title candidate - research-repository.uwa.edu.au · the work described in this thesis was not...
TRANSCRIPT
1
Title
Assessment of the Performance of the Australian Asbestos Job Exposure Matrix
(AsbJEM)
Candidate
Hiroyuki Kamiya, MD, PhD, MM
This thesis is presented for the degree of Master of Philosophy of The University of
Western Australia
School of Population and Global Health
Occupational Respiratory Epidemiology
2018
2
3
Thesis declaration
I, Hiroyuki Kamiya, certify that this thesis has been substantially accomplished during
enrolment in the degree.
This thesis does not contain any material which has been accepted for the award of any
other degree or diploma in my name, in any university or other tertiary institution.
No part of this work will, in the future, be used in a submission in my name, for any
other degree or diploma in any university or other tertiary institution without the prior
approval of The University of Western Australia and where applicable, any partner
institution responsible for the joint-award of this degree.
This thesis does not contain any material previously published or written by another
person, except where due reference has been made in the text and, where relevant, in
the Declaration that follows.
The works are not in any way a violation or infringement of any copyright, trademark,
patent, or other rights whatsoever of any person.
The work described in this thesis was not funded by any grant although the Asbestos
Review Program is continually funded by the Western Australia Department of
Health.
This thesis does not contain any work that I have published, nor work under review
for publication.
The research in this thesis was assessed and exempted from ethics review at the
University of Western Australia Human Research Ethics Committee (RA/4/20/4746).
4
This thesis contains work prepared for publication, some of which has been co-authored.
Hiroyuki Kamiya Peter Franklin
(Student) (Coordinating supervisor)
5
ABSTRACT
Introduction and Aims
Australia used to be one of the largest per capita producers and consumers of asbestos
(Soeberg et al., 2018). Although the use of asbestos-containing products was
completely banned in Australia in 2003 (Soeberg et al., 2018), the incidence of
malignant mesothelioma has just reached its peak in recent years (Musk et al., 2017)
and the prevalence of all asbestos-related diseases (ARDs) remain high (Park et al.,
2008). In Australia, previous reports of assessing risk of ARDs were mostly based on
workers employed in mines and mills (de Klerk et al., 1989) and
general-population-based data regarding occupational asbestos exposures is scarce
although there are large numbers of former workers who are at risk of developing
ARDs (Park et al., 2008).
A job exposure matrix (JEM) is a cross-tabulation of job titles in a variety of industries
and exposure estimates, which enables systematic estimates of occupational asbestos
exposure (Pannett et al., 1985; Hardt et al., 2014; Offermans et al., 2014; Peters et al.,
2016). The Australian Asbestos Job Exposure Matrix (AsbJEM) has been developed to
improve the assessment of dose-response relationships between asbestos exposure and
the diseases as well as to better predict future occurrence of ARDs for Australian
workers (van Oyen et al., 2015). The objective of the research was to assess the
reliability of the AsbJEM and investigate the assumptions that were used to construct
it. We also aimed to clarify dose-response relationships between asbestos exposure and
non-malignant ARDs such as asbestosis and pleural plaques using the AsbJEM.
Methods and Results
6
Two studies were undertaken for this thesis. The first project assessed the relationship
between asbestos exposures, estimated using the AsbJEM, and malignant
mesothelioma. This study was also used to check some of the original assumptions of
the AsbJEM to determine if it needed to be refined. The second project used the
amended AsbJEM to investigate dose-response relationships between asbestos and
both asbestosis and pleural plaques. For both projects subjects were the
non-Wittenoom workers who participated in an asbestos health surveillance program,
the Asbestos Review Program (ARP).
Project 1: This study investigated the relationship between estimated asbestos exposure
and malignant mesothelioma. Exposure was estimated for the original AsbJEM
assumptions and variations of these assumptions. Modifications included duration and
frequency of “peak” and intensity of “mode” exposures. The association of asbestos
exposure level with the occurrence of malignant mesothelioma was estimated using
Cox proportional hazards model for each modification of the AsbJEM. The validity of
the tool was confirmed by the significant association and the best estimate of asbestos
exposure was determined by the steepest dose-response relationship calculated from
these models (Lenters et al., 2011). There were 40 cases of malignant mesothelioma
out of 1602 participants eligible for the analysis. Both cumulative asbestos exposure
and the average intensity of asbestos exposure were significantly associated with the
incidence of malignant mesothelioma using all of the variations of the AsbJEM. The
risk of mesothelioma was estimated as HRs of 1.66, 3.00 and 5.71 for cumulative
asbestos exposures of 0.93, 2.23 and 12.3 fibre-years/ml, respectively, compared with
the reference exposure of 0.09 fibre-years/ml. The model that provided the best
7
estimate (steepest slope) was the model that doubled the frequency of peak exposure of
the published AsbJEM.
Project 2: The modified AsbJEM was applied to the same cohort of participants to
estimate the risk of both asbestosis and pleural plaques. For this study the association
of asbestos exposure with these non-malignant ARDs was estimated using logistic
regression. Both cumulative asbestos exposure and the average intensity of asbestos
exposure were found to be significantly associated with both asbestosis and pleural
plaques. The risk of asbestosis was estimated as odds ratios (ORs) of 1.27, 1.35 and
1.58 for cumulative asbestos exposures of 0.41, 1.11 and 14.9 fibre-years/ml,
respectively compared with the reference exposure of 0.002 fibre-years/ml. The risk of
pleural plaques was estimated with the ORs of 1.25, 1.33 and 1.54 for cumulative
asbestos exposures of 0.30, 1.00 and 16.4 fibre-years/ml, respectively compared with
the reference exposure of 0.004 fibre-years/ml.
Conclusions
These findings indicated that the AsbJEM provides a reasonable estimate of
occupational asbestos exposure including low dose exposure and is an effective tool to
estimate the risk of developing ARDs.
8
TABLE OF CONTENTS
Thesis declaration ........................................................................................................ 3
Abstract ........................................................................................................................ 5
Acknowledgement ..................................................................................................... 12
Authorship declaration ............................................................................................... 13
Significance of research ............................................................................................. 15
List of tables ............................................................................................................... 17
List of figures ............................................................................................................. 18
Chapter One: Introduction ......................................................................................... 19
1.1 Background ........................................................................................................ 20
1.1.1 History of asbestos use in Australia .............................................................. 20
1.1.2 Asbestos burden in Australia ......................................................................... 21
1.1.3 Uncertainty about the relationship between asbestos exposure and
asbestos-related disease .............................................................................................. 22
1.1.4 Asbestos exposure estimates ......................................................................... 24
1.1.5 Asbestos exposure estimates in Australia and the Asbestos Job Exposure
Matrix ......................................................................................................................... 28
1.1.6 Validation of a Job Exposure Matrix ............................................................. 31
1.2 Aim ..................................................................................................................... 33
9
1.3 Overview of thesis.............................................................................................. 33
Chapter Two: Project 1 Validation of the Asbestos Job Exposure Matrix (AsbJEM) in
Australia: Exposure-Response Relationships for Malignant Mesothelioma ............. 34
2.1 Abstract .............................................................................................................. 35
2.2 Introduction ........................................................................................................ 37
2.3 Methods .............................................................................................................. 38
2.3.1 Subjects ......................................................................................................... 38
2.3.2 Asbestos exposure estimations ...................................................................... 39
2.3.3 Manipulation of the AsbJEM ........................................................................ 40
2.3.3.1 Intensity of mode exposure ...................................................................... 40
2.3.3.2 Duration of peak exposure ....................................................................... 41
2.3.3.3 Frequency of peak exposure..................................................................... 41
2.3.4 Case ascertainment ........................................................................................ 41
2.3.5 Statistical analysis ......................................................................................... 42
2.4 Results ................................................................................................................ 44
2.4.1 Characteristics of subjects ............................................................................. 44
2.4.2 Univariate analysis ........................................................................................ 48
2.4.3 Multivariate analysis ..................................................................................... 49
2.4.4 Selection of the best estimation of asbestos exposure .................................. 53
10
2.4.5 Validity of statistical assumptions ................................................................. 54
2.5 Discussion .......................................................................................................... 55
2.6 Conclusion ......................................................................................................... 58
Chapter Three: Project 2 Relationship between asbestos exposure and asbestosis and
pleural plaque in Australian workers ......................................................................... 59
3.1 Abstract .............................................................................................................. 60
3.2 Introduction ........................................................................................................ 62
3.3 Methods .............................................................................................................. 63
3.3.1 Population ..................................................................................................... 63
3.3.2 Asbestos exposure estimation ....................................................................... 63
3.3.3 Case ascertainment ........................................................................................ 64
3.3.4 Statistical analysis ......................................................................................... 65
3.4 Results ................................................................................................................ 67
3.4.1 Characteristics of subjects ............................................................................. 67
3.4.2 Risk estimates of asbestosis .......................................................................... 69
3.4.3 Risk estimates of pleural plaques .................................................................. 71
3.4.4 The fit of the model ....................................................................................... 74
3.5 Discussion .......................................................................................................... 74
3.6 Conclusion ......................................................................................................... 78
11
Chapter Four: Conclusion .......................................................................................... 79
References .................................................................................................................. 87
Appendices ............................................................................................................... 102
12
ACKNOWLEDGEMENTS
I appreciate all the advice and support that our research members in Occupational
Respiratory Epidemiology group provided with me. In particular, I would like to thank
my principal and coordinating supervisor, Dr. Peter Franklin, for his constant assistance
and considerate care that he demonstrated in my entire research period so that I could
keep up with my work and proceed to the right direction.
I would also like to express my gratitude to my family. Minako, as my wife, always
stood by my side and encouraged me to overcome any difficulty that I encountered
throughout my research. Morina, my beloved daughter, was always charming and her
sight relieved my fatigue.
I acknowledge that I could not have completed my research and made this achievement
without all of these people.
13
AUTHORSHIP DECLARATION: CO-AUTHORED PUBLICATIONS
I, Hiroyuki Kamiya, certify that the projects in this thesis were conceived by Dr.
Peter Franklin, my principal supervisor, and supported by other research members in
the Occupational Respiratory Epidemiology group, the University of Western
Australia.
All data required to conduct the research were collected from patients in the
Asbestos Review Program and provided with me in de-identified form by the data
administrator, Mr Robin Mina.
I undertook all the analyses myself and completed the thesis on my own. I take
full responsibility for the integrity of the data and the accuracy of the analysis.
The two main projects will be prepared for publication, namely:
1. Validation of the Asbestos Job-Exposure Matrix (AsbJEM) in Australia:
Exposure-Response Relationships for Malignant Mesothelioma
2. Relationship between asbestos exposure and asbestosis and pleural plaques in
Australian workers
I am the primary author of all the manuscripts and my contribution to these has
been over 75%. All co-authors have provided their consent to include these
manuscripts in this thesis and agreed with this estimated contribution by me.
Susan Peters Nicholas de Klerk
14
Bill Musk Fraser Brims
Alison Reid Nita Sodhi-Berry
Hiroyuki Kamiya
(Student)
8/08/2018
I, Peter Franklin, certify that the student statements regarding his contribution to each
of the works listed above are correct.
Peter Franklin
(Coordinating supervisor)
8/08/2018
15
Significance of research
Asbestos is carcinogenic and causes a number of malignant and non-malignant
asbestos-related diseases (ARDs). Although the complete ban of the use, manufacture
and import of all asbestos and asbestos-containing products was enacted in 2003 in
Australia, there are a large number of former workers exposed occupationally to
asbestos who are still at the risk of developing ARDs. Accurate asbestos exposure
estimates are important to evaluate the risk of future occurrence of ARDs for people
with occupational asbestos exposure since low-quality exposure assessments are
reported to underestimate dose-response relationships. However, population-based data
on occupational asbestos exposure is scarce in Australia with the relationship between
asbestos exposure and ARDs mostly reported based on the research of miners and
millers. The tool that we have constructed called the Asbestos Job Exposure Matrix
(AsbJEM) is based on comprehensive review of previous reports on asbestos
concentrations in a workplace and expert assessments by occupational hygienists. It is
specifically focused on working conditions in Australia and the use of asbestos
products across the nation. The AsbJEM is characterized by its coverage of a wide
range of jobs and industries, which to date includes 537 combinations of 224
occupations and 60 industries over four time periods since 1943 (this is increasing).
Asbestos exposures were estimated quantitatively using two exposure indicators, mode
and peak, and a change of exposure over time was also captured. Although a JEM is
subject to misclassification of exposure estimates and some assumptions used to
construct the AsbJEM have been questioned, it was demonstrated in this thesis that the
AsbJEM is a valid tool to estimate occupational asbestos exposure for Australian
workers. Significant relationships between occupational asbestos exposure and
16
mesothelioma, asbestosis and pleural plaques were all observed in this cohort. The
AsbJEM could improve the risk assessment of ARDs for low-dose exposure as it
detected mesothelioma cases that developed at low-dose estimates of asbestos
exposure. The instrument could be applied to other cohorts of Australian workers with
occupational asbestos exposure to assess dose-response relationships between asbestos
exposure and future occurrence of ARDs because a quantitative evaluation can be
systematically executed to assess occupational asbestos exposure.
17
List of tables
Table 1 Comparison of participants in a cohort with and without mesothelioma ...... 45
Table 2 Comparison of cumulative exposure estimated with different assumptions for
participants with and without mesothelioma ............................................................. 45
Table 3 Comparison of average intensity of asbestos exposure estimated with different
assumptions for participants with and without mesothelioma ................................... 47
Table 4 Regression coefficients of cumulative asbestos exposure estimated from
multivariate Cox proportional hazards models based on different assumptions of mode
intensity and peak duration and frequency................................................................. 49
Table 5 Regression coefficients of average intensity of asbestos exposure estimated
from multivariate Cox proportional hazards models based on different assumptions of
mode intensity and peak duration and frequency ....................................................... 51
Table 6 Risk of mesothelioma associated with cumulative asbestos exposure estimated
from the multivariate Cox proportional hazards model ............................................. 53
Table 7 Comparison of participants in a cohort with and without asbestosis ............ 68
Table 8 Comparison of participants in a cohort with and without pleural plaques .... 68
Table 9 Risk of asbestosis estimated from the multivariate logistic regression model ..
.................................................................................................................................... 69
Table 10 Risk of pleural plaques estimated from the multivariate logistic regression
model .......................................................................................................................... 72
18
List of figures
Figure 1 Risk of asbestosis by cumulative asbestos exposure and the time since first
asbestos exposure ....................................................................................................... 71
Figure 2 Risk of pleural plaques by cumulative asbestos exposure and the time since
first asbestos exposure ............................................................................................... 73
19
Chapter One: Introduction
20
1.1 Background
1.1.1 History of asbestos use in Australia
Asbestos is a fibrotic mineral that exists in natural environment (Dumortier et al.,
1998). It was used extensively in the building and construction industry because of its
excellence as fire-proofing, heat-insulation and resistance (Gray et al., 2016). Australia
used to be one of the largest producers and consumers of asbestos around the world
(Soeberg et al., 2018). Over 700,000 tons of asbestos were consumed in the 1970s and
asbestos consumption per capita was the highest around the globe at that time (Allen et
al., 2018). Asbestos was mined in various places in Australia, in particular, Wittenoom
in Western Australia (crocidolite) (Musk et al., 2008) and Woodsreef in New South
Wales (chrysotile) (Soeberg et al., 2016). Australia also imported both raw asbestos
such as chrysotile and amphibole (crocidolite and amosite) and asbestos-containing
products such as friction material, gaskets and cement from other countries (Soeberg et
al., 2018). In Australia almost all of the asbestos fibres was used in the asbestos cement
manufacturing industry (Soeberg et al., 2018). Asbestos consumption started declining
from 1960s in the United Kingdom and the United States and its use has been banned
for a few decades in many developed countries due to its carcinogenicity (Allen et al.,
2018). In Australia, after Wittenoom and Woodsreef were closed in 1966 and 1983,
respectively, bans on the use of asbestos in building products began in the 1980s and
the complete ban of the use, manufacture and import of all asbestos and
asbestos-containing products was enacted in 2003 (Soeberg et al., 2018). Nevertheless,
many asbestos-containing products and materials remain in situ in residential and
non-residential buildings in Australia (Gray et al., 2016).
21
1.1.2 Asbestos burden in Australia
Asbestos fibre can be released into the air and inhaled into the lungs and causes
malignant diseases such as malignant mesothelioma and lung cancer (Markowitz,
2015). It also causes non-malignant parenchymal and pleural disorders such as
asbestosis, diffuse pleural thickening and pleural plaques. Although non-malignant
diseases are not immediate threats to life in contrast to malignant diseases, the former
diseases impair pulmonary function (Park et al., 2014) and increase the risk of the
malignant diseases (Markowitz et al., 2013). Asbestosis is a progressive disease that
can lead to premature death (Markowitz et al., 1997). Despite the asbestos bans there
has been an increasing number of patients who suffer from these asbestos-related
diseases (ARDs) and it is expected to grow further in the future (Kwak et al., 2017).
