair. Hessel
calculated an OR of 1.04 (95% CI: 0.46–2.22) for the
occupational category ‘brake installation or repair’.
However, aer adjusting for any of eight occupations
with potential asbestos exposure, the OR became 0.82
Mesothelioma among motor vehicle mechanics • Page 5 of 19
by guest on November 5, 2015http://annhyg.oxfordjournals.org/Downloaded from

Table1. Summary of mesothelioma studies and corresponding RR estimates included in the meta-analysis.
First author Year Design Exposure denition Source of cases Comparison group RR estimate (95% CI)
a
Tier 1 studies
b
McDonald 1980 Case-control Garage workers Hospital records Non-pulmonary cancers 0.91 (0.35–2.34)
Teta 1983 Case-control Automobile repair and related
services
CT Tumor Registry Connecticut decedents 0.65 (0.08–5.53)
Teschke 1997 Case-control Vehicle mechanics British Columbia’s
Cancer Registry
Randomly selected from
voters lists
0.8 (0.2–2.3)
Brake lining installation or
repair
0.3 (0.0–1.4)
Vehicle mechanics (ever
employed in at-risk
occupations excluded)
0.4 (0.0–3.2)
Hessel 2004 Case-control Brake lining installation or
repair, all other asbestos
exposures controlled
NY Cancer Registry,
LA County Cancer
Surveillance Program,
VA Hospitals
Deaths from causes other
than cancer, respiratory
disease, suicide or violence
0.82 (0.36–1.80)
Rake 2009 Case-control Vehicle maintenance involving
work with brakes or gaskets
Hospital records, National
Cancer Research Network,
healthcare providers
Randomly selected from
Health Authority registers
0.4 (0.1–1.7)
Tier 2 studies
c
Agudo 2000 Case-control Mechanics, motor vehicles Hospital records Patients with non-asbestos-
related conditions
0.62 (0.11–2.36)
Hansen 2003 Case-control Repair of motor vehicles and
motorcycles
Danish Cancer Registry All other occupations
combined
0.8 (0.4–1.5)
Aguilar-Madrid 2010 Case-control Automobile mechanics Two hospitals Randomly selected active
and retired workers in the
Mexican Institute of Social
Security database
0.74 (0.08–6.68)
Page 6 of 19 • Mesothelioma among motor vehicle mechanics
by guest on November 5, 2015http://annhyg.oxfordjournals.org/Downloaded from

First author Year Design Exposure denition Source of cases Comparison group RR estimate (95% CI)
a
Merlo 2010 Cohort Bus maintenance workers Employees of Azienda
Municipalizzata Trasporti
Ligurian male population 1.27 (0.66–2.43)
Rolland 2010 Case-control Motor vehicle mechanics National Mesothelioma
Surveillance Program
Randomly selected from
the 1999 census
1.50 (0.76–2.95)
Tier 3 studies
d
Woitowitz 1994 Case-control Motor vehicle repair workers Not specied Lung resection patients
and population controls
0.87 (0.43–1.70)
Denitely engaged in brake
service
0.89 (0.31–2.47)
McElvenny 2005 Proportionate
mortality
Motor mechanics British mesothelioma
register
General population of Great
Britain
0.48 (0.37–0.62)
Milham 2011 Proportionate
mortality
Automobile mechanics and
repair workers
Washington State
residents
All other occupations
combined
0.73 (0.50–1.11)
NIOSH 2011 Proportionate
mortality
Auto mechanics National Occupational
Mortality Surveillance
database
All other occupations
combined
0.82 (0.50–1.27)
Health and
Safety Executive
2013 Proportionate
mortality
Motor mechanics; auto
engineers
British mesothelioma
register
General population of
Great Britain
0.41 (0.30–0.55)
Roelofs 2013 Standardized
morbidity
Automobile mechanics Massachuses Cancer
Registry
Patients with cancers not
associated with asbestos
2.1 (1.1–4.0)
a
RR estimates are shown as reported by each study or are rounded to two decimal places if they were calculated based on data obtained from the authors or reported in the papers.
b
Studies of higher methodological quality.
c
Studies of moderate methodological quality.
d
Studies of lower methodological quality.
Table1. Continued
Mesothelioma among motor vehicle mechanics • Page 7 of 19
by guest on November 5, 2015http://annhyg.oxfordjournals.org/Downloaded from

(95% CI: 0.36–1.80). In addition, when cases and
controls with a history of employment in any of these
occupations were removed from the analysis, the OR
for occupational brake installation and repair became
0.62 (95% CI: 0.01–4.71). An exposure-response
analysis showed no evidence of increasing risk of
mesothelioma with increasing duration of brake work.
Subjects whose only brake work was non-occupa-
tional (i.e. do-it-yourself) were not at increased risk
of mesothelioma. ese NCI data have several impor-
tant features: (i) exposure was dened specically
as brake installation and repair, (ii) confounding by
other occupational exposures was addressed, and (iii)
information on duration of employment allowed an
exposure-response analysis. Alimitation of this data
set was the relatively low response rate (74% among
cases’ next-of-kin, 79% among controls’ next-of-kin).
Teschke etal. (1997) compared pathologically con-
rmed mesothelioma cases from the British Columbia
Cancer Agency to matched controls selected from
voter registration lists. Lifetime occupational and
exposure histories were obtained, whenever possible,
directly from cases and controls. e OR for ‘vehicle
mechanics’ was 0.8 (95% CI: 0.2–2.3). e OR for the
Table2. Quality scores of studies evaluating the association between mesothelioma risk and
employment as a motor vehicle mechanic.
