2. Occupational Cancer
Epidemiology in Canada
• Canada has traditionally been a world leader in
occupational cancer research
• Major case-control and cohort studies
• A leader in the development of retrospective
exposure assessment methods, both expert
assessment for population-based studies and
quantitative assessment for industry-based studies
• Where are we now?
2
3. Administrative Health Data in Canada
3
Physician
Billing
Hospital
Discharge
Ambulatory
Care
Tumour
Registry
Where
did they
work?
The missing
piece in Canada
is the lack of
occupation and
industry in
health records
4. Occupational Disease Surveillance System
Compensation
claimants
1983-2014
Registered
Persons
Database
1990-
Cancer
Records
1964-
Hospital
Discharge
2006-
Physician
Billing
1991-
Ambulatory
Care
2006-
4
Cohort (name, sex, DoB)
Occupation/Industry
Nature of Claim
Employer
Residence
Vital Status
Health
Insurance
Number
28 Cancers: lung, breast,
prostate, colon…
Asthma, copd, IPF, silicosis,
asbestosis, AMI, dermatitis,
CTS, Raynaud’s, toxic effects
of CO & Pb, hearing loss
2.2 M
people
5. The CanCHEC Cohort
Record linkage projects – example of 2001 cohort (CanCHEC)
2017-11-28Statistics Canada • Statistique Canada5
Census long-
form (ca. 20%
of population)*
In scope
Income
tax
Deterministic and
probabilistic
linkage
79% linked
Exclusions:
<19 years, overseas,
live in institution
Mortality database:
hospital death registries
Deterministic
linkage
99% linked
Cohort
Ref: Pinault LL, Finès P, Labrecque-Synnott F, Saidi A, Tjepkema M. 2016. The 2001 Canadian Census-Tax-Mortality Cohort: a
10-year follow-up. Analytical Studies: Methods and References. Statistics Canada. Cat No. 11-633-X.
6. The CanCHEC Cohort
2017-11-28Statistics Canada • Statistique Canada6
Timeline*
1991 2001 2011
1991 CanCHEC
1996 CanCHEC
2001 CanCHEC
Postal codes (tax files)
2,6 million
575,700 deaths
3,6 million
565,600 deaths
3,5 million
347,000 deaths
9.7 million
1.5 million deaths
1981
*Also linked to cancer registries – follow cohort for cancer diagnoses.
7. Statistics Canada: Social Data Linkage
Environment (SDLE) – Linked file examples
• Derived Record Depository (DRD) – a national longitudinal
data base of individuals derived from a number of Statistics
Canada data files & containing only basic personal identifiers.
• Vital Statistics – Death Database (1970-2012)
• Canadian Cancer Registry (1992-2013)
• Survey of Labour and Income Dynamics (Panel 3, 1999-2004)
• 2001 Census Tax Mortality Cohort 2001
• Canadian Forces Cancer and Mortality Study II (1976-2015)
• Canadian Health Measures Survey (Cycle 1-cycle 4, 2007-2014)
• Canadian Community Health Survey - Annual and focus
content cycles 2000-01 (Cycle 1.1) to 2014
7
8. Limitations of Administrative Data
and Statistics Canada Linkages
• Availability of a single job & industry title in most
datasets, with no dates or other details
• Accuracy of Occupational Coding
• Inability to apply more advanced exposure
assessment methods
• Lack of potential confounding risk factors
8
10. Assisted Coding of Occupations & Industries
• Codage Assisté des Professions et des Secteurs d'activité
(CAPS.CA)
– Free online tool to assist experts in the coding occupation or
industry developed by Jerome Lavoue and other of the CHUM
Research Centre (Universite de Montreal), in collaboration with InVS
and CREDIM.
• English and a French interface with the following
classifications:
– Canadian Classification and Dictionary of Occupations (1971-1989)
– National Occupational Classification (Canada, 2011)
– Standard Industrial Classification (Canada, 1970 and 1980)
– North American Industry Classification System (2012)
– ISCO, ILO (1968, 1988 and 2008) and ISIC, UN (1968)
– A similar selection of French language coding schemes
10
14. Do We Need an Occupational CANUE?
“Every location in Canada can be described by a complex set of
environmental factors – the amount of nearby traffic, local air
quality, access to greenspaces, opportunities for walking and
cycling, the amount of noise and light pollution, to name a few”
14
16. OMEGA-NET
The overarching concept is to create a network to
optimize the use of occupational, industrial, and
population cohorts at the European level. OMEGA-
NET will advance:
– collaboration of existing cohorts, with extensive
contemporary information on employment and
occupational exposures
– coordination and harmonization of occupational
exposure assessment efforts
– facilitation of an integrated research strategy for
occupational health in Europe
16
17. Exposome and Exposomics - NIOSH
The exposome can be defined as the measure of all exposures
over an individual’s lifetime & how these relate to health.
Exposure begins before birth & includes insults from
environmental & occupational sources.
Exposomics relies on internal & external exposure assessment.
Common themes include:
• biomarkers to determine exposure, effect of exposure,
disease progression, and susceptibility factors
• use of technologies that result in large amounts of data
• data mining techniques to find associations between disease
and exposures, effect of exposures, genetics, other factors
17
20. Conclusions – Opportunities
• Linkable data provides opportunities to advance
the field with low cost studies
• The BIG Cohorts my offer opportunities to leverage
large investments that have already been
committed
– The timing may be good to start a new Network
such as CANUE or OMEGA-Net
20
21. Conclusions - Advancing
• Advances in exposure assessment, under the banner
of exposomics, can take advantage of new
technologies, but are also good marketing
• We need to continue to advance exposure
assessment and address new and upcoming issues
– Creative new approaches are needed
• We still need new case-control and cohort studies
– Reusing existing datasets is cost-effective and
reasonable, but won’t sustain us much longer
21
Large scale occupational disease surveillance has been challenging in many countries, with a few notable exceptions, such as the Nordic countries with their substantial record linkage abilities. We present initial results for lung cancer from a new Canadian Occupational Disease Surveillance System.
The surveillance cohort was created using data from Ontario, Canada time-loss workers’ compensation claims 1983-2016 (96% for injuries) linked to cancer registry records. Follow-up was from first claim date until diagnosis, death, loss-to-follow-up or 2016. Hazard ratios (HRs) were calculated for each industry/occupation using Cox Proportional Hazard models, adjusted for year of birth and stratified on gender.
The study population was 740,000 women and 1,430,000 men. Significant excess risks were observed in many of the a priori suspected occupations and industries, particularly in construction, mining, and transportation occupations. In addition, other relevant associations were observed among both women and men, such as for janitors and cleaners (men: HR=1.22, 95% CI=1.16-1.29, women: HR=1.22, 95% CI=1.13-1.32) and primary metals industry (men: HR=1.18, 95% CI=1.11-1.25, women: HR=1.20, 95% CI=0.89-1.60). Many sex-specific associations were also observed, particularly in women (such as printing and publishing industries: HR=1.42, 95% CI=1.23-1.65 and chemical, rubber and plastic processing occupations HR=1.31, 95% CI=1.15-1.51), which will need further investigation.
The excess risks observed in many a priori suspected groups provides a good confirmation that this study can produce valid results and identify new associations. Triage methods are being developed to target new associations in need of further investigation. Future analyses will use hospital discharge data and outpatient visits.
For occupational health research, missing piece: where did they work? Great data on health outcomes, but none of these data sources include any information about work history