This is mainly due to historical exposures and the long period of latency for most
ARDs (Reid et al., 2014). In Australia, the incidence of mesothelioma started rising
progressively since 1970s and has demonstrated an exponential increase thereafter
(Leigh and Driscoll, 2003). Although it seems to have finally plateaued in recent years,
newly diagnosed mesothelioma cases still amounted to 650 in 2015 (Musk et al., 2017)
and the age-standardized incidence rate is the second highest in the world behind the
United Kingdom, at 3.2 per 100,000 (Bianchi and Bianchi, 2014). In addition,
mesothelioma death rate is still one of the highest around the world (Furuya et al.,
2018). Previous mesothelioma incidence and death are mostly attributed to two waves
of asbestos exposure in Australia. The first wave involved the miners and millers of
asbestos and workers involved in manufacturing asbestos-containing products while
the second one included people working with and using the products. A recently
described third wave relates to people exposed to asbestos in situ (occupationally and
22
environmentally or domestically) and makes additional contributions to the incidence
and death from mesothelioma (Musk et al., 2017; Armstrong and Driscoll, 2016). As
asbestos remains ubiquitous and widespread in the built environment, some people are
still at the risk of occupational asbestos exposure in certain jobs or industries such as
removal and demolition work (Park et al., 2013) while non-occupational asbestos
exposure such as home renovation and maintenance is also a concern (Olsen et al.,
2011).
1.1.3 Uncertainty about the relationship between asbestos exposure and
asbestos-related disease
Some ARDs, such as mesothelioma and asbestosis, are noted to be caused exclusively
by asbestos exposure, while others such as lung cancer are also influenced by other
carcinogenic agents including cigarettes, air pollutants and other occupational agents
(Hodgson and Darnton, 2000). Although it is easily conceivable that the risk of ARDs
would be enhanced with increasing dose of asbestos exposure, the exact relationship
between asbestos exposure and the development of ARDs remains uncertain, in
particular, at low dose exposures. This is mainly because of a variety of methodologies
for asbestos exposure estimates and different circumstances for asbestos exposure. It is
also related to the fact that most of the previous studies only reported asbestos
exposure in high-risk industries (de Klerk et al., 1989; Gasparrini et al., 2008; Ferrante
et al., 2016; Clin et al., 2011; Lacourt et al., 2014).
Many authors use the term, low level asbestos exposure, to describe non-occupational
exposure such as air pollution from asbestos mines (Hansen et al., 1998), domestic
exposure for the family of an asbestos worker (Magnani et al., 1993) and home
23
renovation (Olsen et al., 2011). However, occupational exposure can also occur in
non-classic industries with relatively low dose exposure and some of these examples
include paper industry (Järvholm et al., 1988), sugar refineries (Maltoni et al., 1994)
and jewellery workers (Kern et al., 1992). In addition, exposure to in-situ asbestos in
demolition work under current regulations can also cause low dose exposure (Hillerdal,
1999). Given the extensive use of asbestos in many industries, a number of former
workers may also possibly have been exposed to certain levels of asbestos during their
working life and it is difficult or almost impossible to reveal all of these possible
exposure estimates (Hillerdal, 1999). For these reasons the risk at low dose exposure is
still controversial and extrapolation from high to low risk based on inferential
statistical models is also reported to be fraught with uncertainty (Case et al., 2011).
Therefore, dose-response relationships between asbestos exposure and ARDs should
be further scrutinized, in particular, focusing on low level asbestos exposure and
obtaining detailed and accurate information regarding potential asbestos exposure
which would improve the prediction of future occurrence of the diseases (Lenters et
al., 2011).
Although mesothelioma is a lethal disease regardless of early diagnosis and
intervention, other diseases such as lung cancer may be curable or controlled with
appropriate treatment (Maemondo et al., 2010) and thus it is imperative to closely
follow up people who are at a higher risk of developing the disease and not to miss the
chance for the best therapeutic option available (Musk et al., 2017).
24
1.1.4 Asbestos exposure estimates
It is usually difficult to precisely estimate asbestos exposure. The gold standard of
estimating asbestos burden for an individual person is counting fibres in pulmonary
tissues, which are usually procured from diseased lung (Dufresne A et al., 1995;
Warmock, 1989; Dodson et al., 2003). Dufresne et al (1995) found that the mean fibre
concentration were over 1.0 million/g dry lung in all miners with mesothelioma
irrespective of fibre types. Similarly, Warmock (1989) reported the median
concentration of amphibole fibres as 2.7 million/g dry lung for construction workers
with mesothelioma while former workers in the township of Wittenoom in Western
Australia demonstrated higher mean concentration of fibres of 187 million/g dry lung
(de Klerk et al., 1996). These findings indicate substantial variation in fibre counts
between studies, which is likely due to different working conditions, fibre types and
the difference in procedures to count fibres in pulmonary specimens. As this technique
is almost always performed only on patients with the disease, it is inappropriate to
estimate the risk of ARDs for people who have not developed the disease (Case et al.,
2011). Furthermore, asbestos exposure assessment using this methodology may
possibly overestimate asbestos burden for the development of the disease because it is
only based on diseased cases (Lijmer et al., 1999). Alternatively, bronchoalveolar
lavage fluid (BALF) analysis may be implemented to assess asbestos exposure as it is
less invasive (Cruz et al., 2017). It is reported that asbestos body concentration in
BALF corresponds with occupational asbestos exposure and is related to the presence
of asbestosis or pleural plaques (Nuyts et al., 2017). However, an application of this
procedure to healthy people also gives rise to an ethical issue. It is also likely to be
affected by the experience of operators (Pairon et al., 1994). Although fibre counts
25
inside the body are considered to directly reflect asbestos burden (Velasco-Garcia et
al., 2011), it is unethical and unrealistic to apply these diagnostic procedures to the
general population and thus inappropriate for epidemiological studies to investigate the
relationship between asbestos exposure and the occurrence of ARDs for cohorts with
and without the disease.
The risk of ARDs for workers in certain jobs or industries used to be evaluated by
atmospheric dust sampling. The asbestos concentration in a workplace was calculated
by counting airborne asbestos fibres collected on a filter with microscopy (Baron et al.,
2001). Using this technique de Klerk et al (1996) demonstrated a positive correlation
between airborne asbestos concentration and asbestos fibre counts in pulmonary tissues
in Wittenoom miners and millers while Clin et al (2011) revealed a dose-response
relationship between occupational asbestos exposure calculated by a membrane filter
method and cancer incidence. However, there is uncertainty about the accuracy of this
historical airborne asbestos measurement due to changes over time in measurement
devices and sampling and counting procedures (Case et al., 2011). In addition,
although this method is useful to estimate asbestos exposure in a specific working
environment, it will be insufficient to estimate the risk of ARDs in a population-based
study as participants usually have had multiple jobs where asbestos exposure data may
not be available (Hansell, 2014). Self-reported exposure assessments may be the most
convenient under these circumstances (Villeneuve et al., 2012). However, the
limitation of this method is recall bias as former workers quite often do not remember
details of their previous occupational exposure (Hardt et al., 2014). As a result, it is
well recognized that this method could inflate risk estimates (Delclos et al., 2009;
Quinlan et al., 2009; Adegoke et al., 2004; de Vocht et al., 2005). In addition,
26
self-reported exposure assessments will not usually allow detailed analysis of
dose-response relationship because quantitative exposure data can hardly be obtained
from all exposed individuals. Case-by-case expert assessments will overcome this
limitation of self-reported exposure estimates and were reported to provide accurate
estimates of asbestos exposure (Bourgkard et al., 2013). Using this exposure
assessment method French research groups found dose-response relationship between
cumulative asbestos exposure and mesothelioma where they assigned very high
intensity of exposure to a category of >10 fibres/ml (Iwatsubo et al., 1998; Lacourt et
al., 2014), while an Italian group identified the same trend of dose-response
relationship by assigning the highest level of exposure to a different category of
30-300 fibres/ml (Ferrante et al., 2016). As demonstrated by these reports, there will be
a variation of exposure estimates between studies since working conditions are usually
different depending on regions and exposure estimates will be affected by the quality
of available data and expertise of occupational hygienists. Although expert
assessments are usually used as the reference in a population-based study (Bourgkard
et al., 2013), they are costly and time-consuming as they require experts to evaluate the
level of asbestos exposure on an individual basis (Nam et al., 2005).
A job-exposure matrix (JEM) is a tool that could overcome these problems and provide
a method for estimating occupational asbestos exposure in population-based studies
(Pannett et al., 1985; Coughlin and Chiazze, 1990; Fletcher et al., 1993; Offermans et
al., 2014; Peters et al., 2016). A JEM is depicted as a matrix that contains occupations
and/or industries on one axis and an exposure index such as probability of exposure,
annual mean exposure, intensity and frequency on the other axis (Peters et al., 2016;
Kauppinen et al., 1998; Offermans et al., 2012; Choi et al., 2017). A time axis may
27
also be included and exposure estimates are expressed in each cell of the matrix
semiquantitatively or quantitatively. JEMs are constructed based on expert judgement
of occupational hygienists, published literature and/or databases of direct measures
(Hardt et al., 2014). As JEMs can systematically and cost-effectively evaluate
occupational exposure within a diverse workforce and avoid drawbacks caused by
self-reporting assessments, it has become increasingly common in epidemiological
research involving occupational exposure estimates (Pannett et al., 1985; Coughlin and
Chiazze, 1990; Fletcher et al., 1993; Offermans et al., 2014; Peters et al., 2016). A
number of studies investigated the performance of JEMs compared with other
methodologies to estimate occupational exposure. Hardt et al (2014) demonstrated
more consistent exposure estimates with previous findings if they were based on a
JEM than on self-reported exposure assessments. Similarly, Fletcher et al (1993)
reported that a JEM could be more reliable for estimating asbestos exposure than a
simple question. Pukkala et al (2005) concluded that the JEM-based method gave valid
results. Furthermore, Peters et al (2011) stated that expert assessment approach provids
little, if any, advantage over JEM approach for exposure estimates. The odds ratio for
mesothelioma for low and high asbestos exposure for the NOHS-JEM were 2.0 and
2.5, respectively (Cicioni et al., 1991) while Offermans et al (2014) reported that the
hazard ratio for mesothelioma for ever versus never-exposed people was 3.0 by
FINJEM and 2.6 by DOMJEM. These results derived from JEMs seem to be consistent
between studies as opposed to those obtained from other exposure assessment
methods. This may be partly because a variance of exposure estimates between
individual persons was reduced by categorization based on job titles (Pannett et al.,
1985; Coughlin and Chiazze, 1990; Fletcher et al., 1993).
28
Although the efficiency and reliability of JEMs are emphasized, they also contain
some methodological limitations. These include a lack of coverage of some jobs and/or
industries, insufficient information of a change over time and limited generalizability.
A JEM generated in one country will not necessarily be applicable to other countries
because working conditions will be diverse and types of exposed agents may also be
different. As a result, although a large number of research groups constructed their
own JEMs including FINJEM in Finland (Kauppinen et al., 1998), DOM-JEM
(Offermans et al., 2012) and SYN-JEM in Netherlands (Peters et al., 2016) and
GPJEM in Korea (Choi et al., 2017) to investigate the risk of ARDs, only the FINJEM
has been adopted to other studies conducted in a different region (Offermans et al.,
2014).
1.1.5 Asbestos exposure estimates in Australia and the Asbestos Job Exposure Matrix
In Australia, previous reports of assessing risk of ARDs were mostly based on workers
in mines and mills where exposure estimates were conducted by dust sampling or
post-mortem tissue examination of mesothelioma cases registered in a national
surveillance program that sought to reveal lung asbestos fibre burden. For Wittenoom
miners and millers, de Klerk et al (1989) reported that the odds ratio for mortality from
mesothelioma was 1.26 for one-unit increase in fibre concentration (fibres/ml) while
Rogers et al (1991) stated that the risk of mesothelioma increased by 29.4 with a
10-unit increase in fibre concentration in pulmonary tissues (fibres/microgram) for
mesothelioma cases reported to the Australian Mesothelioma Surveillance Program.
Population-based data on the risk of ARDs with occupational asbestos exposure is
scarce although there are a large number of former workers who are at the risk of
29
developing ARDs. As dose-response relationships between asbestos exposure and the
occurrence of ARDs may be different between miners and millers and other
occupations (Churg and Wiggs, 1986), it is necessary to clarify the risk estimates for
occupational asbestos exposure using exposure assessment tools specific to these
circumstances. In 2010 Hyland et al (2010) reviewed Australian asbestos exposure
measurements for the purpose of providing reference exposure levels that may assist in
determining a worker’s retrospective asbestos exposure and constructed a device called
asbestos task exposure matrix (ASTEM). It was a combination of 12 asbestos-related
tasks and 8 asbestos products over 2 time periods, which was derived from data on
occupational asbestos exposure in New South Wales (Hyland et al., 2010). The
potential advantage of this tool over JEMs was that it could address the variability
between workers within an occupation and day-to-day variability within an individual
worker (Benke et al., 2000). However, the limitation was that the majority of estimates
came from the manufacturing industry and, therefore, might not reflect the conditions
of other industries. In addition, as the ASTEM does not cover the time prior to 1970
due to the lack of reliable data, it cannot be used for occupations before this time
period. There will be a number of former workers who had been exposed to asbestos
occupationally before 1970 that are still at high risk of developing ARDs. Furthermore,
as ASTEM was only derived from data in New South Wales, it may not be applicable
nationwide. To date there have been no published reports investigating the risk of
ARDs with this device for former workers who used to be engaged in asbestos-related
tasks.
30
We recently created a tool called the Asbestos Job Exposure Matrix (AsbJEM), which
is specifically focused on working conditions in Australia and the use of asbestos
products in the whole nation (van Oyen et al., 2015). The AsbJEM is characterized by
its coverage of a wide range of jobs and industries, which included 537 combinations
of 224 occupations and 60 industries over four time periods dating back to 1943. Based
on all available hygiene measurements asbestos exposures were estimated
quantitatively by a panel of qualified and experienced industrial hygienists using two
exposure indicators, mode and intensity and a change of exposure over time was also
captured by time axis. These features are similar to other JEMs such as
INTEROCC-JEM (McElvenny et al., 2018) and NOCCA-JEM (Kauppinen et al.,
2009) where cumulative exposure could be calculated quantitatively. Other established
JEMs such as ECOJEM (Le Moual et al., 2000) and DOMJEM (Olsson et al., 2017)
calculate semiquantitative exposures by classifying exposure into three categories
according to the probability of exposure in an occupation. Therefore, the AsbJEM is
expected to improve the assessment of dose-response relationship between asbestos
exposure and future occurrence of ARDs for Australian workers as a quantitative
evaluation can be systematically executed to assess occupational asbestos exposure
(van Oyen et al., 2015). Furthermore, as the AsbJEM covers the time period after the
complete ban of the use of asbestos-containing products, it could be used to estimate
occupational exposure to in-situ asbestos remaining in the built environment, which
will be associated with a lower dose exposure than other traditional jobs that used
asbestos-containing materials. Although the risks of low dose asbestos exposure are
still uncertain because most of the previous findings were only derived from relatively
high-risk industries (de Klerk et al., 1989; Gasparrini et al., 2008; Ferrante et al., 2016;
31
Clin et al., 2011; Lacourt et al., 2014), the AsbJEM could improve the estimates of
dose-response relationship at lower exposures.
1.1.6 Validation of a Job Exposure Matrix
The AsbJEM has yet to be validated and its performance for estimating exposures also
remains unclear. The accuracy of JEMs can be influenced by the quality of available
data and expertise of the occupational hygienists who developed it. In addition,
misclassification of exposure estimates is also concerning as JEMs disregard potential
differences in exposure among individuals with the same job titles (Fletcher et al.,
1993). Indeed, some questions have been raised against the AsbJEM regarding the
accuracy of its estimates of occupational asbestos exposure (Kottek and Kilpatrick,
2016). Although the reference standard is usually required to confirm the accuracy of a
diagnostic tool, there is no gold standard to confirm asbestos exposure. Counting fibres
in pulmonary tissues can directly estimate asbestos burden. It is, however, not ethical
to people who have not developed a disease. Instead, expert assessments on an
individual basis are usually implemented to evaluate the performance of JEMs
(Solovieva et al., 2012; Sjöström et al., 2013). However, it is also noted that it can
suffer from misclassification and thus the consistency of exposure estimates between
JEMs and expert assessments does not necessarily guarantee the validity of the tool. As
a result it is difficult to achieve a true validation for exposure estimates within a JEM
using traditional methods. On the other hand, Bouyer and Hemon (1993) and Goldberg
et al (1993) suggested that estimates within a JEM could be considered reliable if they
could detect known associations between risk factors and disease occurrence, for
example exclusive association between asbestos exposure and malignant mesothelioma.