Quality criteria Study
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
Overall study design: PMR/PIR/SMOR=0; SMR/
OR=1
1 1 1 1 1 1 1 1 1 1 1 0 0 0 0 0
Asbestos exposure: Job title (e.g. ‘car
mechanic’)=0; Task specic (e.g. brake repairman
or IH-based)=1
0 0 1 1 1 0 1 0 0 0 1 0 0 0 0 0
Was age taken into account adequately?: no=0;
yes=1
1 1 1 1 1 1 1 1 1 1 0 1 1 1 1 1
Lifetime asbestos exposure history: no=0; yes=1 1 1 1 1 1 1 0 1 0 1 1 0 0 0 0 0
Was any aempt made to control for confounding?:
no=0; unlikely/addressed=1
1 0 1 1 1 0 0 0 0 0 0 0 0 0 0 0
Exposure-response analysis performed? no=0;
yes=1
0 0 0 1 1 0 0 0 0 0 0 0 0 0 0 0
Was latency taken into account? no=0; yes=1 0 0 1 0 1 0 1 0 1 0 0 0 0 0 0 0
For cohort studies: follow-up <30 yr=0;
≥30years=1
1 0 0 0 0 0
For case–control studies: response rate:
<80% or not reported=0; ≥80%=1
1 1 1 0 0 1 1 0 0 0
Information bias: possible=0; unlikely/
addressed=1
1 1 0 0 1 0 1 1 1 1 0 1 1 1 1 0
Selection bias: possible=0; unlikely/addressed=1 0 1 0 1 1 1 0 0 1 0 0 1 1 1 1 0
Cases based on pathology review: no=0; yes=1 1 1 1 1 1 1 0 1 0 1 1 0 0 0 0 1
Total score 7 7 8 8 10 6 6 5 6 5 4 3 3 3 3 2
1, (McDonald and McDonald, 1980); 2, (Teta etal., 1983); 3, (Teschke etal., 1997); 4, (Hessel etal., 2004); 5, (Rake etal., 2009); 6, (Agudo etal., 2000);
7, (Hansen and Meersohn, 2003); 8, (Aguilar-Madrid etal., 2010); 9, (Merlo etal., 2010); 10, (Rolland etal., 2010); 11, (Woitowitz and Rodelsperger,
1994b); 12, (McElvenny etal., 2005); 13, (Milham, 2011); 14, (NIOSH, 2011); 15, (Health and Safety Executive, 2013); 16, (Roelofs etal., 2013).
Page 8 of 19 • Mesothelioma among motor vehicle mechanics
by guest on November 5, 2015http://annhyg.oxfordjournals.org/Downloaded from

category ‘brake lining installation or repair’ was 0.3
(95% CI: 0.0–1.4). Aer removing cases and controls
with at-risk occupational asbestos exposures, the OR
for ‘vehicle mechanics’ was 0.4 (95% CI: 0.0–3.2).
e results did not change aer taking into account
20+ years of latency and, in fact, the rst exposure for
all of the vehicle mechanics and brake workers was
>20years prior to diagnosis. Information bias is a con-
cern in this study due to the dierence in proportion
of next-of-kin interviews among cases (33.3%) com-
pared with controls (13.6%). Although fairly small in
size, this study was methodologically strong because:
(i) it considered other asbestos exposures, (ii) speci-
ed exposure as ‘brake lining installation or repair’,
(iii) considered latency, and (iv) had a high response
rate (88% among cases, 81% among controls).
Rake etal. (2009) conducted a case-control study
of 622 mesothelioma patients (92% histologically con-
rmed) and 1420 age-matched population controls in
England, Wales, and Scotland. Cases were identied
through chest physicians, surgeons and nurses, the
National Cancer Research Network, and English and
Scoish hospital records. Apostal questionnaire and
telephone interview assessed lifetime occupational
and residential history, do-it-yourself activities, and
other possible environmental exposures for living
cases and controls. is included work with gaskets
and brake linings. e OR for vehicle maintenance
involving work with brakes or gaskets was 0.4 (95% CI:
0.1–1.7). ough this study used only living cases and
controls, the response rates diered between the cases
(73%) and controls (60%). is study was methodo-
logically strong because the authors accounted for the
impact of other high-risk jobs and examined latency
and duration of exposure.
Tier2
In 1987, Olsen and Jensen published a proportionate
incidence ratio (PIR) analysis that linked cases from
the Danish Cancer Registry with occupational histo-
ries (Olsen and Jensen, 1987). ere were no cases of
mesothelioma for the occupational category ‘repair
of motor vehicles and motorcycles’ or for the indus-
try category ‘garage’. e data were updated in a 2003
case-control study (Hansen and Meersohn, 2003).
For the category ‘repair of motor vehicles and motor-
cycles’ the OR was 0.8 (95% CI: 0.4–1.5), based on
10 mesothelioma cases. Limitations of this study were
lack of requirement of histological conrmation of
mesothelioma diagnosis for study inclusion, inability
to obtain a complete work history or detailed expo-
sure history regarding work with brakes specically,
and lack of latency analysis for repair of motor vehicles
specically. Its strengths included large sample size
and unlikeliness of reporting bias due to the coverage
of all cancer cases in Denmark by the registry.
Agudo etal. (2000), in a hospital-based case-con-
trol study in Spain, compared pathology-conrmed
cases of mesothelioma to population/hospital con-
trols. Acomplete occupational history was collected,
though information was not available on exposure to
brake repair specically. ere were three cases and
14 controls in the category ‘mechanics, motor vehi-
cle’. e non-exposed category included 51 cases and
148 controls that had never worked in any of the at-
risk occupations. e crude OR for ‘mechanics, motor
vehicle’ was 0.62 (95% CI: 0.11–2.36). e compari-
son of motor vehicle mechanics that may have had
other potential asbestos exposure to persons without
exposure may have led to unmeasured confounding.
Because 44% of cases and <1% of controls had next-
of-kin interviews, information bias is possible. e
study’s high response rate, methods of control selec-
tion, and complete occupational histories were among
its strengths.
Aguilar-Madrid (Aguilar-Madrid etal., 2010) con-
ducted a case-control study of 472 workers insured by
the Mexican Institute of Social Security including 119
histologically conrmed incident mesotheliomas and
353 matched hospital controls. In-person interviews
for both cases and controls assessed the complete
employment history and para-occupational exposures
including family members’ occupations and likely
environmental exposures though information was
not available on brake work specically. No next-of-
kin interviews were conducted. ere were one case
and four controls whose most frequent occupation
was ‘mechanic, automobile’, resulting in an OR of 0.74
(95% CI: 0.08–6.68). is OR estimate was based
on relatively few observations, was not adjusted for
potential confounding by other occupational or non-
occupational exposures, and did not take into account
exposure-response or latency.
In Genoa, Italy, Merlo et al. (2010) conducted a
historical cohort mortality study among 9267 male
public transport workers of which 2073 were bus
Mesothelioma among motor vehicle mechanics • Page 9 of 19
by guest on November 5, 2015http://annhyg.oxfordjournals.org/Downloaded from

maintenance workers. Nine deaths from pleural
mesothelioma were observed among bus mainte-
nance workers, resulting in an SMR of 1.27 (95% CI:
0.66–2.43) when compared to the regional male pop-
ulation. is study did not obtain a lifetime asbestos
exposure history, control for potential confounding
by other exposures, assess exposure specic to brake
work, require pathologic review of cases, or perform
exposure-response analyses. is study did, however,
have a long duration of follow-up (1 January 1970 to
31 December 2005) and took latency into account
in the overall cohort but not among the bus mainte-
nance workers specically. Reporting or selection bias
was unlikely as the authors identied all subjects ever
employed at the public transport company inGenoa.