32
In addition, Lenters et al (2011) reported that better quality exposure assessment is
associated with steeper slopes of the dose-response relationships. Based on these
previous reports we aimed to conduct a validation study to prove the accuracy of the
AsbJEM by applying the tool to a cohort of former Australian workers with a history
of asbestos exposure to investigate the dose-response relationship between
occupational asbestos exposure and the development of mesothelioma. According to
the previous reports (Bouyer and Hemon, 1993; Goldberg et al, 1993), the AsbJEM
would be validated if a significant relationship between asbestos exposure, which is
estimated by the instrument, and mesothelioma is confirmed. In addition, the published
AsbJEM could be refined by comparing different assumptions of estimating asbestos
exposure and identifying the steepest slope of the dose-response relationship. A
previous report explained that when a JEM is implemented to estimate a dose-response
relationship, approximately three times more subjects are required to reach the same
precision in the estimation of the odds ratio than that achieved with experts because of
a loss of information between workers in the same job (Bouyer et al., 1995). However,
the same article also stated that an unbiased JEM could provide a statistical power
close to that expected in practice with a good expert (Bouyer et al., 1995). Therefore,
the validated and possibly refined AsbJEM could further be applied to a cohort of
former Australian workers to cost-effectively assess and clarify dose-response
relationship for other ARDs, in particular, non-malignant diseases such as asbestosis
and pleural plaques.
33
1.2 Aim
The aim of the research was:
firstly to validate the AsbJEM as a tool to estimate occupational asbestos exposure and
refine it if it can be improved to produce more reliable estimates,
secondly to estimate exposure-response relationships between occupational asbestos
exposure and non-malignant ARDs such as asbestosis and pleural plaques using the
AsbJEM.
1.3 Overview of thesis
This thesis is composed of reports of two research projects. The first is a validation
study for the AsbJEM, which was created to standardize the estimates of occupational
asbestos exposure and is expected to help evaluate precisely the risk of developing
ARDs. The second is an epidemiological study, which aims to estimate
exposure-response relationships for asbestosis and pleural plaques, using the AsbJEM
with some modification, based on the findings obtained from the first project.
34
Chapter Two: Project 1
Validation of the Asbestos Job Exposure Matrix (AsbJEM) in Australia:
Exposure-Response Relationships for Malignant Mesothelioma
35
2.1 Abstract
Introduction
An Asbestos Job Exposure Matrix (AsbJEM) to estimate cumulative asbestos exposure
for Australian jobs has been published. However, certain assumption in its calculations
has been questioned. We aimed to assess the reliability of the AsbJEM by evaluating
exposure-response relationships between exposure estimates and the incidence of
malignant mesothelioma. A further aim is to investigate the assumptions used in the
development of the AsbJEM.
Subjects
Eligible subjects were participants in an annual health surveillance program who had at
least 3-months’ occupational history with likely asbestos exposure.
Methods
Asbestos exposure estimates included cumulative asbestos exposure and the average
intensity of asbestos exposure, which were calculated using intensity and frequency of
exposure in the AsbJEM and duration of employment. Various exposure estimates
were calculated by manipulating assumptions about the intensity of ‘mode’ exposure
and the duration and frequency of ‘peak’ exposure. Cox proportional hazards models
were used to describe the associations between mesothelioma incidence and each of
the calculated cumulative exposure estimates and the average intensity estimates.
Results
Data were collected from 1602 male participants. Of these, 40 developed mesothelioma
during the study period (incidence rate; 45 per 100,000 person-years). Twenty-four
36
exposure estimates were calculated using three exposure intensities (50th
, 75th
and 90th
percentile of the range of mode exposure), 4 peak durations (15, 30, 60 and 120 minutes)
and 2 patterns of peak frequency. There were significant associations between
mesothelioma incidence and both cumulative asbestos exposure and the average intensity
of asbestos exposure for all estimates. The strongest association, based on the
regression-coefficient from the proportional hazards models, was found for the 50th
percentile of mode exposure and duration of 15 minutes and doubled, from the original
model, frequency of peak exposure. Using these assumptions, the Hazard Ratios for
mesothelioma were 1.66, 3.00 and 5.71 for cumulative asbestos exposures of 0.93, 2.23
and 12.3 fibre-years/ml, respectively, compared with the reference exposure of 0.09
fibre-years/ml. It was also estimated as 1.84, 2.31 and 4.40 for the average intensity of
asbestos exposure of 0.03, 0.06 and 0.42 fibres/ml, respectively, compared with the
reference exposure of 0.003 fibres/ml.
Conclusion
We have validated the AsbJEM by detecting a positive exposure-response trend
between mesothelioma incidence and both estimated cumulative asbestos exposure and
average intensity of asbestos exposure. The strongest relationship was found when the
frequency of peak exposure was doubled from the original published version. The
AsbJEM may be a useful and non-invasive tool for identifying exposure-response
relationships between occupational asbestos exposures and other asbestos related
diseases.
37
2.2 Introduction
Asbestos is a known cause of a number of diseases, including malignant mesothelioma,
lung cancer and asbestosis (Jamrozik et al., 2011). Exposure-response relationships
have been established between asbestos exposure and these asbestos-related diseases
(ARDs) (Clin et al., 2011; Paris et al., 2009; Lenters, et al., 2011; Olsson, et al., 2017).
However, there has been a considerable variation in the slope of the exposure-response
relationship across individual studies, which is partly due to the diversity of the quality
of the exposure assessments in different studies (Lenters et al., 2011). A robust
measure of exposure is important as poor quality exposure estimates will lead to
misclassification of exposure and misleading exposure-response relationships.
A Job Exposure Matrix (JEM) is a standardized method to transform job histories into
specific exposure-estimates. It is shown as a cross-tabulation of job titles in various
industries and the estimated exposure of each job (Pannett, 1985). JEMs have become
an increasingly common method to systematically and cost-effectively evaluate
occupational exposures within the wider workforce for a variety of exposure-agents in
various industries and occupations over different time periods and are specific for
different countries (Coughlin and Chiazze, 1990; Peters et al., 2016). The use of a JEM
allows standardized exposure assessment and reduces reporting bias (Hardt et al.,
2014). However, the accuracy of a JEM depends on the quality of the exposure data
and can be subject to intra-occupational misclassification since workers at the same job
are not necessarily exposed to the same level of occupational asbestos (Coughlin and
Chiazze, 1990; Bouyer et al., 1995).
38
While true validation of estimates within a JEM can rarely be achieved, the
performance of estimates can be evaluated. Bouyer and Hemon (1993) and Goldberg et
al (1993) suggested that estimates within a JEM could be considered reliable if they
could detect known associations between risk factors and a disease, for example the
well-established association between asbestos exposure and malignant mesothelioma.
We recently published an Asbestos Job Exposure Matrix (AsbJEM) to estimate
occupational asbestos exposure for Australian workers (van Oyen et al., 2015).
However, it has been suggested that exposures may have been underestimated for
some job titles because a higher level of asbestos exposure was confirmed for many
reported cases of ARDs for these jobs (Kottek and Kilpatrick, 2016). Particular
concern was raised about the duration and frequency of peak exposure for some of the
job titles (Kottek and Kilpatrick, 2016). Therefore, the aim of this study was, firstly, to
validate the AsbJEM by assessing the exposure-response relationship with
mesothelioma based on the fact that it is almost exclusively caused by asbestos
exposure and secondly, to investigate different assumptions of duration, intensity and
frequency of exposure so as to refine the AsbJEM if required.
2.3 Methods
2.3.1 Subjects
Subjects were participants of an annual health surveillance program for people with
significant occupational or environmental exposure to asbestos (the Asbestos Review
Program (ARP)) (Hansen et al., 1997; Hansen et al., 1998; Musk et al., 1998). People
eligible to join the ARP include former workers of the crocidolite mine and mill at
Wittenoom, Western Australia, ex-residents of the mining township of Wittenoom,
39
people who have had at least 3-months full-time jobs that involved exposure to
asbestos, and people who had evidence of pleural plaques. Asbestos exposure
estimates for the former Wittenoom workers (de Klerk et al., 1989) and ex-residents of
the Wittenoom township (Hansen et al., 1997) have previously been published and
validated (de Klerk et al., 1996). Only asbestos-exposed workers, who were not former
workers or residents of Wittenoom, were included in this study. This study was
exempted from ethics review by the University of Western Australia Human Research
Ethics Committee (RA/4/20/4746).
2.3.2 Asbestos exposure estimations
Cumulative asbestos exposure (fibre-years/ml) for each participant in the cohort was
estimated using the AsbJEM (van Oyen et al., 2015). Exposure estimates for
occupation-industry combinations over four time periods (1943-1966, 1967-1986,
1987-2003 and ≥2004) were evaluated by an expert panel of industrial hygienists
familiar with likely job exposures. ‘Mode’ was the most common exposure level and
‘peak’ was the short-term intense exposure in a particular job. ‘Background’ was the
same as the exposure from the general environment. The frequency and intensity of
mode, peak and background exposures were pre-defined for each time period, except
for the frequency of background exposure, which was calculated by subtracting the
mode frequency from the total number of working days in the working year (240 days
assuming two days off a week and four weeks of holidays). The background, peak and
mode exposures were each calculated as the multiplicative product of intensity and
frequency. The average exposure was calculated as the sum of all these exposures
divided by 240. This exposure index is expressed as fibre-year/ml and 1 fibre-year/ml
40
indicated that a daily exposure of 1 fibre/ml continued for one year. Accordingly,
cumulative asbestos exposure for an individual in a specific occupation was calculated
as the annual average exposure for that job multiplied by the length of employment
(van Oyen et al., 2015).
2.3.3 Manipulation of the AsbJEM
The AsbJEM was manipulated regarding the intensity of mode exposure, and the
duration and frequency of peak exposure. The exposure-response relationship was
iteratively examined in order to determine the best estimation of exposure, which was
determined by the slope of the relationship. A steeper slope was assumed to indicate a
better estimation of asbestos exposure as it suggests less exposure misclassification
and less underestimation of exposure-response relationships (Lenters et al., 2011).
2.3.3.1 Intensity of mode exposure
The intensity of exposure was classified into background (0.0001 fibres/ml), low
(0.01-0.1 fibres/ml), medium (0.1-1 fibres/ml), high (1-25 fibres/ml) and very high
(25-50 fibres/ml) (van Oyen et al., 2015). The AsbJEM adopted the mid-point of the
range of exposure in each category to represent the intensity of mode exposure.
However, it is possible that for ARP participants, that value may have underestimated
exposure because the AsbJEM is estimating the average occupational asbestos
exposure for the industry/occupation pairing while the subjects in this study joined the
ARP due to their awareness that they may have undertaken tasks which were at the
higher end of the occupation-exposure combination. Therefore, we assessed whether
the 75th and 90th percentile of the range of exposure better represented the level of
exposure for this cohort than the 50th percentile.
41
2.3.3.2 Duration of peak exposure
During the development of the AsbJEM, the duration of peak exposure was arbitrarily
set at 15 minutes. This had been discussed extensively at that time and was finalized
based on the expertise of occupational hygienists. However, as pointed out by Kottek
and Kilpatrick (2016), “many of the historic reports of peak exposure are short-term
measurements made during a task that was carried out for a much longer period”.
Therefore, we further assessed other peak exposure durations, specifically 30, 60 and
120 minutes.
2.3.3.3 Frequency of peak exposure
Frequency of exposure peaks was distributed at 1, 2, 11 or 48 days per year across
various industry-occupation combinations in the AsbJEM, which was built on the
experts’ assessment of exposure in a particular job or industry. However, Kottek and
Kilpatrick (2016) mentioned the possibility of the expert panel underestimating the
frequency. Therefore, we also investigated whether twice the original frequency of
peak exposure, in particular for the first two time periods (<1943 and 1943-1966),
would result in a stronger exposure-response relationship for malignant mesothelioma.
2.3.4 Case ascertainment
All incident cases of mesothelioma and related deaths in Western Australian are
reported to the Western Australian Mesothelioma Registry and the Australian
Mesothelioma Registry. Cases were determined by data linkage of ARP participants
with the Mesothelioma Registry. The Registry records both the date of diagnosis and
42
the date of death, if the patient had died. Data were available until 31st Dec 2014,
which served as our censoring date for this study.
2.3.5 Statistical analysis
When the association of asbestos exposure with mesothelioma incidence is estimated,
cumulative asbestos exposure will often be utilized as the exposure index. However,
cumulative exposure is calculated as the multiplication of the average asbestos
exposure estimated using the AsbJEM and the length of employment, which can be
obtained without the tool. Accordingly, exposure-response relationship estimated using
cumulative exposure will be affected by the information not related to the AsbJEM and
as a result the meaning of the tool will be undermined. Therefore, exposure-response
relationship between asbestos exposure and the occurrence of malignant mesothelioma
was evaluated in two multivariate statistical models using Cox Proportional Hazards
regression. The rate of mesothelioma incidence with a particular cumulative asbestos
exposures and the average intensity of asbestos exposure was calculated in each model,
respectively. Time to mesothelioma diagnosis following entry into the ARP was
defined as the outcome variable while explanatory variables included cumulative
asbestos exposure in the first model and the average intensity of asbestos exposure and
duration of exposure in the second model. Duration of exposure was not included in
the first model due to its high collinearity with cumulative asbestos exposure.
Cumulative asbestos exposure was re-calculated based on new assumptions of the
intensity of mode exposure and the duration and frequency of peak exposure as
described above. The average intensity of asbestos exposure was calculated as the
division of cumulative asbestos exposure estimated using the AsbJEM by duration of
43
exposure, which was defined as the total number of years of occupational asbestos
exposure above the background level.
For all calculations the distribution of cumulative asbestos exposure and the average
intensity of asbestos exposure were skewed and logarithmic transformation was
performed and used for the analyses. The incidence rate of mesothelioma was
calculated as the number of mesothelioma cases divided by person-years since the
entry to the ARP. Age at first exposure was included in all models regardless of its
statistical significance as this is known to influence mesothelioma incidence (Reid et
al., 2007).
The linearity of continuous variables such as cumulative asbestos exposure was
examined in two different ways. Firstly, the log rank test for trend was performed to
see the linear trend for each variable by categorizing it into four groups using quartiles.
Secondly, it was checked graphically by plotting the β coefficient for each group,
which was estimated from the model, against the mid-point of each quartile. If the
linear assumption had been violated, the variables were handled as categorical.
The presence of the effect modification of cumulative asbestos exposure and the
average intensity of asbestos exposure by other potential confounders was evaluated
using interaction terms. A variable with the least significance showing p value over
0.01 was eliminated one by one from the model until it reached the least variables.
The proportional hazards assumption was evaluated in two different ways. Firstly, the
Schoenfeld residuals of all explanatory variables were examined graphically.
Secondly, a time-dependent covariate for each explanatory variable was created and its
significance was checked in a multivariate Cox proportional hazards model.
44
Continuous data were expressed as either mean (standard deviation) or median
(interquartile range). Student’s t-test or Wilcoxon rank sum test were performed for a
comparison of the mean or median of two independent groups, respectively. For
categorical data, a chi-square or Fisher’s exact tests were conducted. Statistical
significance was set at p-value ≤0.05. The risk of developing the outcome was
estimated by the Hazard Ratio (HR). The strongest association of mesothelioma
incidence with cumulative asbestos exposure and the average intensity of asbestos
exposure was identified by the largest value of the coefficient estimated from the Cox
regression model. Accordingly, the AsbJEM model that provided the steepest slope of
the relationship, was considered as the best estimate of occupational asbestos exposure
based on the fact that mesothelioma is caused exclusively by asbestos exposure. The
result estimated from the model with the average intensity as the asbestos exposure
index was prioritized over cumulative exposure to discover the best estimates of
asbestos exposure if inconsistent results were obtained for the same reason as
described above. Participants lost to follow-up were assumed to have lived up to the
end of study or 90 years of age. All statistical analyses were conducted using SAS 9.4
(SAS Institute Inc., Cary, NC, USA).
2.4 Results
2.4.1 Characteristics of subjects
A total of 2084 people, who were not former workers or ex-residents of Wittenoom,
attended the ARP as of December, 2015. Of these 1602 men were included in the
analyses after excluding possible non-occupational asbestos exposure (n=388),
participants with an uncertain occupational history (n=81) and women (n=13). At data
45
linkage of ARP participants with the Mesothelioma Registry in 2015, 40 cases of
mesothelioma were identified out of all the participants included in the analysis (45
cases per 100,000 person-years) (Table 1). Compared to participants who did not
develop mesothelioma, the mean age at first exposure was significantly lower for
mesothelioma cases (15.7±2.2 vs. 17.0±4.6, p=0.0008) and the mean duration of
exposure was significantly longer (37.5±11.2 vs. 31.9±13.9 years, p=0.01) (Table 1).