Rolland et al. (2005, 2010) conducted a popu-
lation-based case-control study of primary malig-
nant pleural tumors, nested within France’s National
Mesothelioma Surveillance Program (PNSM); 86.8%
of the cases were pathologically conrmed. A self-
administered survey and interviewer-administered
questionnaire assessed lifetime residential, educa-
tional, and occupational history, including specic job
tasks and do-it-yourself repair. No next-of-kin inter-
views were conducted. Of the 371 and 732 male cases
and controls, 17 cases and 22 controls were motor
vehicle mechanics. e OR for motor vehicle mechan-
ics was 1.50 (95% CI: 0.76–2.95). e authors did not
control for multiple occupational exposures despite
the fact that this information was included in the ques-
tionnaire. e response rate was low, with overall par-
ticipation under 50%.
Tier3
Woitowitz and Rodelsperger (1994) compared life-
time occupational histories of 324 pathology-con-
rmed mesothelioma cases in Germany with two
groups of controls: 315 hospital control patients who
underwent lung resection and 182 population controls.
Lung resection controls were used in order to obtain
lung tissue for ber analysis. Sixteen cases, 16 hospi-
tal controls and 12 population controls were listed as
‘motor vehicle repair workers’. ese data yielded an
OR of 0.97 (95% CI: 0.45– 2.12) using hospital con-
trols and an OR of 0.74 (95% CI: 0.32–1.75) using
population controls. For people denitely engaged in
brake service, the OR was 0.75 (95% CI: 0.25–2.23)
using hospital controls and 1.32 (95% CI: 0.30–6.51)
using population controls. e use of hospital con-
trols that were largely lung cancer patients raises the
possibility of confounding, whereas the population
controls were not subject to this issue. When the two
types of controls were combined for a larger sample
size, the OR was 0.87 (95% CI: 0.43–1.70) for motor
vehicle mechanics and 0.89 (95% CI: 0.31–2.47) for
persons denitely engaged in brake servicing. (ese
calculations were based on updated numbers pro-
vided by the authors, Prof. H.-J. Woitowitz and Dr
K. Rödelsperger, Justus-Liebig University Giessen,
Germany.) e strengths of this study were its ability
to examine the association with brake repair and its
high response rate. e most important shortcomings
included the lack of adjustment for age and an inad-
equate description of subject selection.
Roelofs et al. (2013) conducted a standardized
morbidity odds ratio (SMOR) analysis using inci-
dent mesothelioma cases in the Massachuses Cancer
Registry (MCR) from 1988 to 2003. Atotal of 80 184
cancer controls were used to compute SMORs for
each occupation and industry, controlling for gender
and age. Of the 564 mesothelioma cases coded for
occupation, 10 were identied as automobile mechan-
ics, resulting in an SMOR of 2.1 (95% CI: 1.1–4.0).
Although there were 1424 total cases of mesothelioma
in the MCR during the time period of interest, only
564 (40%) were able to be coded for ‘usual’ occupa-
tion. e majority of cases were identied as retired, of
unknown occupation, or disabled, and were excluded.
e authors did not obtain a lifetime asbestos expo-
sure history or brake work-specic exposure informa-
tion, did not perform exposure-response analysis, and
did not take latency into account. When calculating
the risk estimate for each occupation, they omied
subjects with occupations known to carry an increased
risk from the non-exposed group; however, similar
at-risk individuals could not be excluded from the
exposed group (only the usual occupation was coded).
us many occupations, including vehicle mechanics,
showed elevated risk estimates. e authors noted that
it was not possible to determine if the cases in their
study resulted from exposure to the reported usual
occupation.
Four studies included in this review and meta-
analysis were PMR analyses. Two of these studies took
place in Great Britain (McElvenny etal., 2005; Health
and Safety Executive, 2013) and two in the USA
Page 10 of 19 • Mesothelioma among motor vehicle mechanics
by guest on November 5, 2015http://annhyg.oxfordjournals.org/Downloaded from

(Milham, 2011; NIOSH, 2011). ese studies did
not collect task-specic asbestos exposure histories
(i.e. brake repairman versus automobile mechanic) or
lifetime asbestos exposure histories. e PMR stud-
ies did not control for confounding by other asbestos
exposures and the reference group for each occupa-
tion was the general population, which included those
with known risk-related exposures to asbestos. ese
studies did not assess exposure-response or latency.
e cases of mesothelioma in these studies were not
necessarily based on pathologic review.
Meta-analysis results
Primary meta-analyses
Meta-analysis of all 16 eligible studies (all tiers com-
bined) of mesothelioma risk among motor vehicle
mechanics resulted in an SRRE of 0.80 (95% CI:
0.61–1.05), with statistically signicant heteroge-
neity [Q-test P-value (P-H) < 0.000, I
2
 = 62.90]
(Table3, Fig.2). Meta-regression by study quality
did not yield a statistically signicant eect (Fig.3).
e paerns of associations across tiers were rela-
tively similar with overlapping CIs (Fig.2). All ve
Tier 1 studies reported no increased risk of mesothe-
lioma among motor vehicle mechanics resulting in
an SRRE=0.76 (95% CI: 0.46–1.25), with no het-
erogeneity between studies (P-H=0.912, I
2
=0.00)
(Table3, Fig.2). In this model, the study by Hessel
contributed the most relative weight (38.18%),
while the study by Teta contributed the lowest
(5.51%). Meta-analysis of the ve Tier 2 studies
produced a non-signicant SRRE of 1.09 (95% CI:
0.76–1.58) with lile heterogeneity (P-H=0.637,
I
2
 = 0.00). ree of the individual study ORs in
this model were below 1.0 and two were above 1.0
(none of which was statistically signicant) and all
CIs overlapped. Meta-analysis of the Tier 3 studies
resulted in an SRRE of 0.73 (95% CI: 0.49–1.08),
with statistically signicant heterogeneity (P-
H<0.000, I
2
=81.39). Of the six Tier 3 studies, ve
reported point estimates ranging between 0.41 and
0.87, while one study (Roelofs etal., 2013) reported
a point estimate of 2.10. e study by Roelofs etal.
was identied as an outlier (overall and within the
Tier 3 model) based on visual inspection of a funnel
plot of standard error by log OR. Removal of this
outlier study in a sensitivity analysis modied the
Tier 3 eect size to 0.59, which was statistically sig-
nicant (95% CI: 0.44–0.79).
Combination of the Tier 1 and Tier 2 studies
produced a non-signicant SRRE (0.96, 95% CI:
0.72–1.29) with no heterogeneity (P-H = 0.845,
I
2
=0.00). e Tier 1 studies provided ~36% of the
relative weight in this model. is model was robust
to the inuence of any single study as reected by our
one-study-removed sensitivity analyses where SRREs
ranged between 0.87 (Rolland removed) and 1.01
(Hansen removed).
Four studies reported data specic to brake
work; meta-analysis of these studies resulted in an
SRRE=0.66 (95% CI: 0.39–1.10), with no appreci-
able heterogeneity (P-H=0.472, I
2
=0.00) (Table3,
Fig.4). All four studies reported ORs below1.0.
Sub-group meta-analyses
Meta-analysis by study design generated SRREs <1.0
for both case-control and PMR/SMOR/cohort stud-
ies (Table 3). While no heterogeneity was apparent
for the model including only case-control studies
(P-H = 0.911, I
2
 = 0.00), signicant heterogeneity
was found in the model including only studies that
used PMR/SMOR/cohort methods (P-H = 0.000,
I
2
 = 83.86). Analysis of only PMR/SMOR studies
resulted in an SRRE of 0.71 (95% CI: 0.46–1.10) with
signicant heterogeneity. Summary results of studies
conducted in the USA/Canada (SRRE = 0.93, 95%
CI: 0.69–1.27) diered slightly from the non-US/
Canadian studies (SRRE=0.70, 95% CI: 0.49–1.00),
but the CIs largely overlapped and removal of Roelofs
in the US/Canada model resulted in an SRRE of 0.79
(95% CI: 0.60–1.04). Analysis including only stud-
ies that controlled for confounding by other asbestos
exposure resulted in an SRRE of 0.77 (95% CI=0.46–
1.28) with no apparent heterogeneity.
Publication bias assessment
Statistical testing for possible publication bias pro-
duced signicant observations. Egger’s regression
test—a test for funnel plot asymmetry on a linear
regression of a standard normal deviate—produced a
P-value of 0.05. If a meta-analysis captured all relevant
studies, it would be expected that the funnel plot would
be symmetric, and that the studies would be dispersed
equally on both sides of the overall eect (Rothstein
etal., 2005). us, the Duval and Tweedie trim and
Mesothelioma among motor vehicle mechanics • Page 11 of 19
by guest on November 5, 2015http://annhyg.oxfordjournals.org/Downloaded from

Table3. Summary of meta-analysis results for mesothelioma among motor vehicle mechanics.
Model (number of studies) SRRE 95% CI P-value for
heterogeneity, I
2
Notes
Primary analyses
All studies (n=16) 0.80 0.61–1.05 0.000, 62.90 Tier 1, 2, 3 studies included
All studies (n=16), adjusted data
from Teschke
0.79 0.60–1.05 0.000, 62.90 Teschke: Cases and controls
removed if ever held at-risk
occupations
All studies, Roelofs excluded
(n=15)
0.72 0.57–0.91 0.018, 48.44 Roelofs identied as an
outlier study, and had the
lowest study quality rating
Tier 1 studies (n=5) 0.76 0.46–1.25 0.912, 0.00 Studies with the highest study
quality rating
Tier 1 studies (n=5), adjusted
data from Teschke
0.74 0.43–1.26 0.885, 0.00 Teschke: Cases and controls
removed if ever held at-risk
occupations
Tier 2 studies (n=5) 1.09 0.76–1.58 0.637, 0.00 Studies determined to be of
moderate quality
Tier 3 studies (n=6) 0.73 0.49–1.08 0.000, 81.39 Studies determined to be of
low methodological quality
a
Tier 1 and 2 studies combined
(n=10)
0.96 0.72–1.29 0.845, 0.00 Tier 1 and 2 studies included
Studies specic to brake work
(n=4)
0.64 0.38–1.09 0.486, 0.00 Analysis of studies that
reported data specically for
brake work
Sub-group analyses
Case-control studies
(n=10)
0.89 0.66–1.20 0.911, 0.00 Analysis of case-control stud-
ies only
SMOR/PMR/cohort studies
(n=6)
0.77 0.51–1.18 0.000, 83.86 Analysis of SMOR/PMR and
cohort studies
SMOR/PMR studies
(n=5)
0.71 0.46–1.10 0.000, 84.30 Analysis of SMOR/PMR
studies
Studies conducted in the
USA/Canada (n=7)
0.93 0.69–1.27 0.253, 23.12 Studies conducted in the
USA/Canada
Studies conducted in the
USA/Canada (n=6),
Roelofs excluded
0.79 0.60–1.04 0.998, 0.00 Roelofs identied as an out-
lier study, and had the lowest
study quality rating
Non-US/Canadian studies
(n=9)
0.70 0.49–1.00 0.004, 64.60 Studies conducted in Europe,
with the exception of one
(Mexico)
Page 12 of 19 • Mesothelioma among motor vehicle mechanics
by guest on November 5, 2015http://annhyg.oxfordjournals.org/Downloaded from

ll method (Duval and Tweedie, 2000) imputes data
to account for variable results between studies. Eight
data points were imputed using this method, result-
ing in a change of the overall model SRRE from 0.80
(95% CI: 0.61–1.05) to 0.52 (95% CI: 0.39–0.69),
further reinforcing the evidence of lack of an increased
risk of mesothelioma among motor vehicle mechanics
(Supplementary Fig. S1).
DISCUSSION
Our results indicate that mesothelioma risk is not asso-
ciated with either motor vehicle repair or with brake
repair work. e SRREs of mesothelioma showed no
trend either upward or downward according to study
quality. ere have been a number of studies published
since 2004, when we last reviewed this topic, including
four studies in Tiers 1 and 2 (Rake etal., 2009; Aguilar-
Madrid etal., 2010; Merlo etal., 2010; Rolland etal.,
2010), a new study in Tier 3 (Roelofs et al., 2013),
and updates to PMR studies (Milham, 2011; Health
and Safety Executive, 2013). e addition of this sub-
stantial body of scientic evidence did not change
the previously reported conclusion (Goodman etal.,
2004) of no association between mesothelioma and
either motor vehicle repair or brake repair work. e
single study that showed a signicantly increased risk
Figure2 Meta-analysis of mesothelioma risk among motor vehicle mechanics by study tier.