Table 1 Comparison of participants in a cohort with and without mesothelioma
Mesothelioma
(n=40)
No mesothelioma
(n=1562)
p-value
Duration of exposure (years) 37.5±11.2 31.9±13.9 0.01
Age at first exposure (years) 15.7±2.2 17.0±4.6 0.0008
Time since first exposure (years) 55.1±8.2 56.8±10.5 0.33
Time since entry to the ARP (years) 8.7±6.1 12.1±7.6 0.001
Data expressed as mean±SD; ARP, Asbestos Review Program;
The median of cumulative asbestos exposure was significantly higher for
mesothelioma cases compared with non-cases for all calculations of the AsbJEM
(Table 2).
Table 2 Comparison of cumulative exposure estimated with different assumptions for
participants with and without mesothelioma
Mesothelioma
(n=40)
No mesothelioma
(n=1562)
p-value
Median cumulative exposure (IQR) (fibre-years/ml)
Original frequency of peak exposure
46
Mode 50%_Peak duration_15 min 1.63 (3.72) 0.49 (1.22) <0.0001
30 min 2.11 (4.33) 0.72 (1.77) <0.0001
60 min 2.53 (8.01) 1.13 (2.66) <0.0001
120 min 4.35 (13.40) 1.76 (4.82) 0.0001
Mode 75%_Peak duration_15 min 1.96 (4.96) 0.61 (1.59) <0.0001
30 min 2.78 (6.12) 0.85 (2.15) <0.0001
60 min 3.36 (8.21) 1.31 (3.04) <0.0001
120 min 4.53 (15.49) 2.01 (5.10) 0.0001
Mode 90%_Peak duration_15 min 2.02 (5.83) 0.68 (1.81) <0.0001
30 min 3.09 (6.94) 0.92 (2.32) <0.0001
60 min 3.86 (8.41) 1.39 (3.35) <0.0001
120 min 4.78 (15.96) 2.16 (5.24) <0.0001
Doubled frequency of peak exposure
Mode 50%_Peak duration_15 min 1.93 (4.30) 0.57 (1.42) <0.0001
30 min 2.32 (6.18) 0.85 (2.13) <0.0001
60 min 3.36 (11.35) 1.32 (3.76) <0.0001
120 min 5.99 (17.21) 2.11 (6.71) <0.0001
Mode 75%_Peak duration_15 min 2.16 (5.71) 0.71 (1.76) <0.0001
30 min 3.05 (6.87) 0.98 (2.52) <0.0001
60 min 3.95 (12.17) 1.56 (3.93) <0.0001
120 min 6.25 (20.12) 2.37 (7.25) <0.0001
Mode 90%_Peak duration_15 min 2.20 (6.42) 0.78 (2.00) <0.0001
30 min 3.53 (7.82) 1.09 (2.73) <0.0001
60 min 4.41 (12.25) 1.66 (4.13) <0.0001
47
120 min 6.50 (12.79) 2.51 (7.38) <0.0001
Mode 50%, 75% and 90% indicate the 50th, 75
th and 90
th percentile of the range of mode
intensity, respectively. Mode 50%_Peak duration_15 min indicates mode intensity set at the
50th percentile of the range of mode intensity and peak duration of 15 minutes. Other labels are
defined likewise. IQR, interquartile range;
Mesothelioma cases also demonstrated higher average intensity of asbestos exposure
than participants without mesothelioma for all calculations of the AsbJEM (Table 3).
Table 3 Comparison of average intensity of asbestos exposure estimated with different
assumptions for participants with and without mesothelioma
Mesothelioma
(n=40)
No mesothelioma
(n=1562)
p-value
Median of average intensity of asbestos exposure
(IQR) (fibres/ml)
Original frequency of peak exposure
Mode 50%_Peak duration_15 min 0.04 (0.09) 0.02 (0.04) 0.0007
30 min 0.06 (0.12) 0.03 (0.05) 0.0011
60 min 0.08 (0.21) 0.04 (0.09) 0.0018
120 min 0.12 (0.40) 0.06 (0.15) 0.0028
Mode 75%_Peak duration_15 min 0.05 (0.11) 0.02 (0.06) 0.0007
30 min 0.07 (0.14) 0.03 (0.07) 0.0008
60 min 0.09 (0.22) 0.05 (0.10) 0.0014
120 min 0.14 (0.41) 0.07 (0.16) 0.0024
Mode 90%_Peak duration_15 min 0.06 (0.13) 0.02 (0.06) 0.0006
48
30 min 0.07 (0.16) 0.03 (0.08) 0.0007
60 min 0.10 (0.24) 0.05 (0.10) 0.0011
120 min 0.15 (0.41) 0.08 (0.18) 0.0021
Doubled frequency of peak exposure
Mode 50%_Peak duration_15 min 0.04 (0.10) 0.02 (0.05) 0.0007
30 min 0.06 (0.16) 0.03 (0.07) 0.0009
60 min 0.11 (0.27) 0.05 (0.12) 0.0011
120 min 0.18 (0.54) 0.07 (0.22) 0.0013
Mode 75%_Peak duration_15 min 0.06 (0.13) 0.02 (0.06) 0.0006
30 min 0.07 (0.18) 0.04 (0.08) 0.0008
60 min 0.12 (0.28) 0.06 (0.13) 0.0010
120 min 0.20 (0.53) 0.08 (0.23) 0.0011
Mode 90%_Peak duration_15 min 0.06 (0.15) 0.03 (0.07) 0.0006
30 min 0.08 (0.19) 0.04 (0.09) 0.0007
60 min 0.13 (0.30) 0.06 (0.13) 0.0010
120 min 0.21 (0.54) 0.09 (0.24) 0.0011
Mode 50%, 75% and 90% indicate the 50th, 75
th and 90
th percentile of the range of mode
intensity, respectively. Mode 50%_Peak duration_15 min indicates mode intensity set at the
50th percentile of the range of mode intensity and peak duration of 15 minutes. Other labels are
defined likewise. IQR, interquartile range;
2.4.2 Univariate analysis
The age at first exposure was handled as a categorical variable as the linear assumption
failed to hold by both the log rank test for trend and a graphical examination while
49
cumulative asbestos exposure, average intensity of asbestos exposure and duration of
exposure were all included as continuous variables in Cox proportional hazards model.
Univariate analyses with Cox proportional hazards model revealed that mesothelioma
incidence was significantly associated with cumulative asbestos exposure and average
intensity of exposure for all calculations of the AsbJEM. There was also significant
association between mesothelioma incidence and duration of exposure but the age at
first exposure demonstrated no significant association with the disease.
2.4.3 Multivariate analysis
All explanatory variables of interest were included to build multivariate Cox
proportional hazards model regardless of their statistical significance in univariate
analysis. As a result, the first model was composed of cumulative exposure and the age
at first exposure while the second model consisted of average intensity of exposure,
duration of exposure and the age at first exposure. As there was also no effect
modification of cumulative asbestos exposure and the average intensity of asbestos
exposure by other variables in each model, the final models for the analysis were set as
described above.
After adjusting for the age at first exposure, there was a significant association of
mesothelioma incidence with cumulative asbestos exposure for all calculations of the
AsbJEM (Table 4).
Table 4 Regression coefficients of cumulative asbestos exposure estimated from multivariate
Cox proportional hazards models based on different assumptions of mode intensity and peak
duration and frequencya
50
Median cumulative exposure (IQR)
(fibre-years/ml)
β-coefficient (SE) p-value
Original frequency of peak exposure
Mode 50%_Peak duration_15 min 0.2683 (0.0777) 0.0006
30 min 0.2668 (0.0809) 0.0010
60 min 0.2595 (0.0832) 0.0018
120 min 0.2472 (0.0837) 0.0032
Mode 75%_Peak duration_15 min 0.2588 (0.0748) 0.0005
30 min 0.2606 (0.0781) 0.0008
60 min 0.2579 (0.0810) 0.0015
120 min 0.2497 (0.0827) 0.0025
Mode 90%_Peak duration_15 min 0.2546 (0.0737) 0.0005
30 min 0.2576 (0.0769) 0.0008
60 min 0.2566 (0.0801) 0.0014
120 min 0.2503 (0.0822) 0.0023
Doubled frequency of peak exposure
Mode 50%_Peak duration_15 min 0.2723 (0.0794) 0.0006
30 min 0.2712 (0.0826) 0.0010
60 min 0.2648 (0.0845) 0.0017
120 min 0.2550 (0.0848) 0.0026
Mode 75%_Peak duration_15 min 0.2629 (0.0764) 0.0006
30 min 0.2653 (0.0798) 0.0009
60 min 0.2631 (0.0826) 0.0015
120 min 0.2565 (0.0840) 0.0023
Mode 90%_Peak duration_15 min 0.2587 (0.0753) 0.0006
51
30 min 0.2623 (0.0787) 0.0009
60 min 0.2618 (0.0818) 0.0014
120 min 0.2568 (0.0836) 0.0021
a: Models including age at first asbestos exposure as a covariate aside from cumulative
asbestos exposure; Mode 50%, 75% and 90% indicate the 50th, 75
th and the 90
th percentile of
the range of mode intensity, respectively; IQR: interquartile range; SE: standard error;
After adjusting for duration of exposure and the age at first exposure, there was also a
significant association of mesothelioma incidence with the average intensity of
exposure for all calculations of the AsbJEM (Table 5).
Table 5 Regression coefficients of average intensity of asbestos exposure estimated from
multivariate Cox proportional hazards models based on different assumptions of mode
intensity and peak duration and frequencya
Median of average intensity of asbestos
exposure (IQR) (fibres/ml)
β-coefficient (SE) p-value
Original frequency of peak exposure
Mode 50%_Peak duration_15 min 0.2344 (0.0795) 0.0032
30 min 0.2291 (0.0824) 0.0054
60 min 0.2181 (0.0841) 0.0095
120 min 0.2033 (0.0841) 0.0157
Mode 75%_Peak duration_15 min 0.2275 (0.0765) 0.0029
30 min 0.2259 (0.0796) 0.0046
60 min 0.2191 (0.0822) 0.0076
120 min 0.2077 (0.0833) 0.0127
52
Mode 90%_Peak duration_15 min 0.2243 (0.0753) 0.0029
30 min 0.2239 (0.0784) 0.0043
60 min 0.2190 (0.0813) 0.0070
120 min 0.2091 (0.0829) 0.0116
Doubled frequency of peak exposure
Mode 50%_Peak duration_15 min 0.2367 (0.0811) 0.0035
30 min 0.2317 (0.0839) 0.0057
60 min 0.2220 (0.0851) 0.0091
120 min 0.2101 (0.0849) 0.0134
Mode 75%_Peak duration_15 min 0.2300 (0.0780) 0.0032
30 min 0.2286 (0.0812) 0.0049
60 min 0.2226 (0.0835) 0.0077
120 min 0.2132 (0.0843) 0.0115
Mode 90%_Peak duration_15 min 0.2268 (0.0769) 0.0032
30 min 0.2268 (0.0801) 0.0046
60 min 0.2224 (0.0827) 0.0072
120 min 0.2142 (0.0840) 0.0108
a: Models including duration of exposure and age at first asbestos exposure as a covariate aside
from the average intensity of asbestos exposure; Mode 50%, 75% and 90% indicate the 50th,
75th and the 90
th percentile of the range of mode intensity, respectively; IQR: interquartile
range; SE: standard error;
However, no significant association was identified between mesothelioma incidence
and the age at first exposure in both models (p>0.05).
53
2.4.4 Selection of the best estimation of asbestos exposure
The results demonstrated that a revised AsbJEM with doubled peak frequency and no
change for other indexes (the 50th
percentile of the range of each exposure category as
mode intensity and peak duration of 15 minutes) produced the strongest dose-response
relationship between mesothelioma incidence and cumulative asbestos exposure (Table
4). The same finding was obtained for the relationship between mesothelioma
incidence and the average intensity of exposure (Table 5). Although the differences
between all assumptions were small, the β-coefficient showed the largest value of
0.2723 and 0.2367 with this revised AsbJEM for both cumulative asbestos exposure
and the average intensity of asbestos exposure, respectively (Table 4 and Table 5). The
risk of mesothelioma was estimated as HR of 1.31 (95%CI: 1.12-1.53) for every 1-log
fibre-year/ml increase of cumulative asbestos exposure. Mesothelioma risk was also
estimated as HRs of 1.66, 3.00 and 5.71 for cumulative asbestos exposures of 0.93,
2.23 and 12.3 fibre-years/ml, respectively, compared with the reference exposure of
0.09 fibre-years/ml (Table 6). The risk of mesothelioma was also estimated as HR of
1.27 (95%CI: 1.08-1.49) for every 1-log fibre/ml increase of the average intensity of
asbestos exposure. Mesothelioma risk was estimated as HRs of 1.84, 2.31 and 4.40 for
the average intensity of asbestos exposure of 0.03, 0.06 and 0.42 fibres/ml, respectively,
compared with the reference exposure of 0.003 fibres/ml (Table 6).
Table 6 Risk of mesothelioma associated with cumulative asbestos exposure and average
intensity of asbestos exposure estimated from the multivariate Cox proportional hazards model
Cumulative exposure (fibre-years/ml)a Hazard ratio (95% confidence interval)
54
Continuous 1.31 (1.12-1.53) (per log fibre-years/ml)
Categorized
0.09 1 (reference)
0.93 1.66 (0.69-3.99)
2.23 3.00 (1.22-7.38 )
12.3 5.71 (2.41-13.55)
Median of average intensity of asbestos
exposure (fibres/ml)b
Hazard ratio (95% confidence interval)
Continuous 1.27 (1.08-1.49) (per log fibres/ml)
Categorized
0.003 1 (reference)
0.03 1.84 (0.75-4.53)
0.06 2.31 (0.96-5.60 )
0.42 4.40 (1.78-10.91)
a, Cumulative asbestos exposure was calculated using the AsbJEM with a modification to peak
frequency (peak duration of 15 minutes, mode intensity at the 50th percentile of the range and
doubled peak frequency during the first two periods); b, Average intensity of asbestos exposure
was calculated using the AsbJEM with a modification to peak frequency (peak duration of 15
minutes, mode intensity at the 50th percentile of the range and doubled peak frequency during
the first two periods);
2.4.5 Validity of statistical assumptions
The proportional hazards assumption was confirmed to hold for all explanatory
variables by checking the Schoenfeld residuals, which were randomly scattered around
zero and the time-dependent covariate method, which showed non-significant results.
55
2.5 Discussion
The AsbJEM provided estimates of occupational asbestos exposure, enabling the
detection of an exposure-response relationship with mesothelioma incidence in
Western Australian workers. This was consistent with previously published literature
(Hansen et al., 1998; Iwatsubo et al., 1998; Gasparrini et al., 2008; Clin B et al., 2011;
Lacourt et al., 2014; Offermans et al., 2014; Ferrante et al., 2016). As the association of
mesothelioma incidence with asbestos exposure has been established, the significant
result of this study using the AsbJEM indicates that the tool serves to provide realistic
estimations for occupational asbestos exposure in a wide range of jobs throughout
Australia as it incorporates all available exposure data relevant to Australian conditions.
The significant result of exposure-response relationship between malignant
mesothelioma and asbestos exposure was demonstrated in two different models
including cumulative asbestos exposure and the average intensity of asbestos exposure,
respectively. This finding reinforces the effectiveness of the AsbJEM because the latter
exposure index is only calculated based on the AsbJEM while the former exposure
index also relies on information obtained irrespective of the tool, i.e. duration of
occupation/exposure. We explored various assumptions of the AsbJEM and found that
although differences between models were small, the published formula could be better
refined with some modification to the frequency of peak intensity since the strongest
dose-response relationship was demonstrated when it was doubled in the first two
periods of time in the AsbJEM.
Mesothelioma incidence was 45 per 100,000 person-years in our study. This figure is
lower than 135-203 per 100,000 patient-years among former asbestos miners and
56
millers in Western Australia (Berry et al., 2012) while it is higher than 5.8 per 100,000
patient-years from a population-based study in Italy (Mensi et al., 2016). Previous
studies of railway rolling stock workers in Italy and shipyard workers in Sweden
reported mesothelioma incidence as 22 and 54 per 100,000 patient-years, respectively
(Gasparrini et al., 2008; Sanden et al., 1992), which were comparable with the rate
observed in this study. These findings indicate that the ARP cohort is representative of
people at the risk of developing mesothelioma from occupational asbestos exposures
and the AsbJEM can adequately estimate the level of such exposures for this group.
Mesothelioma was also observed in workers with relatively low-dose occupational
asbestos exposure, again consistent with previous studies. The lowest quartile of
cumulative asbestos exposure for mesothelioma cases in this study was 0.54
fibre-years/ml (data only shown as the median and interquartile range in Table 1).