Model (number of studies) SRRE 95% CI P-value for
heterogeneity, I
2
Notes
Studies that controlled for confounding
by other asbestos exposures (n=4)
0.77 0.46–1.28 0.809, 0.00 Analysis of studies that con-
trolled for confounding due
to asbestos exposure outside
of motor vehicle repair
Lifetime asbestos exposure history
ascertained (n=9)
0.92 0.66–1.28 0.868, 0.00 Analysis of studies that
obtained information on
prior history of potential
asbestos exposure
a
ese studies were determined to be of lower quality but provided risk estimates for motor vehicle mechanic workers and mesothelioma, and were
evaluated in the context of ‘results consistency’ with other studies.
Table3. Continued
Mesothelioma among motor vehicle mechanics • Page 13 of 19
by guest on November 5, 2015http://annhyg.oxfordjournals.org/Downloaded from

of mesothelioma (Roelofs et al., 2013) had several
methodological deciencies and received the lowest
quality score. Moreover, analyses of heterogeneity
indicated that this study was appreciably dierent than
the entire body of literature on thistopic.
It should be emphasized that this body of research
was conducted using dierent study designs, in dif-
ferent populations in Europe and North America, by
dierent teams of researchers, and over a period span-
ning three decades. ese dierent circumstances
yielded consistent results with no appreciable hetero-
geneity. Although it can be argued that not all motor
vehicle mechanics are engaged in brake repairs, clearly
many are. While in theory exposure misclassication
at the occupational title level could miss an associa-
tion if the exposure was low level or of low frequency,
we sought to minimize this possibility by focusing on
studies that reported brake, clutch, and gasket repairs
specically and on studies that examined duration of
exposure. Moreover, the studies that examined brake
repairs specically (Woitowitz and Rodelsperger,
1994; Teschke etal., 1997; Hessel etal., 2004; Rake
etal., 2009) provided results that were not appreciably
dierent than the results for motor vehicle mechan-
ics across all studies. us, the lack of association
seen in the studies specic to exposure to brakes was
not meaningfully dierent than studies that were less
specic. is was not unexpected as brake repair is
a very common activity among vehicle mechanics.
Similarly, vehicle mechanics were commonly exposed
to asbestos from clutch and gasket repairs. e lack of
mesothelioma risk across studies indicates that vehicle
mechanics’ work with asbestos-containing clutches
and gaskets does not increase the risk for mesothe-
lioma. It is unlikely that the healthy worker eect
biased the results. Anumber of the studies compared
Figure3 Regression of log odds ratio on study quality. e circle size represents the weight in the regression analysis.
Figure4 Meta-analysis of mesothelioma risk among workers involved with brake repair.
Page 14 of 19 • Mesothelioma among motor vehicle mechanics
by guest on November 5, 2015http://annhyg.oxfordjournals.org/Downloaded from

vehicle mechanics to other working populations or
were case-control studies. Moreover, mesothelioma
has a very long latency and is oen diagnosed aer the
exposed workers have retired—it is dicult to imag-
ine a healthy worker eect exists 40+ years aer start-
ing work and into retirement.
Although some of the individual studies had limited
power to detect a statistically signicant association,
the meta-analysis had greatly increased study power as
reected in the narrow CIs. Furthermore, the SRREs
indicated associations near and slightly below 1.0.
us, power to detect a statistically signicant positive
association in individual studies was irrelevant.
In general, the relevant exposure periods of the
study populations coincided with the timeframe in
which chrysotile asbestos was used in brakes. In addi-
tion, the studies had adequate elapsed time from rst
exposure to case identication to detect mesothelioma
if this outcome had been causally related to brake expo-
sures. is is of particular importance because of the
long latency period of mesothelioma. For example,
in Woitowitz (Woitowitz and Rodelsperger, 1994),
the median rst date of exposure was 1949 for both
cases and controls, and case ascertainment occurred
between 1988 and 1994. In the study by Teschke etal.
(1997), where latency was considered, all of the vehi-
cle mechanics and brake workers began their exposures
prior to the minimum latency period (i.e. 20years).
Both study design and statistical heterogeneity are
important considerations in a rigorous meta-analytic
approach. Indeed, we made an eort to conduct an
objective and transparent study quality scoring exer-
cise based on recommendations by the MOOSE
(Stroup et al., 2000) and PRISMA (Moher et al.,
2009) guidelines. is facilitated several additional
meta-analyses based on parameters such as study
design, potential bias, the inuence of confounding,
and the specicity of exposure designations. SRREs
by study design, study quality, study region, and
ascertainment of lifetime asbestos exposure were cal-
culated. However, virtually all analyses produced the
same results—no association between mesothelioma
risk and either motor vehicle repair or brake repair.
Some study designs are inherently weaker than oth-
ers, such as PMR studies, which are not designed to
include lifetime occupational histories, task-specic
Table4. Summary of mesothelioma studies and corresponding RR estimates not included in the
meta-analysis.
First author Year Design RR estimate 95% CI Reason for exclusion
Malker
a
1985 Cohort 2.4 (P<0.01) NA Exposure dened as
‘mechanics’
Jarvholm 1988 Cohort NA (1 case) NA Unable to calculate
relative risk
Schiman 1988 Case-control 0 (no cases, 2 controls) NA Exposure dened as
‘mechanics’
Hansen 1989 Cohort NA (1 case) NA Unable to calculate
relative risk
Gustavsson 1990 Cohort NA (2 cases) NA Unable to calculate
relative risk
Pan
b
2005 Case-control 1.02 0.66– 1.57 Exposure dened as
‘mechanics, not specied’
Welch 2005 Case-control 1.50 0.43– 5.26 Exposure dened as ‘Tire
or brake lining work’
CI, condence interval; NA, not available; RR, estimate of relative risk.
a
Of the 16 mechanics with mesothelioma in this study, only one was likely to be an auto mechanic (W. Blot, personal communication, 2003).
b
Mechanics were not restricted to motor vehicle mechanics (D. Garabrant, personal communication, 2014).