Offermans et al. (2014) described HR of 2.69 for cumulative asbestos exposure of 0.20
fibre-years/ml compared with never-exposed subjects. Lacourt et al. (2014)
demonstrated that when cumulative asbestos exposure ranged from 0 to 0.1
fibre-years/ml, the risk of mesothelioma became 4 times higher than never-exposed
subjects. They also stated that 15% of men with mesothelioma had cumulative asbestos
exposure below 0.1 fibre-years/ml. In addition, Ferrante et al. (2016) reported that the
odds ratio of mesothelioma rose to 4.4 when cumulative asbestos exposure increased
from the background level below 0.1 to a range of 0.1 to 1 fibre-years/ml. Furthermore,
our study confirmed four cases of mesothelioma at an even lower cumulative asbestos
exposure ranging from 0.001 to 0.1 fibre-years/ml. The occurrence of mesothelioma
with such a low dose of exposure is not anecdotal as Iwatsubo et al. (1998) reported
the risk of mesothelioma as OR of 1.2 with cumulative exposure from 0.001 to 0.49
57
fibre-years/ml. Therefore, the AsbJEM will be helpful to estimate the risk of
mesothelioma for workers with a low dose of occupational asbestos exposure. There
were some differences between our study and others. Most previous studies were
population-based and thus subjects never-exposed to asbestos were included in the
analysis (Ferrante et al., 2016; Lacourt et al., 2014; Offermans et al., 2014). This was
also true for a hospital-based study (Iwatsubo et al., 1998). However, in our study we
used a cohort with known occupational asbestos exposure and participants with only
background level of asbestos exposure were not included. If never-exposed subjects
had been included in our analyses, the results would have been overestimated because
those people were theoretically at no risk of developing mesothelioma. This would
then be analogous to a previous study where a significant exposure-response
relationship was obtained only when never-exposed subjects were included in the
analysis (Offermans et al., 2014).
There are some caveats to interpret the findings of this study. First, we aimed to
validate the AsbJEM based on the assumption that the strongest dose-response
relationship, identified through the steepest slope, for the well-established association
between asbestos and mesothelioma incidence would indicate the best estimation of
asbestos exposure. Some may argue that this is not necessarily correct and the gold
standard of counting the number of asbestos fibres inside the lung is necessary to
confirm the exposure level (Tuomi et al., 1991). However, it is unrealistic and
unethical to apply such an invasive procedure to the whole population included in this
study and adopting a substitute analytical method is more practical to estimate the level
of exposure (Hardt et al., 2014). In fact, the above-mentioned assumption seems
58
reasonable as dose-response relationships are expected to improve if the
misclassification of the estimation can be reduced (Lenters et al., 2011). Second, some
participants were excluded from the analysis due to an unclear occupational history
regardless of potential asbestos exposure. This may have caused under- or
overestimation of the relationship between the incidence of mesothelioma and asbestos
exposure. However, as the number of these patients was relatively small compared
with the entire sample size (5%), the influence of this issue was likely to be trivial.
Furthermore, there was some uncertainty regarding asbestos exposure before 1943. We
assumed that the exposure level before 1943 was the same as the subsequent period
from 1943 to 1966 and applied the same level of exposure. This decision was
supported by an occupational hygienist from the expert panel that was involved in the
development of the AsbJEM (personal communication with L.G.) although there were
uncertainties for some industries. However, this extrapolation was only necessary for a
small number of jobs (n=88) and thus would not have affected the findings.
2.6 Conclusion
The recently published AsbJEM was validated using a cohort with a diverse
occupational asbestos exposure. The original published calculation can be better
refined with some modification to the frequency of peak intensity as it demonstrated
the strongest dose-response relationship and therefore was considered the best estimate
of occupational asbestos exposure. We expect the results of this study to facilitate more
widespread use of the AsbJEM in future research including risk assessment of other
ARDs.
59
Chapter Three: Project 2
Relationship between asbestos exposure and asbestosis and pleural plaques in
Australian workers
60
3.1 Abstract
Introduction
Dose-response relationships between asbestos exposure and non-malignant
asbestos-related diseases (ARDs) have been well described. However, there has been
substantial variation in the estimates of the relationship, which may, in part, be due to
misclassification of occupational asbestos exposure. An Asbestos Job Exposure Matrix
(AsbJEM) has been validated as a tool to estimate occupational asbestos exposure for
Australian workers. The aim of this study was to investigate dose-response
relationships between asbestos and both asbestosis and pleural plaques using the
AsbJEM.
Methods
Subjects were participants in an asbestos health surveillance program who had at least
3 months’ occupational exposure to asbestos. The level of asbestos exposure for a
participant was estimated using the AsbJEM. Asbestosis and pleural plaques were
identified from plain chest x-ray films. Asbestosis was diagnosed as the International
Labour Office (ILO) profusion classification of 1/0 or higher while pleural plaques
were defined as localized pleural thickening. The association of asbestosis and pleural
plaques with cumulative asbestos exposure and the average intensity of asbestos
exposure was analysed using multivariate logistic regression models.
Results
Of the 1602 men included in the analysis, there were 477 cases of asbestosis (30.0%)
and 901 cases of pleural plaques (66.2%). After adjusting for the age at first asbestos
61
exposure and smoking history, there were significant associations of asbestosis with
both cumulative asbestos exposure and the average intensity of asbestos exposure
(odds ratio (OR) (95% confidence interval (CI)): 1.06 (1.01-1.10) and 1.06 (1.02-1.11),
respectively). There was a monotonic increase in risk across exposure quartiles with
ORs of 1.27, 1.35 and 1.58 for cumulative asbestos exposures of 0.41, 1.11 and 14.9
fibre-years/ml, respectively compared with the reference exposure of 0.002
fibre-years/ml. The risk of asbestosis also steadily increased to ORs of 1.29, 1.38 and
1.50 as the average intensity of asbestos exposure increased from the reference
exposure of 0.0002 to 0.01, 0.04 and 0.17 fibres/ml, respectively. Results for pleural
plaques were similar and there was a significant association of pleural plaques with
cumulative asbestos exposure (OR (95%CI): 1.05 (1.00-1.11)). As cumulative asbestos
exposures elevated to 0.30, 1.00 and 16.4 fibre-years/ml from the reference exposure
of 0.004 fibre-years/ml, the risk of pleural plaques increased to ORs of 1.25, 1.33 and
1.54, respectively. However, there was no significant association of pleural plaques
with the average intensity of asbestos exposure (OR (95%CI): 1.04 (0.99-1.09)).
Conclusion
The risk of asbestosis and pleural plaques was closely related to cumulative asbestos
exposure. These diseases developed at a low level of asbestos exposure.
62
3.2 Introduction
Asbestos can cause non-malignant diseases such as asbestosis and pleural plaques
(PPs) (Hillerdal, 1994). Asbestosis is a fibrotic change of pulmonary parenchyma
mixed with inflammation while PPs are focal thickness of thoracic pleura, which often
involves calcification (American Thoracic Society, 2004). Although these
non-malignant asbestos related diseases (ARDs) are not usually an immediate threat to
life (American Thoracic Society, 2004), which is different from malignant
mesothelioma, asbestosis can damage pulmonary function and respiratory failure may
ensue whereas PPs have negligible effect on lung function (Algranti et al., 2001;
Markowitz et al., 1997; Berry et al., 1979). Positive dose-response relationships
between asbestos and both asbestosis and PPs have consistently been reported
(Jakobsson et al., 1995; Järvholm, 1992; Boffetta, 1998). However, there has been
substantial variation in the estimates of dose-response relationships for both of these
conditions (Matrat et al., 2004; Huang, 1990; Courtice et al., 2016; Paris et al., 2009;
Ehrlich et al., 1992). Some of this variation may be due to misclassification of
occupational asbestos exposure (Lenters et al., 2011) and other possible reasons
include different types of fibres and follow-up periods of time (Berman and Crump,
2008). To better understand these relationships, particularly at the lower end of
exposure where data is sparce, robust exposure measures are required (Fletcher et al.,
1993; van der Bij et al., 2013). We have recently published an Asbestos Job Exposure
Matrix (AsbJEM) for Australian workers (van Oyen et al., 2015) and the reliability of
the AsbJEM has been demonstrated using malignant mesothelioma (refer to the second
chapter of this thesis). It has also been demonstrated that the tool can accurately
63
estimate lower levels of occupational asbestos exposure (refer to the second chapter of
this thesis). Therefore, the aim of this study was to estimate the risk of both asbestosis
and PPs using the AsbJEM for a cohort of Australian workers with occupational
asbestos exposure.
3.3 Methods
3.3.1 Population
Eligible subjects in this study were participants of an annual health surveillance
program for people with significant occupational or environmental exposure to
asbestos (the Asbestos Review Program (ARP)) (Hansen et al., 1997; Hansen et al.,
1998; Musk et al., 1998). People recruited to join the ARP included former workers of
the crocidolite mine and mill at Wittenoom, Western Australia, ex-residents of the
township that serviced the mine and people who had had at least 3-months’ working
history with possible occupational asbestos exposure, and people who had evidence of
PPs. Only asbestos-exposed workers, who were not former workers or residents of
Wittenoom, were included in this study.
3.3.2 Asbestos exposure estimation
The level of asbestos exposure for an individual person in the cohort was estimated
using the AsbJEM based on their work histories (van Oyen et al., 2015). Exposure
estimates for occupation-industry combinations over four time periods (1943-1966;
1967-1986; 1987-2003; ≥2004) were developed by an expert panel. ‘Mode’ was the
most common exposure level and ‘peak’ was the short-term intense exposure in a
particular job. ‘Background’ was the same as the exposure from the general
64
environment. The frequency and intensity of mode, peak and background exposures
were pre-defined for each time period aside from the frequency of background
exposure, which was calculated by subtracting the mode frequency from the total
number of working days in the working year (240 days assuming two days off a week,
four weeks of holidays and Christmas). The background, peak and mode exposures
were each calculated as the multiplicative product of intensity and frequency and an
annual average exposure was calculated as the sum of all these exposures.
Accordingly, a cumulative asbestos exposure for an individual person in a specific
occupation was calculated as the annual average exposure multiplied by the length of
employment (van Oyen et al., 2015). As the previous research confirmed that the
published AsbJEM could be better improved if the frequency of peak exposure was
doubled for the first two periods of time, the AsbJEM with this modification was
implemented in this analysis.
3.3.3 Case ascertainment
All participants in the ARP had annual chest screening using plain chest x-ray (CXR)
film before 2013 and it was subsequently replaced by low dose computed tomography
(LDCT) scan. In this study, CT was utilized for 241 people who participated in the
program after this year and only the result of asbestosis was available for these people.
Asbestosis and PPs were ascertained by experienced readers. Each CXR was read by at
least two and mostly three readers. Blinding for radiological evaluation was not
attempted because the readers were aware that participants had occupational asbestos
exposure. Asbestosis was diagnosed using the International Labour Office (ILO)
profusion classification 1/0 or higher on CXR or reticular opacities, interlobular septal
65
thickening, traction bronchiectasis or honeycombing on LDCT (ILO, 2003; ILO, 2011;
Akira and Morinaga, 2016). PPs were defined as localized pleural thickening
regardless of the site and extent and the presence of calcification (ILO, 2003; ILO,
2011). For the analyses both asbestosis and PPs were recorded as binary variables
(present/absent). The date of the diagnosis of each disease was set at the time when
corresponding radiological findings were identified for the first time during the
follow-up periods of time or at the first visit if they were present at that time. For the
latter case, the profusion level was higher than 1/0 for some cases if they attended the
program at an advanced stage. Non-cases were those without any corresponding
radiological findings during the follow-up periods of time and ascertained at the last
visit.
3.3.4 Statistical analysis
The association of asbestosis and PPs with asbestos exposure was estimated by two
differet models using the multivariate logistic regression analysis. As an index of
asbestos exposure, cumulative asbestos exposure was included in the first model while
the second model included the average intensity of asbestos exposure alongside
duration of employment. Cumulative asbestos exposure was calculated for an
individual person using the AsbJEM and the average intensity of asbestos exposure
was calculated as the division of cumulative asbestos exposure by duration of
employment. Asbestosis and PPs were analyzed separately although some people had
both conditions (n=326).
The first occupational asbestos exposure was defined as the exposure beyond the
background level and the time since first occupational asbestos exposure was
66
calculated as the period up to the ascertainment of the disease for cases and the last
visit for non-cases.
As the distribution of cumulative asbestos exposure and the average intensity of
asbestos exposure was skewed, logarithmic transformation was performed and used for
the analyses.
Other covariates included in the model were the age at first occupational asbestos
exposure and the time since first occupational asbestos exposure (Paris et al., 2008).
Age was excluded from the model due to concerns of collinearity with the time since
first asbestos exposure. Smoking history was only included for the analysis of
asbestosis as it has no relationship with PPs (Hillerdal, 1994). As all these variables
were likely to influence the prevalence of the diseases, they were included in the
multivariate analysis regardless of statistical significance in the univariate analysis.
The presence of the effect modification of cumulative asbestos exposure and the
average intensity of asbestos exposure by other potential confounders was evaluated
using interaction terms. A variable with the least significant p-value over 0.01 was
eliminated one by one from the model until it reached the least variables.
The fit of the model was assessed by Hosmer and Lemeshow goodness of fit test.
P-value ≤0.05 was considered as the model not fitting the data well.
Through all the process of building logistic regression models, fractional polynomials
were used to fit the best numerical form of continuous variables into the models.
Continuous data was expressed as either mean (standard deviation (SD)) or median
(interquartile range (IQR)). Student’s t-test or Wilcoxon rank sum test was performed
for a comparison of the mean or median of two independent groups, respectively. For
67
categorical data, chi-square test was conducted. Statistical significance was set at
p-value ≤0.05. All statistical analyses were conducted using SAS 9.4 (SAS Institute
Inc., Cary, NC, USA) or Stata 14 (STATA Corp LLC., College Station, TX, USA).
This study was exempted from ethics review by the University of Western Australia
Human Research Ethics Committee (RA/4/20/4746).
3.4 Results
3.4.1 Characteristics of subjects
A total of 2084 people, who were not former workers or ex-residents of Wittenoom,
attended the ARP. Of these, 1602 men were included in the analyses after excluding
those with additional non-occupational asbestos exposure (n=373), participants with an
uncertain occupational history (n=96) and women (n=13). There were 477 cases of
asbestosis (30.0%) and 901 cases of PPs (66.2%) (Table 7 and Table 8). The profusion
level for asbestosis included 1/0 in 450, 1/1 in 8, 1/2 in 6, 2/1 in 7, 2/3 in 3 and 3/2 in 3.
Compared with participants who did not develop asbestosis or PPs, the mean age at
first exposure was significantly higher for asbestosis cases (17.5±5.5 vs. 16.7±4.2,
p=0.0006) while it was lower for PP cases (16.6±4.5 vs. 17.9±5.4, p<0.0001). The
calculated median cumulative asbestos exposure was significantly higher for both
asbestosis and PP cases compared with non-cases (0.64 vs. 0.45 fibre-years/ml,
p=0.0013 and 0.63 vs. 0.44 fibre-years/ml, p=0.0002, respectively) (Table 7 and Table
8). The median of the calculated average intensity of asbestos exposure was also
significantly higher for asbestosis cases compared with non-cases (0.027 vs. 0.018
fibres/ml, p=0.0003) (Table 7) while it was not significantly different between PP cases
and non-cases (0.023 vs. 0.026 fibres/ml, p=0.14) (Table 8).
68
Table 7 Comparison of participants in a cohort with and without asbestosis
Asbestosis
(n=477)
No asbestosis
(n=1125)
p-value
Age (years) 66.9±8.7 67.5±11.2 0.20
Duration of exposure (years) 31.6±14.0 32.3±13.9 0.43
Age at first exposure (years) 17.5±5.5 16.7±4.2 0.004
Time since first exposure (years) 49.4±10.5 50.9±12.2 0.02
Cumulative exposure (fibre-years/ml) 0.64 (1.67) 0.45 (1.17) 0.001
Average intensity of exposure (fibres/ml) 0.027 (0.061) 0.018 (0.043) 0.0003
Presence of smoking history (current or former) 353 (74.0) 834 (74.2) 0.94
Continuous data expressed as mean±SD or median (IQR) while categorical data expressed as
number (%); Student’s t-test and Wilcoxon rank sum test was performed for a comparison of
the mean and median between the two groups, respectively. For categorical data, chi-square
test was conducted.
Table 8 Comparison of participants in a cohort with and without pleural plaquea
Plural plaques
(n=901)
No pleural
plaques (n=460)
p-value
Aage (years) 70.5±8.5 62.7±12.5 <0.0001
Duration of exposure (years) 34.0±13.4 27.3±13.9 <0.0001
Age at first exposure (years) 16.6±4.5 17.9±5.4 <0.0001
Time since first exposure (years) 53.9±9.5 44.7±13.5 <0.0001
Cumulative exposure (fibre-years/ml) 0.63 (1.52) 0.44 (1.20) 0.0002
Average intensity of exposure (fibres/ml) 0.023 (0.051) 0.026 (0.050) 0.14
Presence of smoking history (current or former) 674 (74.8) 346 (75.4) 0.82
Continuous data expressed as mean±SD or median (IQR) while categorical data expressed as
number (%); Student’s t-test and Wilcoxon rank sum test was performed for a comparison of
69
the mean and median between the two groups, respectively. For categorical data, chi-square
test was conducted. a, a total number of participants for this analysis was 1361 out of 1602
participants due to 241 missing data for the outcome
3.4.2 Risk estimates of asbestosis
After adjusting for the age at first asbestos exposure and smoking history, there was a
significant association of asbestosis with calculated cumulative asbestos exposure (OR
(95%CI): 1.06 (1.01-1.10)). The risk of asbestosis was estimated as the ORs of 1.27,
1.35 and 1.58 for calculated cumulative asbestos exposures of 0.41, 1.11 and 14.9
fibre-years/ml, respectively compared with the reference exposure of 0.002
fibre-years/ml (Table 9).
There was also a significant association of asbestosis with the average estimated
intensity of asbestos exposure (OR (95%CI): 1.06 (1.02-1.11)). The risk of asbestosis
was estimated as the ORs of 1.29, 1.38 and 1.50 for the average estimated intensity of
asbestos exposure of 0.01, 0.04 and 0.17 fibres/ml, respectively compared with the
reference exposure of 0.0002 fibres/ml (Table 9).
Table 9 Risk of asbestosis estimated from the multivariate logistic regression model
Cumulative exposure (fibre-years/ml) Odds ratio (95% confidence interval)
Continuous 1.06 (1.01-1.10) (/Iloga+1.31 fibre-years/ml)
Categorized
0.002 1 (reference)
0.41 1.27 (1.05-1.52)
1.11 1.35 (1.06-1.72 )
70
14.9 1.58 (1.10-2.26)
Average intensity of exposure (fibres/ml) Odds ratio (95% confidence interval)
Continuous 1.06 (1.01-1.10) (/Ilogb+4.47 fibres/ml)
Categorized
0.0002 1 (reference)
0.01 1.29 (1.07-1.55)
0.04 1.38 (1.09-1.74 )
0.17 1.50 (1.12-2.02)
Time since first exposure (years) Odds ratio (95% confidence interval)
10 1 (reference)
20 2.05 (1.39-3.03)
30 9.04 (12.92-27.94)
40 11.68 (3.15-43.30)
50 11.02 (12.84-42.73)
The multivariate logistic regression model included the age at first asbestos exposure and
smoking history as covariates. a, logarithmic form of cumulative exposure; b, logarithmic form
of average intensity of exposure;
The time since first asbestos exposure was also significantly related to asbestosis with
the incremental risk over the follow-up periods of time. The ORs elevated to 2.05, 9.04
and 11.7 for 20, 30 and 40 years, respectively with the reference of 10 years (Figure 1).
71
Figure 1 Risk of asbestosis by cumulative asbestos exposure and the time since first asbestos
exposure
After adjusting for the age at first exposure and smoking history by multivariate logistic
regression model, there was a trend that the risk of asbestosis elevated constantly with an
increase of cumulative asbestos exposure from 0.002 to 14.9 fibre-years/ml. There was also a
consistent increase of the risk of asbestosis with the progress of time since first asbestos
exposure from 10 to 40 years. The risk was estimated with cumulative exposure of 0.002
fibre-years/ml and the time since first exposure of 10 years as the reference.
3.4.3 Risk estimates of pleural plaque
After adjusting for the age at first asbestos exposure, there was a significant association
of PPs with calculated cumulative asbestos exposure (OR (95%CI): 1.05 (1.00-1.11)).
The risk of PPs was estimated as the ORs of 1.25, 1.33 and 1.54 for calculated
05
10
15
20
Odd
s R
atio
10 20 30 40 50Time since the first exposure (years)
0.002 fibres-years/ml 0.41 fibres-years/ml
1.11 fibres-years/ml 14.9 fibres-years/ml
72
cumulative asbestos exposures of 0.30, 1.00 and 16.4 fibre-years/ml, respectively
compared with the reference exposure of 0.004 fibre-years/ml (Table 10).
However, there was no significant association of PPs with the average calculated
intensity of asbestos exposure (OR (95%CI): 1.04 (0.99-1.09)). The risk of PPs was
estimated as the ORs of 1.18, 1.23 and 1.29 for the average calculated intensity of
asbestos exposure of 0.01, 0.04 and 0.17 fibres/ml, respectively compared with the
reference exposure of 0.0002 fibres/ml (Table 10).
Table 10 Risk of pleural plaques estimated from the multivariate logistic regression model
Cumulative exposure (fibre-years/ml) Odds ratio (95% confidence interval)
Continuous 1.05 (1.00-1.11) (/Iloga+1.15 fibre-years/ml)
Categorized
0.004 1 (reference)
0.30 1.25 (1.01-1.54)
1.00 1.33 (1.02-1.74 )
16.4 1.54 (1.03-2.30)
Average intensity of exposure (fibres/ml) Odds ratio (95% confidence interval)
Continuous 1.06 (1.01-1.10) (/Ilogb+4.47 fibres/ml)
Categorized
0.0002 1 (reference)
0.01 1.18 (0.96-1.46)
0.04 1.23 (0.95-1.59 )
0.17 1.29 (0.94-1.77)
73
Time since first exposure (years) Odds ratio (95% confidence interval)
20 1 (reference)
30 5.68 (4.24-7.61)
40 16.10 (10.08-25.72)
50 32.89 (18.25-59.26)
60 55.79 (28.32-109.90)
The multivariate logistic regression model included the age at first asbestos exposure. a,
logarithmic form of cumulative exposure; b, logarithmic form of average intensity of exposure
The time since first asbestos exposure was also significantly related to PPs with the
incremental risk over the follow-up periods of time. The ORs elevated to 5.68, 16.1,
32.9 and 55.8 for 30, 40, 50 and 60 years, respectively with the reference of 20 years
(Figure 2).
02
04
06
08
0
Odd
s R
atio
20 30 40 50 60Time since the first exposure (years)
0.004 fibres-years/ml 0.30 fibres-years/ml
1.00 fibres-years/ml 16.4 fibres-years/ml
74
Figure 2 Risk of pleural plaques by cumulative asbestos exposure and the time since first
asbestos exposure
After adjusting for the age at first exposure by multivariate logistic regression model, there was
a trend that the risk of PPs elevated constantly with an increase of cumulative asbestos
exposure from 0.004 to 16.4 fibre-years/ml. There was also a consistent increase of the risk of
PPs with the progress of time since first asbestos exposure from 20 to 60 years. The risk was
estimated with cumulative exposure of 0.004 fibre-years/ml and the time since first exposure
of 20 years as the reference.
3.4.4 The fit of the model
As Hosmer and Lemeshow goodness of fit test demonstrated p-value >0.05 in all
models for both asbestosis and PPs, they were considered as fitting the data well.
3.5 Discussion
This study demonstrated that the risk of asbestosis and PPs was significantly associated
with calculated cumulative asbestos exposure and the time since first asbestos
exposure for asbestos-exposed workers after adjusting for the effect of other potential
confounders such as the age at first asbestos exposure. The average calculated intensity
of asbestos exposure was also significantly associated with asbestosis although it was
not significant for PPs. These findings are compatible with previous reports that
cumulative exposure is most closely associated with the incidence of asbestosis
(Boffetta, 1998) although short-time exposure to high level of asbestos could also
cause the disease as Lacourt et al (2017) described the importance of time dose
relationship over the average concentration. On the other hand, PPs were reported to be
observed for low levels of calculated cumulative asbestos exposure although it would
take a longer latency period. This may be explained by the assumption that asbestos
75
fibres would reach the pleura either directly passing through the lung or transported by
the lymphatic system, both of which would take some time in addition to the
subsequent long-standing fibrotic process (Järvholm, 1992). The importance of time
dose relationship over the simple average concentration may also be related to the
finding that the average calculated intensity of asbestos exposure was not significant
for PPs as it was calculated as the product of cumulative exposure divided by duration
of exposure (Lacourt et al., 2017).
A number of articles described the dose-response relationship between occupational
asbestos exposure and non-malignant ARDs and there was substantial variation in their
estimates. Berry et al (1979) and Huang (1990) reported that the prevalence of
asbestosis for British workers in a textile factory and Chinese workers in chrysotile
product factory was 1% with estimated cumulative asbestos exposure of 45 and 22
fibre-years/ml, respectively. The finding in the latter report was commensurate with
asbestos concentration of 0.63 fibres/ml while in the former report asbestosis occurred
with the concentration between 0.3 and 1.1 fibres/ml. According to the study by Matrat
et al (2004), the majority of French workers in buildings with asbestos-containing
materials demonstrated the profusion classification of 1/0 or higher with cumulative
asbestos exposure of less than 1.5 fibre-years/ml while out of 772 male Italian asbestos
workers, 2% developed asbestosis with the mean cumulative exposure of 370
fibre-years/ml (Mastrangelo et al., 2009). Courtice et al (2016) reported the median
cumulative asbestos exposure of 132 fibre-years/ml for workers in an asbestos product
factory and 22% developed asbestosis. Such a variation of previous findings is also
true for PPs. Paris et al (2009) demonstrated that over 80% of PPs developed with
cumulative asbestos exposure of over 30 fibre-years/ml and another study by the same
76
research group reported the mean of cumulative asbestos exposure as 135
fibre-years/ml for workers with the disease (Paris et al., 2004). Whereas Matrat et al
(2004) reported that 30% of pleural thickening was identified with cumulative asbestos
exposure of less than 0.25 fibre-years/ml.
As compared with these prior reports, calculated cumulative asbestos exposure in our
study was even lower with the median value of 0.64 and 0.63 fibre-years/ml for
workers with asbestosis and PPs, respectively. Similarly, the average intensity of
asbestos exposure was as low as 0.027 and 0.023 fibres/ml, respectively. It was also
discovered that the prevalence of the diseases was 30% for asbestosis and 65% for PPs,
which were higher than previous studies (Koskinen et al., 1998; Stayner et al., 1997;
Algranti et al., 2001) regardless of such a low level of occupational asbestos exposure.
There are several factors that may have caused the diversity of previous reports and
differences between these studies and our research. First, misclassification of the level
of occupational asbestos exposure is a major cause of varied results (Lenters et al.,
2011), in particular, when it is estimated based on an individual worker’s self-report.
As the AsbJEM can improve the estimates of asbestos exposure, in particular, at a
lower level of exposure, the dose-response relationship between occupational asbestos
exposure and non-malignant ARDs demonstrated in this study may be indicating more
reliable estimates. However, the same trend needs to be ascertained in another cohort
of Australian people using the AsbJEM to further investigate the performance of this
tool for identifying dose-response relationships between occupational asbestos
exposure and non-malignant ARDs. In addition, the results in this study are not
necessarily comparable to previous studies that were conducted in other countries
where working conditions related to occupational asbestos exposure will be different
77
from the setting in this study and the results based on a JEM are often not generalizable
to other situations (Lavoue et al., 2012). Second, the difference of fibre types used may
be related to diverse results. Participants in this study were former workers in
asbestos-related industry in Western Australia where crocidolite was mined from 1943
until 1966 and many asbestos products used were mixed types containing crocidolite
(van Oyen et al., 2015). In addition, the percentage of crocidolite in the products may
have been higher in Western Australia than in other areas in Australia. As crocidolite is
more potent than chrysotile for mesothelioma (Berman and Crump, 2008), the use of
products composed of mixed asbestos fibres may explain higher prevalence of ARDs
for a lower level of asbestos exposure in this study. Third, there is also an issue of
sensitivity and specificity for the diagnosis of the disease by conventional x-ray films
(Algranti et al., 2001). It is often difficult to detect early changes of asbestosis and PPs
on plain films and misdiagnoses are not uncommon due to artefacts inherently
accompanied by the modality. Variability between readers can also be involved
(Hopstaken et al., 2004). Some research groups which have implemented CT scan to
screen and more precisely evaluate pulmonary changes have stated that over 85% of
asbestosis developed with cumulative asbestos exposure of over 30 fibre-years/ml
(Paris et al., 2009) and only 2% of the persons with that of less than 25 fibre-years/ml
had CT asbestosis (Paris et al., 2004). However, CT is likely to reveal irrelevant and
incidental pulmonary changes in addition to an early phase of asbestosis and PPs
(Vierikko et al., 2007). This potential overdiagnosis may lead to underestimation of
exposure required for the development of the disease, particularly at lower level
exposure. Furthermore, smoking is noted to cause a mixture of fibrosis and
inflammation in pulmonary parenchyma and develop interstitial lung disease (Kumar
78
et al., 2017), which will be manifested as similar radiological changes as early
asbestosis (Copley et al., 2003). Accordingly, it is possible that smoking history might
have confounded the diagnosis of asbestosis since smoking history was prevalent in
this study although theoretically it cannot cause the disease as they are
pathophysiologically different (Agusti et al., 1988). Finally, although the AsbJEM with
modification was implemented in this study based on the result of the first project, it is
possible that the tool is still underestimating occupational asbestos exposure for the
subjects in the study because they joined the health surveillance program due to their
awareness of asbestos exposure.
Regardless of these limitations that might have caused over- or underdiagnosis of the
disease, this study remains meaningful as it has demonstrated a relationship between
occupational asbestos exposure and non-malignant ARDs using the AsbJEM, which
has proved to be a valid and reliable tool to estimate the level of occupational asbestos
exposure, in particular, at a lower level. Further research needs to be conducted in
another cohort of Australian people to confirm the result of this study and the risk
assessment for ARDs would be improved with implementation of CT in addition to the
introduction of the AsbJEM.
3.6 Conclusion
This study estimated the risk of asbestosis and PPs for ordinary Australian workers
with occupational asbestos exposure using the AsbJEM. Both calculated cumulative
asbestos exposure and the time since first asbestos exposure were significantly related
to the diseases. Both diseases were noted to develop after a long latency since the
exposure to a low level of asbestos.
79
Chapter Four: Conclusion
80
Asbestos can cause serious health problems and the risk of ARDs has been
investigated by a myriad of research groups around the world (Gasparrini et al., 2008;
Ferrante et al., 2016; Clin et al., 2011; Lacourt et al., 2014). In Australia, previous
reports of assessing risk of ARDs were mostly based on miners and millers (de Klerk
et al., 1989) while population-based data regarding occupational asbestos exposure has
been lacking. As dose-response relationship between asbestos exposure and the
occurrence of ARDs for this high exposure industry may be different from other
ordinary jobs (Churg and Wiggs, 1986), it is difficult and possibly inappropriate to
extrapolate the result of mining workers to estimate the risk of ARDs for general
population. A job exposure matrix (JEM) is a combination of job titles and exposure
estimates, which enables systematic calculations of occupational asbestos exposure
(Pannett et al., 1985; Hardt et al., 2014; Offermans et al., 2014; Peters et al., 2016).
The Asbestos Job Exposure Matrix (AsbJEM) has been developed to estimate
dose-response relationship between occupational asbestos exposure and ARDs for
ordinary Australian workers (van Oyen et al., 2015).
The objective of this thesis was to assess the performance of the AsbJEM, as well as
refine it if required. The thesis was composed of two studies. In the first study, the
AsbJEM was demonstrated to be a reliable and useful tool to systematically and
cost-effectively estimate occupational asbestos exposure. This project found that the
AsbJEM could be refined with some modification, specifically increasing the
frequency of peak exposures. The validity of the tool was demonstrated by a
significant dose-response relationship between asbestos exposure and mesothelioma
incidence. This study also found that some cases developed at very low dose exposure
of less than 0.01fibres/ml. In the second study, we demonstrated a significant
81
dose-response relationship between occupational asbestos exposure and benign ARDs
such as asbestosis and pleural plaques by applying the modified AsbJEM based on the
results of the first project. These findings suggest that the AsbJEM could be utilized as
a reliable instrument to calculate asbestos exposure in various types of job and
industries and improve the estimates of dose-response relationship for Australian
workers.
Epidemiological studies of asbestos-related health problems in Australia have mostly
been conducted for mesothelioma cases (Leigh and Driscoll, 2003). This was
facilitated by a national health surveillance program called Australian Mesothelioma
Surveillance Program and Australian Mesothelioma Registry (Yeung et al., 1999).
Otherwise, it was based on asbestos mining and milling where detailed data for
exposure was available (de Klerk et al., 1989). Leigh and Driscoll (2003) investigated
mesothelioma risks in different occupational groups and reported that the most
common occupational exposures were building industry followed by shipbuilding,
asbestos cement production, railways and Wittenoom crocidolite mining and milling.