Mesothelioma among motor vehicle mechanics • Page 15 of 19
by guest on November 5, 2015http://annhyg.oxfordjournals.org/Downloaded from

information, duration of employment, latency, or
covariates for other sources of exposure. ese types
of studies received zero scores on all of these criteria,
as they should. In the PMR studies, the results for
motor vehicle mechanics are below 1.0 but sometimes
are above other occupations. is likely reects the
fact that mechanics may be more likely than workers
in other occupations (e.g. farmers, teachers, clerical
workers) to have worked in other asbestos-exposed
jobs. For this reason, we believe the analyses that con-
trolled for other sources of asbestos exposure (Hessel
etal., 2004; Rake etal., 2009) provide a more reliable
estimate of RR and warrant great analytical weight.
ere were seven potentially relevant studies that
could not be included in the meta-analysis (Table4).
ree were excluded because no measures of associa-
tion were provided or could be calculated. Jarvholm
and Brisman (1988) reported a single case of meso-
thelioma in a cohort of 21 905 Swedish car mechanics
born between 1890 and 1945, and followed through
1979. Hansen (1989) reported a single case of meso-
thelioma in a cohort of 21 800 Danish auto mechanics
and followed from 1970 to 1980, accumulating 192
000 person years. Gustavsson etal. (1990) reported
two cases of mesothelioma among 695 Stockholm
bus garage workers followed from 1958 to 1994.
ree other studies (Malker et al., 1985; Schiman
etal., 1988; Pan etal., 2005) reported mesothelioma
risks for ‘mechanics’, but could not be included in the
meta-analysis because occupations were not restricted
to motor vehicle mechanics (W. Blot and H.Malker,
2003, personal communication; D. Garabrant and
M. Shenker, 2013, personal communication). e
seventh study (Welch etal., 2005), which compared
cases of peritoneal mesothelioma to cases of appen-
diceal cancer, was excluded because it grouped brake
installation and repair with tire installation and repair
and did not report brake repair activities or work as
a motor vehicle mechanic. ese studies did not pro-
vide evidence of mesothelioma risks among motor
vehicle mechanics.
e observed lack of association between work
as a motor vehicle mechanic or brake mechanic and
mesothelioma may result from several factors. First,
the asbestos bers found in dust and airborne sam-
ples of vehicle mechanics workplaces or simulated
workplaces are generally short (<5 μm) chrysotile
bers (Hatch, 1970; Anderson et al., 1973; Rohl
etal., 1976; Roberts and Zumwalde, 1982; Williams
and Muhlbaier, 1982; Cha etal., 1983; Rodelsperger
et al., 1986; Rodelsperger, 1987; NIOSH et al.,
1989). ere appears to be increasing consensus that
short bers, particularly those <5μm in length, are
associated with lile (if any) pathologic response,
including risk of lung cancer and mesothelioma
(Platek etal., 1985; Eastern Research Group, 2003;
Berman, 2011; Bernstein etal., 2013). In an analysis
of dust obtained from sanding brake drums manu-
factured with chrysotile, no signicant pathological
response was shown following short-term inhalation
in rats (Bernstein etal., 2014). Secondly, asbestos b-
ers in brake pads are embedded in resin (Weir and
Meraz, 2001) and therefore are less likely to become
airborne. irdly, much of the chrysotile in brake
pads is transformed to forsterite during the brak-
ing process (Anderson etal., 1973). Forsterite does
not appear to have asbestos ber properties and
is not considered carcinogenic in humans (Wong,
1992). More importantly, Langer demonstrated that
chrysotile’s biological activity becomes virtually nil
hundreds of degrees below the forsterite transforma-
tion temperature. us, complete transformation of
the mineral is not required to result in loss of activ-
ity (Langer, 2003). A large body of IH literature
indicates the mean time weighted average asbestos
exposure during brake servicing was ~0.04 bers per
cubic centimeter during the post-1974 time period
(Weil and Delpire, 1985).
Two studies addressed the risk of mesothelioma
among bystanders (i.e. persons presumed to be indi-
rectly exposed to asbestos) and do-it-yourself mechan-
ics (Hessel etal., 2004; Rake etal., 2009). Both studies
reported no association in these circumstances. If
workers whose occupation involving low-level chry-
sotile exposure are not associated with an increased
risk of mesothelioma, it follows that co-habitants of
these workers also would not have an increased risk
of mesothelioma (Goswami etal., 2013). Take-home
exposure from friction products has been estimated to
be lower than the exposures of automobile mechanics
(0.0001 f/cc versus 0.04 f/cc) (Goswami etal., 2013).
An association between an exposure and a disease
should clearly exist before a determination of causa-
tion is made (Hill, 1965). Even in the absence of an
association between employment as a motor vehi-
cle mechanic and mesothelioma, our meta-analysis
Page 16 of 19 • Mesothelioma among motor vehicle mechanics
by guest on November 5, 2015http://annhyg.oxfordjournals.org/Downloaded from

allowed evaluating some key considerations of a causal
relationship such as strength of association and con-
sistency of ndings. All of the studies with the excep-
tion of Roelofs et al., 2013 provided homogeneous
eect estimates showing no association. e SRREs
do not support an increased risk of mesothelioma.
Two studies evaluated the duration of work and nei-
ther found a positive association in the group with
the longest duration of work (Hessel etal., 2004; Peto
etal., 2009). Further, a positive dose-response paern
would be unlikely given the fact that few studies (and
of lower quality) reported risks above 1.0. Finally,
study-specic risk estimates were consistent across
studies. We sought to nd an eect by analyzing the
data in a number of alternative ways to ensure that the
results were not overly inuenced by any single study
or group of studies. erefore, based on this compre-
hensive quantitative assessment of the totality of epi-
demiologic literature, neither work as a motor vehicle
mechanic nor work in brake repair is associated with
an increased risk of mesothelioma. is conclusion is
clearly supported by the data reported in a large group
of studies spanning decades of research across multi-
ple study populations using a variety of study design
and exposure assessment methods.
SUPPLE MENTARYDATA
Supplementary data can be found at hp://annhyg.
oxfordjournals.org/.
ACKNOWLEDGEMENTS
ere was no direct nancial support for this work.
No one other than the listed authors prepared the
research material, wrote, reviewed, or approved the
submied manuscript. D.H.G, D.D.A., M.J.T., P.A.H.,
V.A.C., and M.K.have provided legal testimony and/
or consultancy on behalf of companies that manufac-
ture automobiles, brakes, clutches, and automotive
gaskets. P.B. has provided legal testimony in a case
of asbestos exposure in the manufacture of synthetic
polymers and risk of mesothelioma.
REFERENCE S
Agudo A, Gonzalez CA, Bleda MJ et al. (2000) Occupation
and risk of malignant pleural mesothelioma: a case-control
study in Spain. Am J Indust Med; 37: 159–68.
Aguilar-Madrid G, Robles-Perez E etal. (2010) Case-control
study of pleural mesothelioma in workers with social secu-
rity in Mexico. Am J Ind Med; 53: 241–51.
Anderson AE, Gealer RL, McCune RC etal. (1973) Asbestos
emissions om brake dynamometer tests. Scientic research
sta, Ford Motor Co., pp. 1–19.