Yeung et al (1999) also examined distribution of mesothelioma cases in various
occupations and industries and reported a similar trend. The latter research group also
pointed out that mesothelioma cases from the traditional asbestos industry such as
crocidolite mining and milling and asbestos cement production had been declining or
plateaued and more cases had been observed from occupations such as plumbers,
carpenters and machinists. A recent report from Musk et al (2015) revealed a shift of
major occupational groups for mesothelioma cases over the last five decades in
Western Australia. According to their report the percentage of Wittenoom workers for
82
mesothelioma cases declined from 54.4% in the years between 1960 and 1979 to
13.8% in the years between 2000 and 2009 while building and construction workers
increased from 3.5% to 29.4% in the same time span. Although these trends may be
currently changing over time and the risk at low dose exposure to in-situ asbestos have
recently been attracting much attention after the complete ban of using
asbestos-containing materials, it should still be kept in mind that a large number of
former workers exposed to occupational asbestos are still at the risk of developing the
lethal disease. However, most of the previous reports of the quantitative risk estimates
of asbestos exposure in Australia were based on miners and millers (de Klerk et al.,
1989) and residents of the township that served the industry (Hansen et al., 1998). On
the other hand, there has been scarce data for Australian workers in other occupations
and industries. It is conceivable that it may cause incorrect prediction if the findings of
miners and millers are extrapolated to non-mining and milling occupations and
industries as the intensity and duration of exposure are likely different and these index
of exposure are closely related to the development of ARDs (Gasparrini et al., 2008;
Ferrante et al., 2016; Clin et al., 2011; Lacourt et al., 2014).
Although a large number of studies investigated dose-response relationship between
occupational asbestos exposure and ARDs, results are diverse between studies
(Gasparrini et al., 2008; Ferrante et al., 2016; Clin et al., 2011; Lacourt et al., 2014).
This variability is attributed to the difference of working conditions and asbestos fibres
used, and varied tools of estimating asbestos exposure. de Klerk et al (1996) and Rees
et al (2001) reported asbestos fibre concentrations in lung tissues of crocidolite and
chrysotile mining workers in Australia and South Africa, respectively using the same
83
fibre counting measure called transmission electron microscope fitted with an energy
dispersive X-ray analysis system. The former study group found crocidolite fibres with
the mean concentration of 183 million fibres/g dry tissue while the mean chrysotile
fibre concentration was 0.69 million fibres/g dry tissue in the latter study. Although
both of these two studies investigated mining workers, lung burden of asbestos fibres
was remarkably different and this disparity can be mostly explained by the difference
of the property of fibres regarding turnover in human lungs. Chrysotile is reported to
be more rapidly cleared from the lungs while crocidolite is more likely to be retained
inside the lung (Albin et al., 1994). In addition to the difference of carcinogenicity
(Hodgson and Darnton, 2000), this disparity of physical characteristics of asbestos
fibres may explain the findings of this thesis that both malignant and non-malignant
ARDs developed at relatively low dose asbestos exposure with some cases at very low
level exposure (<0.01fibres/ml) and the prevalence of asbestosis and pleural plaques
was larger in the cohort of this thesis than in previous reports since most of the
asbestos products used in Australia were mixed types containing crocidolite (van Oyen
et al., 2015).
It is important to establish a reliable methodology to calculate occupational asbestos
exposure because it will help more accurately estimate the risk of ARDs (Lenters et al.,
2011). In this thesis the AsbJEM was demonstrated to be a reliable exposure
assessment tool to estimate dose-response relationship for ordinary Western Australian
workers. JEMs are reported to improve the drawbacks of self-reported exposure
estimates (Hardt et al., 2014) and reduce the cost and labour of expert evaluation on an
individual basis (Nam et al., 2005). In addition, the AsbJEM covers a wide range of
84
jobs and industries over a long period of time and could quantitatively estimate
asbestos exposure. This gives it an advantage over the only other published tool for
asbestos task exposure in Australia, asbestos task exposure matrix (ASTEM). The
ASTEM was constructed based on data in New South Wales and only describes 19
task-product combinations over two periods of time (Hyland et al., 2010). Therefore,
the AsbJEM would improve the risk estimates of ARDs and could be accepted for
widespread use across the country to clarify the occupational asbestos burden for the
entire population. As the AsbJEM covers the period after 2003 when the use of
asbestos was completely banned, it could estimate the risk at low dose exposure to
in-situ asbestos such as demolition work under current regulations (Park et al., 2013),
which will be different from previous occupational exposure patterns (Musk et al.,
2017) although the risk of non-occupational exposure to in-situ asbestos might need to
be assessed with a different approach aside from the AsbJEM.
The AsbJEM could also give an impact on clinical practice by improving the estimates
for the risk of ARDs. For example, accurate assessment of asbestosis and pleural
plaques could improve the estimates for the risk of lung cancer as the presence of the
former conditions could enhance the risk of the latter disease (Weiss, 1999). In contrast
to mesothelioma, lung cancer is a curable disease and can be controlled with currently
available pharmacological agents if implemented at an appropriate timeline
(Maemondo et al., 2010). In addition, a new therapeutic option has recently become
available to suppress the progression of parenchymal fibrosis in the lung (Noble et al.,
2011) and may possibly be effective for asbestosis. Therefore, the AsbJEM could make
85
contributions to better estimate the risk of ARDs and help improve clinical outcome for
individual patient with appropriate treatment.
Despite their benefits, JEMs are noted to suffer from misclassification due to
incapability of discriminating individuals in the same occupation as exposure estimates
are classified based on job and industry combinations although the level of exposure
may be varied between workers for some reasons such as different tasks and the
presence of preventive measures. The AsbJEM was validated and refined using a
cohort of former workers who participated in the health surveillance program. They
were involved in the program because they were referred by a doctor, had evidence of
pleural plaques or because of their awareness of probable asbestos exposure. As a
result, the calculation in the AsbJEM may still be underestimated for the cohort and the
modification to the AsbJEM proposed in this thesis may possibly overestimate the risk
of ARDs in another population. Therefore, it is preferable to apply the AsbJEM to a
different cohort in Australia, which may have a different exposure background from
the cohort in this thesis, in order to confirm the same trend of dose-response
relationship between occupational asbestos exposure and ARDs. If a similar result is
obtained, the reliability of the AsbJEM will be reinforced and the applicability will be
facilitated. A direct comparison of estimates of the risk of ARDs will be achieved
between different populations. Conversely, if the result is divergent from the findings
obtained in this thesis, the AsbJEM may need to be further refined using that cohort or
another in Australia. Although there will be a limitation regarding the applicability of
JEMs, FINJEM, one of the most well-recognized JEMs, is reported to be applicable to
other regions (Kauppinen et al., 2014). Therefore, it is a subject of interest to compare
86
the performance of this internationally renowned JEM with the AsbJEM for Australian
workers in the future research. In addition, it is also interesting to see if the AsbJEM is
applicable to people in other countries. This may be possible as international exposure
data was referred in the development of the AsbJEM for the early time periods (van
Oyen et al., 2015). If it is proved to be useful, it would help estimate the risk of
developing ARDs in that region, in particular in a country where a JEM or other
reliable exposure assessment tools are not available yet to facilitate standardized
exposure estimates. Although there are some methodological limitations for the
research projects comprising this thesis, which were described in each section, they are
not serious enough to undermine their significance. The result of dose-response
relationship between occupational asbestos exposure and non-malignant diseases
presented in this thesis will be a reference value for further research of estimating the
asbestos burden for the entire population across the nation as the number of
non-malignant ARDs is far beyond that of mesothelioma cases and it would only be
reliable if it is calculated based on accurate exposure estimates (Lenters et al., 2011).
Additional findings need to be accumulated through further research to acquire
widespread recognition of the AsbJEM and better refine the instrument to
appropriately address the situation in Australia.
87
References
Adegoke OJ, Blair A, Ou Shu X, et al. Agreement of job-exposure matrix (JEM)
assessed exposure and self-reported exposure among adult leukemia
patients and controls in Shanghai. Am J Ind Med. 2004; 45: 281-8.
Agusti AG, Roca J, Rodriguez-Roisin R, et al. Different patterns of gas exchange
response to exercise in asbestosis and idiopathic pulmonary fibrosis. Eur
Respir J. 1988; 1: 510-6.
Akira M, Morinaga K. The comparison of high-resolution computed tomography
findings in asbestosis and idiopathic pulmonary fibrosis. Am J Ind Med.
2016; 59: 301-6.
Albin M, Pooley FD, Strömberg U, et al. Retention patterns of asbestos fibres in lung
tissue among asbestos cement workers. Occup Environ Med.
1994; 51: 205-11.
Algranti E, Mendonca EM, DeCapitani EM, et al. Non-malignant asbestos-related
diseases in Brazilian asbestos-cement workers. Am J Ind Med. 2001; 40:
240-54.
Allen LP, Baez J, Stern MEC, et al. Trends and the economic effect of asbestos bans
and decline in asbestos consumption and production worldwide. Int J
Environ Res Public Health. 2018; 15: 531.
American Thoracic Society. Diagnosis and initial management of nonmalignant
diseases related to asbestos. Am J Respir Crit Care Med.
2004; 170: 691-715.
88
Armstrong B, Driscoll T. Mesothelioma in Australia: cresting the third wave. Public
Health Res Pract. 2016; 26: e2621614.
Baron PA. Measurement of airborne fibers. Ind Health. 2001; 39: 39-50.
Benke G, Sim M, Fritschi L, et al. Beyond the job exposure matrix (JEM): the task
exposure matrix (TEM). Ann Occup Hyg. 2000; 44: 475-82.
Berman DW, Crump KS. A meta-analysis of asbestos-related cancer risk that addresses
fiber size and mineral type. Crit Rev Toxicol. 2008; 38: 49-73.
Berry G, Gilson JC, Holmes S, et al. Asbestosis: a study of dose-response relationships
in an asbestos textile factory. Br J Ind Med. 1979; 36: 98-112.
Berry G, Reid A, Aboagye-Sarfo P, et al. Malignant mesotheliomas in former miners
and millers of crocidolite at Wittenoom (Western Australia) after more than
50 years follow-up. Br J Cancer. 2012; 106: 1016-20.
Bianchi C, Bianchi T. Global mesothelioma epidemic: trend and features. Indian J
Occup Environ Med. 2014; 18: 82-8.
Boffetta P. Health effects of asbestos exposure in humans: a quantitative assessment.
Med Lav. 1998; 89: 471-80.
Bourgkard E, Wild P, Gonzalez M, et al. Comparison of exposure assessment methods
in a lung cancer case-control study: performance of a lifelong task-based
questionnaire for asbestos and PAHs. Occup Environ Med. 2013; 70:
884-91.
Bouyer J, Dardenne J, Hémon D. Performance of odds ratios obtained with a
job-exposure matrix and individual exposure assessment with special
reference to misclassification errors. Scand J Work Environ Health. 1995;
21: 265-71.
89
Bouyer J, Hemon D. Studying the performance of a job exposure matrix. Int J
Epidemiol. 1993; 22 Suppl 2: S65-71.
Case BW, Abraham JL, Meeker G, et al. Applying definitions of "asbestos" to
environmental and "low-dose" exposure levels and health effects,
particularly malignant mesothelioma. J Toxicol Environ Health B Crit Rev.
2011; 14: 3-39.
Choi S, Kang D, Park D, et al. Developing asbestos job exposure matrix using
occupation and industry specific exposure data (1984-2008) in Republic of
Korea. Saf Health Work. 2017; 8: 105-115.
Churg A, Wiggs B. Fiber size and number in workers exposed to processed chrysotile
asbestos, chrysotile miners, and the general population. Am J Ind Med.
1986; 9: 143-52.
Cicioni C, London SJ, Garabrant DH, et al. Occupational asbestos exposure and
mesothelioma risk in Los Angeles county: application of an occupational
hazard survey job-exposure matrix. Am J Ind Med. 1991; 20: 371-9.
Clin B, Morlais F, Launoy G, et al. Cancer incidence within a cohort occupationally
exposed to asbestos: a study of dose--response relationships. Occup
Environ Med. 2011; 68: 832-6.
Copley SJ, Wells AU, Sivakumaran P, et al. Asbestosis and idiopathic pulmonary
fibrosis: comparison of thin-section CT features. Radiology.
2003; 229: 731-6.
Coughlin SS, Chiazze L, Jr. Job-exposure matrices in epidemiologic research and
medical surveillance. Occup Med. 1990; 5: 633-46.
Courtice MN, Wang X, Lin S, et al. Exposure-response estimate for lung cancer and
90
asbestosis in a predominantly chrysotile-exposed Chinese factory cohort.
Am J Ind Med. 2016; 59: 369-78.
Cruz MJ, Curull V, Pijuan L, et al. Utility of bronchoalveplar lavage for the diagnosis
of asbestos-related diseases. Arch Bronconeumol (English Edition).
2017; 53: 318-23.
de Klerk NH, Armstrong BK, Musk AW, et al. Cancer mortality in relation to measures
of occupational exposure to crocidolite at Wittenoom Gorge in Western
Australia. Br J Ind Med. 1989; 46: 529-36.
de Klerk NH, Musk AW, Williams V, et al. Comparison of measures of exposure to
asbestos in former crocidolite workers from Wittenoom Gorge, W.
Australia. Am J Ind Med. 1996; 30: 579-87.
Delclos GL, Gimeno D, Arif AA, et al. Occupational exposures and asthma in
health-care workers: comparison of seld-reports with a workplace-specific
job exposure matrix. Am J Epidemiol. 2009; 169: 581-7.
de Vocht F, Zock JP, Kromhout H, et al. Comparison of self-reported occupational
exposure with a job exposure matrix in an international community-based
study on asthma. Am J Ind Med. 2005; 47: 434-42.
Dodson RF, O'Sullivan M, Brooks DR, et al. Quantitative analysis of asbestos burden
in women with mesothelioma. Am J Ind Med. 2003; 43: 188-95.
Dufresne A, Harrigan M, Masse S, et al. Fibers in lung tissues of mesothelioma cases
among miners and millers of the township of asbestos, Quebec. Am J Ind
Med. 1995; 27: 581-92.
Dumortier P, Coplü L, de Maetelaer V, et al. Assessment of environmental asbestos
exposure iin Turkey by bronchoalveolar lavage. Am J Respir Crit Care
91
Med. 1998; 158: 1815-24.
Ehrlich R, Lilis R, Chan E, et al. Long term radiological effects of short term exposure
to amosite asbestos among factory workers. Br J Ind Med. 1992; 49:
268-75.
Ferrante D, Mirabelli D, Tunesi S, et al. Pleural mesothelioma and occupational and
non-occupational asbestos exposure: a case-control study with quantitative
risk assessment. Occup Environ Med. 2016; 73: 147-53.
Fletcher AC, Engholm G, Englund A. The risk of lung cancer from asbestos among
Swedish construction workers: self-reported exposure and a job exposure
matrix compared. Int J Epidemiol. 1993; 22: S29-35.
Furuya S, Chimed-Ochir O, Takahashi K, et al. Global asbestos disaster. Int J Environ
Res Public Health. 2018; 15: 1000.
Gasparrini A, Pizzo AM, Gorini G, et al. Prediction of mesothelioma and lung cancer
in a cohort of asbestos exposed workers. Eur J Epidemiol. 2008; 23: 541-6.
Goldberg M, Kromhout H, Guenel P, et al. Job exposure matrices in industry. Int J
Epidemiol. 1993; 22 Suppl 2: S10-5.
Gray C, Carey RN, Reid A. Current and future risks of asbestos exposure in the
community. Int J Occup Environ Health. 2016; 22: 292-9.
Hansell A, Ghosh RE, Poole S, et al. Occupational risk factors for chronic respiratory
disease in a New Zealand population using lifetime occupational history. J
Occup Environ Med. 2014; 56: 270-80.
Hansen J, de Klerk NH, Musk AW, et al. Environmental exposure to crocidolite and
mesothelioma: exposure-response relationships. Am J Respir Crit Care
Med. 1998; 157: 69-75.
92
Hansen J, de Klerk NH, Musk AW, et al. Individual exposure levels in people
environmentally exposed to crocidolite. Appl Occup Environ Hyg. 1997;
12: 485-90.
Hardt JS, Vermeulen R, Peters S, et al. A comparison of exposure assessment
approaches: lung cancer and occupational asbestos exposure in a
population-based case-control study. Occup Environ Med. 2014; 71: 282-8.
Hillerdal G. The human evidence: parenchymal and pleural changes. Ann Occup Hyg.
1994; 38: 561-7.
Hillerdal G. Mesothelioma: cases associated with non-occupational and low dose
exposures. Occup Environ Med. 1999; 56: 505-13.
Hodgson JT, Darnton A. The quantitative risks of mesothelioma and lung cancer in
relation to asbestos exposure. Ann Occup Hyg. 2000; 44: 565-601.
Hopstaken RM, Witbraad T, van Engelshoven JM, et al. Inter-observer variation in the
interpretation of chest radiographs for pneumonia in community-acquired
lower respiratory tract infections. Clin Radiol. 2004; 59: 743-52.
Huang JQ. A study on the dose-response relationship between asbestos exposure level
and asbestosis among workers in a Chinese chrysotile product factory.
Biomed Environ Sci. 1990; 3: 90-8.
Hyland RA, Yates DH, BenkeG, et al. Occupational exposure to asbestos in New South
Wales, Australia (1970-1989): development of an asbestos task exposure
matrix. Occup Environ Med. 2010; 67: 201-6.
International Labour Office. International classification of radiographs of
pneumoconioses. Geneva, Switzerland: International Labour Organization;
2003.
93
International Labour Office. Guidelines for the use of the ILO international
classification of radiographs of pneumoconiosis, revised edition 2011.
Occupational Safetry and Health Series 22.
Iwatsubo Y, Pairon JC, Boutin C, et al. Pleural mesothelioma: dose-response relation at
low levels of asbestos exposure in a French population-based case-control
study. Am J Epidemiol. 1998; 148: 133-42.
Jakobsson K, Stromberg U, Albin M, et al. Radiological changes in asbestos cement
workers. Occup Environ Med. 1995; 52: 20-7.
Jamrozik E, de Klerk N, Musk AW. Asbestos-related disease. Intern Med J. 2011; 41:
372-80.
Järvholm B, Malker H, Malker B, et al. Pleural mesothelioma and asbestos exposure in
the pulp and paper industries: a new risk group identified by linkage of
official registries. Am J Ind Med. 1988; 13: 561-7.
Kauppinen T, Heikkilä P, Plato N, et al. Construction of job-exposure matrices for the
Nordic Occupational Cancer Study (NOCCA). Acta Oncologica. 2009; 48:
791-800.
Kauppinen T, Toikkanen J, Pukkala E. From cross-tabulations to multipurpose
exposure information system: a new job-exposure matrix. Am J Ind Med.
1998; 33: 409-17.
Kauppinen T, Uuksulainen S, Saalo A. Use of the Finnish Information System on
Occupational Exposure (FINJEM) in epidemiologic, surveillance, and
other applications. Am Occup Hyg. 2014; 58: 380-96.
Kern DG, Hanley KT, Roggli VL. Malignant mesothelioma in the jewelry industry. Am
J Ind Med. 1992; 21: 409-16.
94
Koskinen K, Zitting A, Tossavainen A, et al. Radiographic abnormalities among
Finnish construction, shipyard and asbestos industry workers. Scand J
Work Environ Health. 1998; 24: 109-17.
Kottek M, Kilpatrick DJ. Estimating Occupational Exposure to Asbestos in Australia.
Ann Occup Hyg. 2016; 60: 531-2.
Kumar A, Cherian SV, Vassallo R, et al. Current Concepts in Pathogenesis, Diagnosis,
and Management of Smoking-Related Interstitial Lung Diseases. Chest.
2017.
Kwak KM, Paek D, Hwang SS, et al. Estimated future incidence of malignant
mesothelioma in South Korea: projection from 2014 to 2033. PLoS One.
2017; 12: e0183404.
Lacourt A, Gramond C, Rolland P, et al. Occupational and non-occupational
attributable risk of asbestos exposure for malignant pleural mesothelioma.
Thorax. 2014; 69: 532-9.
Lacourt A, Leveque E, GuichardE, et al. Dose-time-response association between
occupational asbestos exposure and pleural mesothelioma. Occup Environ
Med. 2017; 74: 691-7.
Lavoue J, Pintos J, Van Tongeren M, et al. Comparison of exposure estimates in the
Finnish job-exposure matrix FINJEM with a JEM derived from expert
assessments performed in Montreal. Occup Environ Med. 2012; 69:
465-71.
Le Moual N, Bakke P, Orlowski E, et al. Performance of population specific job
exposure matrices (JEMs): European collaborative analyses on
occupational risk factors for chronic obstructive pulmonary disease with
95
job exposure matrices (ECOJEM). Occup Environ Med. 2000; 57: 126-32.
Leigh J, Driscoll T. Malignant mesothelioma in Australia, 1945-2002. Int J Occup
Environ Health. 2003; 9: 206-17.
Lenters V, Vermeulen R, Dogger S, et al. A meta-analysis of asbestos and lung cancer:
is better quality exposure assessment associated with steeper slopes of the
exposure-response relationships? Environ Health Perspect. 2011; 119:
1547-55.
Lijmer JG, Mol BW, Heisterkamp S, et al. Empirical evidence of design-related bias in
studies of diagnostic tests. JAMA. 1999; 282: 1061-6.
Maemondo M, Inoue A, Kobayashi K, et al. Gefitinib or chemotherapy for
non-small-cell lung cancer with mutated EGFR. New Engl J Med.
2010; 362: 2380-8.
Magnani C, Terracini B, Ivaldi C, et al. A cohort study on mortality among wives of
workers in the asbestos cement industry in Casale Monferrato, Italy. Br J
Ind Med. 1993; 59: 779-84.
Maltoni C, Pinto C, Valenti D, et al. Mesotheliomas following exposure to asbestos
used in sugar refineries: report of 11 Italian cases. In: A Mehlman, A Upton,
eds. Advances in modern environmental toxicology. Vol ⅩⅩⅡ. The
identification and control of environmental and occupational diseases:
hazards and risks of chemicals in the oil refining industry. Prinecteon, NJ:
Princeton Scientific, 1994: 629-34.
Markowitz S. Asbestos-related lung cancer and malignant mesothelioma of the pleura:
selected current issues. Semin Respir Crit Care Med. 2015; 36: 334-46.
Markowitz SB, Levin SM, Miller A, et al. Asbestos, asbestosis, smoking, and lung
96
cancer. New findings from the North American insulator cohort. Am J
Respir Crit Care Med. 2013; 188: 90-6.
Markowitz SB, Morabia A, Lilis R, et al. Clinical predictors of mortality from
asbestosis in the North American Insulator Cohort, 1981 to 1991. Am J
Respir Crit Care Med. 1997; 156: 101-8.
Mastrangelo G, Ballarin MN, Bellini E, et al. Asbestos exposure and benign asestos
diseases in 772 formerly exposed workers: dose-response relationships.
Am J Ind Med. 2009; 52: 596-602.
Matrat M, Pairon JC, Paolillo AG, et al. Asbestos exposure and radiological
abnormalities among maintenance and custodian workers in buildings with
friable asbestos-containing materials. Int Arch Occup Environ Health.
2004; 77: 307-12.
McElvenny DM, van Tongeren M, Turner MC, et al. The INTEROCC case-control
study: risk of meningioma and occupational exposure to selected
combustion products, dusts and other chemical agents. Occup Environ Med.
2018; 75: 12-22.
Mensi C, De Matteis S, Dallari B, et al. Incidence of mesothelioma in Lombardy, Italy:
exposure to asbestos, time patterns and future projections. Occup Environ
Med. 2016; 73: 607-13.
Musk AW, de Klerk NH, Ambrosini GL, et al. Vitamin A and cancer prevention I:
observations in workers previously exposed to asbestos at Wittenoom,
Western Australia. Int J Cancer. 1998; 75: 355-61.
Musk ABW, de Klerk N, Brims FJ. Mesothelioma in Australia: a review. Med J Aust.
2017; 207: 449-52.
97
Musk AW, de Klerk NH, Reid A, et al. Mortality of former crocidolite (blue asbestos)
miners and millers at Wittenoom. Occup Environ Med. 2008; 65: 541-3.
Musk AW, Olsen N, Alfonso H, et al. Pattern of malignant mesothelioma incidence and
occupational exposure to asbestos in Western Australia. Med J Aust. 2015;
203: 251-2.
Nam JM, Rice C, Gail MH. Comparison of asbestos exposure assessments by
next-of-kin respondents, by an occupational hygienist, and by a
job-exposure matrix from the National Occupational Hazard Survey. Am J
Ind Med. 2005; 47: 443-50.
Noble PW, Albera C, Bradford WZ, et al. Pirfenidone in patients with idiopathic
pulmonary fibrosis (CAPACITY): two randomised trials. Lancet. 2011;
377: 1760-9.
Nuyts V, Vanhooren H, Begyn S, et al. Asbestos bodies in bronchoalveolar lavage in
the 21st century: a time-trend analysis in a clinical population. Occup
Environ Med. 2017; 74: 59-65.
Offermans NS, Vermeulen R, Burdorf A, et al. Occupational asbestos exposure and
risk of pleural mesothelioma, lung cancer, and laryngeal cancer in the
prospective Netherlands cohort study. J Occup Environ Med. 2014; 56:
6-19.
Offermans NS, Vermeulen R, Burdorf A, et al. Comparison of expert and job-exposure
matrix-based retrospective exposure assessment of occupational
carcinogens in the Netherlands Cohort Study. Occup Environ Med. 2012;
69: 745-51.
Olsen NJ, Franklin PJ, Reid A, et al. Increasing incidence of malignant mesothelioma
98
after exposure to asbestos during home maintenance and renovation. Med J
Aust. 2011; 195: 271-4.
Olsson AC, Vermeulen R, Schüz J, et al. Exposure-response analyses of asbestos and
lung cancer subtypes in a pooled analysis of case-control studies.
Epidemiology. 2017; 28: 288-99.
Pairon JC, Orlowski E, Iwatsubo Y, et al. Pleural mesothelioma and exposure to
asbestos: evaluation from work histories and analysis of asbestos bodies in
bronchoalveolar lavage fluid or lung tissue in 131 patients. Ocuup Environ Med.
1994; 51: 244-9.
Pannett B, Coggon D, Acheson ED. A job-exposure matrix for use in population based
studies in England and Wales. Br J Ind Med. 1985; 42: 777-83.
Paris C, Benichou J, Raffaelli C, et al. Factors associated with early-stage pulmonary
fibrosis as determined by high-resolution computed tomography among
persons occupationally exposed to asbestos. Scand J Work Environ Health.
2004; 30: 206-14.
Paris C, Martin A, Letourneux M, et al. Modelling prevalence and incidence of fibrosis
and pleural plaques in asbestos-exposed populations for screening and
follow-up: a cross-sectional study. Environmental health : a global access
science source. 2008; 7: 30.
Paris C, Thierry S, Brochard P, et al. Pleural plaques and asbestosis: dose- and time-
response relationships based on HRCT data. Eur Respir J. 2009; 34: 72-9.
Park EK, Hannaford-Turner KM, Hyland RA, et al. Asbestos-related occupational lung
diseases in NSW, Australia and potential exposure of the general
population. Ind Health. 2008; 46: 535-40.
99
Park EK, Yates DH, Hyland RA, et al. Asbestos exposure during home renovation in
New South Wales. Med J Aust. 2013; 199: 410-3.
Park EK, Yates DH, Wilson D. Lung function profiles among individuals with
nonmalignant asbestos-related disorders. Saf Health Work. 2014; 5: 234-7.
Peters S, Vermeulen R, Cassidy A, et al. Comparison of exposure assessment methods
for occupational carcinogens in a multi-centre luung cancer case-control
study. Occup Environ Med. 2011; 68: 148-53.
Peters S, Vermeulen R, Portengen L, et al. SYN-JEM: A quantitative job-exposure
matrix for five lung carcinogens. Ann Occup Hyg. 2016; 60: 795-811.
Pukkala E, Guo J, Kyyrönen P, et al. National job-exposure matrix in analyses of
census-based estimates of occupational cancer risk. Scand J Work, Environ
Health. 2005; 31: 97-107.
Quinlan PJ, Earnest G, Eisner MD, et al. Performance of self-reported occupational
exposure compared to a job exposure matrix approach in asthma and
chronic rhinitis. Occup Environ Med. 2009; 66: 154-60.
Rees D, Phillips JI, Garton E, et al. Asbestos lung fibre concentrations in South African
chrysotile mine workers. Ann Occup Hyg. 2001; 45: 473-7.
Reid A, Berry G, de Klerk N, et al. Age and sex differences in malignant mesothelioma
after residential exposure to blue asbestos (crocidolite). Chest. 2007; 131:
376-82.
Reid A, de Klerk NH, Magnani C, et al. Mesothelioma risk after 40 years since first
exposure to asbestos: a pooled analysis. Thorax. 2014; 69: 843-50.
Rogers AJ, Leigh J, Berry G, et al. Relationship between lung asbestos fiber type and
concentration and relative risk of mesothelioma. a case-control study.
100
Cancer. 1991; 67: 1912-20.
Sanden A, Järvholm B, Larsson S, et al. The risk of lung cancer and mesothelioma
after cessation of asbestos exposure: a prospective cohort study of shipyard
workers. Eur Respir J. 1992; 5: 281-5.
Sjöström M, Lewne W, Alderling M, et al. A job-exposure matrix for occupational
noise: development and validation. Ann Occup Hyg. 2013; 57: 774-83.
Soeberg MJ, Creighton N, Currow DC, et al. Patterns in the incidence, mortality and
survival of malignant pleural and peritoneal mesothelioma, New South
Wales, 1972-2009. Aust N Z J Public Health. 2016; 40: 255-62.
Soeberg MJ, Vallance DA, Keena V, et al. Australia`s ongoing legacy of asbestos:
significant chalenges remain even after the complete banning of asbestos
almost fifteen years ago. Int J Environ Res Public Health. 2018; 15: 384.
Solovieva S, Pehkonen I, Kausto J, et al. Development and validation of a job exposure
matrix for physical risk factors in low back pain. PLoS One.
2012; 7: e48680.
Stayner L, Smith R, Bailer J, et al. Exposure-response analysis of risk of respiratory
disease associated with occupational exposure to chrysotile asbestos.
Occup Environ Med. 1997; 54: 646-52.
Tuomi T, Huuskonen MS, Tammilehto L, et al. Occupational exposure to asbestos as
evaluated from work histories and analysis of lung tissues from patients
with mesothelioma. Br J Ind Med. 1991; 48: 48-52.
van der Bij S, Koffijberg H, Lenters V, et al. Lung cancer risk at low cumulative
asbestos exposure: meta-regresson of the exposure-response relationship.
Cancer Causes Control. 2013; 24: 1-12.
101
van Oyen SC, Peters S, Alfonso H, et al. Development of a Job-Exposure Matrix
(AsbJEM) to Estimate Occupational Exposure to Asbestos in Australia.
Ann Occup Hyg. 2015; 59: 737-48.
Velasco-Garcia MI, Cruz MJ, Ruano L, et al. Reproducibility of asbestos body counts
in digestions of autopsy and surgical lung tissue. Am J Ind Med.
2011; 54: 597-602.
Vierikko T, Jarvenpaa R, Autti T, et al. Chest CT screening of asbestos-exposed
workers: lung lesions and incidental findings. Eur Respir J. 2007; 29:
78-84.
Villeneuve PJ, Parent ME, Harris SA, et al. Occupational exposure to asbestos and
lung cancer in men: evidence from a population-based case-control study
in eight Canadian provinces. BMC cancer. 2012; 12: 595.
Warmock ML. Lung asbestos burden in shipyard and constructioon workers with
mesothelioma: comparison with burdens in subjects with asbestosis or lung
cancer. Environ Res. 1989; 50: 68-85.
Weiss W. Asbestosis: a marker for the increased risk of lung cancer among workers
exposed to asbestos. Chest. 1999; 115: 536-49.
Yeung P, Rogers A, Johnson A. Distribution of mesothelioma cases in different
occupational groups and industries in Australia, 1979-1995. Appl Occup
Environ Hyg. 1999; 14: 759-67.
102
Appendices
T +61 8 6488 3703 / 4703F +61 8 6488 8775E [email protected]
Human Ethics
Office of Research Enterprise
The University of Western AustraliaM459, 35 Stirling HighwayCrawley WA 6009 Australia
CRICOS Provider Code: 00126G
Our Ref: RA/4/20/4746
02 August 2018
Dr Peter FranklinSchool of Population and Global HealthMBDP: M431
Dear Doctor Franklin
HUMAN RESEARCH ETHICS OFFICE – EXEMPTION FROM ETHICS REVIEW
Assessment of the performance of the Asbestos Job Exposure Matrix in Australia.
Based on the information you have provided to the the Human Ethics office in relation to the above project, the described activity has beenassessed as exempt from ethics review at the University of Western Australia.
However, should there be any significant changes to the protocol, you must contact the HREO to determine whether your exempt status remainsvalid or whether you will be required to submit an application for ethics approval.
If you have any queries please contact the Human Ethics office at [email protected].
Please ensure that you quote the file reference – RA/4/20/4746 – and the associated project title in all future correspondence.
Yours sincerely
Mark DaviesManager, Human Ethics
Name Faculty / School RoleDr Peter Franklin School of Population and Global Health Chief Investigator
The University of Western AustraliaCrawley WA 6009 Australia
T +61 8 6488 3703 / 4703F +61 8 6488 8775
E [email protected] Provider Code: 00126G