Berman DW. (2011) Apples to apples: the origin and magni-
tude of dierences in asbestos cancer risk estimates derived
using varying protocols. Risk Anal; 31: 1308–26.
Bernstein D, Dunnigan J, Hesterberg T etal. (2013) Health risk
of chrysotile revisited. Crit Rev Toxicol; 43: 154–83.
Bernstein DM, Rogers R, Sepulveda R etal. (2014) Evaluation
of the deposition, translocation and pathological response
of brake dust with and without added chrysotile in compar-
ison to crocidolite asbestos following short-term inhalation:
Interim results. Toxicol Appl Pharmacol; 276: 28–46.
Breslow NE, Day NE. (1987) Statistical methods in cancer research.
Volume II: the design and analysis of cohort studies. Lyon, France:
International Agency for Research on Cancer. pp. 1–406.
Cha S, Carter P, Bradow RL. (1983) Simulation of automobile
brake wear dynamics and estimation of emissions. Warrendale,
PA: Society of Automotive Engineers, Inc. pp. 1–20.
DerSimonian R, Laird N. (1986) Meta-analysis in clinical trials.
Controlled Clinical Trials. New York, NY: Elsevier Science
Publishing Co., Inc. pp. 177–88.
Duval S, Tweedie R. (2000) Trim and ll: a simple funnel-plot-
based method of testing and adjusting for publication bias
in meta-analysis. Biometrics; 56: 455–63.
Eastern Research Group. (2003) Report on the expert panel
health eects of asbestos and synthetic vitreous bers: the
inuence of ber length. Atlanta, GA: Agency for Toxic
Substances and Disease Registry. pp. 1-1–5-6.
EPA. (1986a) Controlling brake dust to protect your health...What
every auto mechanic should know. Washington, DC: U.S.
Environmental Protection Agency, Asbestos Action Program.
EPA. (1986b) Guidance for preventing asbestos disease among
auto mechanics. Washington, DC: US Environmental
Protection Agency. pp. 1–16.
EPA. (1989) Asbestos; manufacture, importation, process-
ing, and distribution in commerce prohibitions: nal rule.
Federal Register; 54: 29460–513.
Freeman MD, Kohles SS. (2012) Assessing specic causation of
mesothelioma following exposure to chrysotile asbestos-con-
tainting brake dust. Int J Occup Environ Health; 18: 329–36.
Goodman M, Teta MJ, Hessel PA etal. (2004) Mesothelioma
and lung cancer among motor vehicle mechanics: a meta-
analysis. Ann Occup Hyg; 48: 309–26.
Goswami E, Craven V, Dahlstrom DL etal. (2013) Domestic
asbestos exposure: a review of epidemiologic and exposure
data. Int J Environ Res Public Health; 10: 5629–70.
Gustavsson P, Plato N, Lidston E-B etal. (1990) Lung cancer
and exposure to diesel exhaust among bus garage workers.
Scand J Work Environ Health; 16: 348–54.
Hansen ES. (1989) Mortality of auto mechanics. Aten-year
follow-up. Scand J Work Environ Health; 15: 43–6.
Hansen J, Meersohn A. (2003) Report on cancer and occu-
pation (1970–97). Kobenhavn, Denmark: Institut for
Epidemiologisk Kraeforskning. pp. 1–114.
Mesothelioma among motor vehicle mechanics • Page 17 of 19
by guest on November 5, 2015http://annhyg.oxfordjournals.org/Downloaded from

Hatch D. (1970) Possible alternatives to asbestos as a friction
material. Ann Occup Hyg; 13: 25–9.
Health and Safety Executive. (2013) Mesothelioma occupation
statistics: male and female deaths aged 16–74 in Great Britain
2002–2010. London: Health and Safety Executive.
Hennekens CH, Buring JE. (1987) Descriptive studies. In
Mayrent SL, editor. Epidemiology in medicine. Boston, MA:
Lile, Brown and Company. pp. 101–31.
Hessel PA, Teta MJ, Goodman M etal. (2004) Mesothelioma
among brake mechanics: an expanded analysis of a case-
control study. Risk Anal; 24: 547–52.
Hickish DE, Knight KL. (1970) Exposure to asbestos during
brake maintenance. Ann Occup Hyg; 13: 17–21.
Higgins JP, ompson SG. (2002) Quantifying heterogeneity
in a meta-analysis. Stat Med; 21: 1539–58.
Hill AB. (1965) e environment and disease: association or
causation. Proc R Soc Med; 58: 295–300.
IARC. (2012) A review of human carcinogens: arsenic, met-
als, bres and dusts (Asbestos Monograph). Lyon, France:
World Health Organization, International Agency for
Research on Cancer. pp. 219–309.
Jacko MG, DuCharme RT. (1973) Brake emissions: emission
measurements om brake and clutch linings om selected mobile
sources. Southeld, MI: Bendix Research Laboratories.
Jacko MG, Spurgeon WM, Rusnak RM etal. (1975) ermal
stability and fade characteristics of iction materials. USA:
Ford/e Bendix Corp. Research Lab. pp. 1–15.
Jarvholm B, Brisman J. (1988) Asbestos associated tumours in
car mechanics. Br J Ind Med; 45: 645–6.
Langer AM. (2003) Reduction of the biological potential of
chrysotile asbestos arising from conditions of service on
brake pads. Regul Toxicol Pharmacol; 38: 71–7.
Lemen . (2004) Asbestos in brakes: exposure and risk of
disease. Am J Ind Med; 45: 229–37.
Lorimer WV, Rohl AN, Miller A etal. (1976) Asbestos expo-
sure of brake repair workers in the United States. Mount
Sinai J Med; 43: 207–18.
Lynch JR. (1968) Brake lining decomposition products. J Air
Pollut Control Assoc; 18: 824–6.
Madl AK, Sco LL, Murbach DM et al. (2008) Exposure to
chrysotile asbestos associated with unpacking and repacking
boxes of automobile brake pads and shoes. Ann Occup Hyg;
52: 463–79.
Malker HSR, McLaughlin JK, Malker BK et al. (1985)
Occupational risks for pleural mesothelioma in Sweden,
1961–79. J Natl Cancer Inst; 74: 61–5.
McDonald AD, McDonald JC. (1980) Malignant mesothe-
lioma in North America. Cancer; 46: 1650–6.
McElvenny DM, Darnton A, Price MJ et al. (2005)
Mesothelioma mortality in Great Britain from 1968 to
2001. Occup Med; 55: 87.
Merlo DF, Stagi E, Fontana V etal. (2010) A historical mortal-
ity study among bus drivers and bus maintenance workers
exposed to urban air pollutants in the city of Genoa, Italy.
Occup Environ Med; 67: 611–9.
Milham S, Ossiander E. (2001) Occupational mortal-
ity in Washington State, 1950–1999. Washington State
Department of Health.
Milham S. (2011) Washington State occupational mortality
database, 1950–2010. Available at hps://fortress wa gov/
doh/occmort/OMQuery aspx [serial online] Accessed
31 July 2013.
Moher D, Liberati A, Tetzla J etal. (2009) Preferred report-
ing items for systematic reviews and meta-analyses: the
PRISMA statement. PLoS Med; 6: e1000097.
NIOSH. (2011) National occupational mortality surveillance
(NOMS). Available at hp://www cdc gov/niosh/top-
ics/surveillance/noms/icd-occ-form html [serial online]
Accessed 7 February 2014.
NIOSH; Sheehy JW, O’Brien DM, McGlothlin JD etal. (1989)
Control of asbestos exposure during brake drum service.
Cincinnati, OH: U.S. Department of Health and Human
Services. pp. 1–70.
Olsen JH, Jensen OM. (1987) Occupation and risk of cancer in
Denmark: an analysis of 93,810 cancer cases, 1970–1979.
Scand J Work Environ Health; 13: 1–91.
Pan XL, Day HW, Wang W etal. (2005) Residential proxim-
ity to naturally occurring asbestos and mesothelioma risk in
California. Am J Respir Crit Care Med; 172: 1019–25.
Paustenbach DJ, Finley BL, Lu ET etal. (2004) Environmental and
occupational health hazards associated with the presence of
asbestos in brake linings and pads (1900 to present): A“state-
of-the-art” review. J Toxicol Environ Health B; 7: 33–110.
Peto J, Rake C, Gilham C etal. (2009) Occupational, domestic
and environmental mesothelioma risks in Britain: a case-con-
trol study. London: Health and Safety Executive.
Platek SF, Groth DH, Ulrich CE etal. (1985) Chronic inhalation
of short asbestos bers. Fundam Appl Toxicol; 5: 327–40.
Rake C, Gilham C, Hatch J etal. (2009) Occupational, domes-
tic and environmental mesothelioma risks in the British
population: a case-control study. Br J Cancer; 110: 1175–83.
Roberts DR, Zumwalde RD. (1982) Industrial hygiene summary
report of asbestos exposure assessment for brake mechanics.
Cincinnati, OH: National Institute for Occupational Safety
and Health.
Rodelsperger K. (1987) Asbestos ber dust hazard om automo-
bile brakes. Giessen, German: Justus-Liebig University.
Rodelsperger K, Jahn H, Bruckel B etal. (1986) Asbestos dust
exposure during brake repair. Am J Ind Med; 10: 63–72.
Roelofs CR, Kernan GJ, Davis LK etal. (2013) Mesothelioma
and employment in Massachuses: analysis of cancer regis-
try data 1988–2003. Am J Ind Med; 56: 985–92.
Rohl AN, Langer AM, Wol MS etal. (1976) Asbestos expo-
sure during brake lining maintenance and repair. Environ
Res; 12: 110–28.
Rolland P, Gramond C, Berron H etal. (2005) Pleural meso-
thelioma: professions and occupational areas at risk among
humans. Saint-Maurice Cedex, France: Institut de Veille
Sanitaire. Available at hp://www.invs.sante.fr. Accessed
9 July 2007. pp. 1–6.
Page 18 of 19 • Mesothelioma among motor vehicle mechanics
by guest on November 5, 2015http://annhyg.oxfordjournals.org/Downloaded from

Rolland P, Gramond C, Lacourt A etal. (2010) Occupations
and industries in France at high risk for pleural mesotheli-
oma: a population-based case–control study (1998–2002).
Am J Ind Med; 53: 1207–19.
Rothstein HR, Suon AJ, Borenstein M. (2005) Publication
bias in meta-analysis: prevention, assessment, and adjustments.
Chichester, West Sussex: John Wiley & Sons Ltd.
Rowson DM. (1978) e chrysotile content of the wear debris
of brake linings. Wear; 17: 315–21.
Schiman MH, Pickle LW, Fontham E et al. (1988) Case-
control study of diet and mesothelioma in Louisiana.
Cancer Res; 48: 2911–5.
Spirtas R, Heineman EF, Bernstein L etal. (1994) Malignant
mesothelioma: aributable risk of asbestos exposure. Occup
Environ Med; 51: 804–11.
Spirtas R, Keehn R, Wright W etal. (1985) Mesothelioma risk
related to occupational or other asbestos exposure: pre-
liminary results from a case-control study. Am J Epidemiol;
122: 518.
Stroup DF, Berlin JA, Morton SC etal. (2000) Meta-analysis
of observational studies in epidemiology: a proposal for
reporting. Meta-analysis Of Observational Studies in
Epidemiology (MOOSE) group. JAMA; 283: 2008–12.
Teschke K, Morgan MS, Checkoway H et al. (1997)
Mesothelioma surveillance to locate sources of exposure to
asbestos. Can J Public Health; 88: 163–8.
Teta MJ, Lewinsohn HC, Meigs JW etal. (1983) Mesothelioma
in Connecticut, 1955–1977. Occupational and geographic
associations. JOM; 25: 749–56.
Weil S, Delpire L. (1985) Asbestos exposure - brakes and ambient.
Washington, DC: U.S. Environmental Protection Agency.
Weir FW, Meraz LB. (2001) Morphological characteristics of
asbestos bers released during grinding and drilling of fric-
tion products. Appl Occup Environ Hyg; 16: 1147–9.
Welch LS, Acherman YI, Haile E etal. (2005) Asbestos and
peritoneal mesothelioma among college-educated men. Int
J Occup Environ Health; 11: 254–8.
Williams RL, Muhlbaier JL. (1982) Asbestos brake emissions.
Environ Res; 29: 70–82.
Woitowitz HJ, Rodelsperger K. (1994) Mesothelioma among
car mechanics? Ann Occup Hyg; 38: 635–8.
Wong O. (1992) Chrysotile asbestos, mesothelioma, and
garage mechanics. Am J Ind Med; 21: 449–51.
World Trade Organization. (2000) European communities-
measures aecting asbestos and asbestos-containing products.
Report of the Panel. pp. 1–465.
Mesothelioma among motor vehicle mechanics • Page 19 of 19
by guest on November 5, 2015http://annhyg.oxfordjournals.org/Downloaded from