Article Text
PDF
Abstract
Objectives To understand rates of work-related COVID-19 (WR-C19) infection by occupational exposures across waves of the COVID-19 pandemic in Ontario, Canada.
Methods We combined workers’ compensation claims for COVID-19 with data from Statistics Canada’s Labour Force Survey, to estimate rates of WR-C19 among workers spending the majority of their working time at the workplace between 1 April 2020 and 30 April 2022. Occupational exposures, imputed using a job exposure matrix, were whether the occupation was public facing, proximity to others at work, location of work and a summary measure of low, medium and high occupational exposure. Negative binomial regression models examined the relationship between occupational exposures and risk of WR-C19, adjusting for covariates.
Results Trends in rates of WR-C19 differed from overall COVID-19 cases among the working-aged population. All occupational exposures were associated with increased risk of WR-C19, with risk ratios for medium and high summary exposures being 1.30 (95% CI 1.09 to 1.55) and 2.46 (95% CI 2.10 to 2.88), respectively, in fully adjusted models. The magnitude of associations between occupational exposures and risk of WR-C19 differed across waves of the pandemic, being weakest for most exposures in period March 2021 to June 2021, and highest at the start of the pandemic and during the Omicron wave (December 2021 to April 2022).
Conclusions Occupational exposures were consistently associated with increased risk of WR-C19, although the magnitude of this relationship differed across pandemic waves in Ontario. Preparation for future pandemics should consider more accurate reporting of WR-C19 infections and the potential dynamic nature of occupational exposures.
- COVID-19
- Epidemiology
- Public health
Data availability statement
Data may be obtained from a third party and are not publicly available.
Statistics from Altmetric.com
WHAT IS ALREADY KNOWN ON THIS TOPIC
Occupational exposures, such as working in close proximity and having direct contact with the public, are associated with a risk of COVID-19 infection.
Few studies have been able to restrict COVID-19 infections to only those that are work related, or able to restrict the population at risk for work-related COVID-19 infection to only those working outside of home.
WHAT THIS STUDY ADDS
We observed an increased risk between occupational exposures and work-related COVID-19 infections, among Ontario workers who were working outside of home. While the direction of risk for occupational exposures was consistent across waves of the pandemic, the magnitude of risk was not.
We also observed a divergence between rates of work-related COVID-19 infection and rates of overall COVID-19 infection among the working-aged population.
HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY
Preparation for, and response to, future pandemics should consider more accurate reporting of work-related COVID-19 infections, compared with overall COVID-19 infection.
This study also highlights the potentially dynamic nature of occupational exposures and COVID-19 risk across waves of the pandemic.
Introduction
The COVID-19 pandemic has had profound effects on labour markets around the world.1 The workplace was identified in the early phases of the pandemic as a potentially important setting for COVID-19 transmission, and examinations in many countries have focused on rates of confirmed COVID-19 infections across occupational or industrial groups. Workplace exposures associated with increased risk of COVID-19 infections include having direct contact with potentially infected individuals (eg, healthcare), where workers are in close proximity to each other at work, indoor settings with poor ventilation and where infection control practices are inadequate.2
However, important information gaps remain. For example, much of what we know about risk of COVID-19 across occupations has used broad occupational groups, with occupational information sometimes captured at a single time point, many years prior to the pandemic.3–5 A smaller number of studies have examined specific occupational exposures associated with COVID-19 risk.6 In addition, COVID-19 outcomes have ranged from positive antigen tests to hospitalisations to mortality from COVID-19. While each of these represents important health outcomes, the direct attribution of occupation per se to more distal health outcomes such as hospitalisation and mortality is more challenging than for initial COVID-19 infection, given the competing roles of age, ethnicity and other chronic conditions for more severe COVID-19 outcomes.7 Further, while studies to date have adjusted for other factors associated with occupation that might also be associated with risk of COVID-19, such as household crowding or other living conditions, COVID-19 outcomes used across most studies to date have lacked a specific attribution to work. The use of general COVID-19 infections likely reduces the specificity of the occupational exposure risk estimates, given occupation may be a marker of multiple different work-related and non-work-related risks. A final challenge has been the inability of studies to exclude the changing proportion of the labour market who worked from home during various phases of the pandemic. Labour market participants who worked from home are likely not at risk of work-related COVID-19 (WR-C19) infection.8 An exception is a paper from Ontario which was able to examine rates of COVID-19 cases due to workplace outbreaks compared with the overall incidence rates among the working-aged population. In this study, while overall workplace outbreak standardised incidence ratios were below the overall incidence rates among adults aged 16–69 years of age, elevated rates were observed in particular industries.9
Workers’ compensation data can offer unique insights into rates of WR-C19 infections,10 as this data source contains only COVID-19 infections that are certainly due to transmission at the workplace; thus, enabling more precise estimates of occupational exposures associated with increased risk of WR-C19 infections. As such, the objective of this study is to estimate rates of WR-C19 infection across different occupational exposures in general, and across different waves of the COVID-19 pandemic in Ontario, Canada. Second, to enable greater comparison to previous studies examining work-exposures and COVID-19 infection, we also examine trends in WR-C19 infections compared with overall COVID-19 infections among the working-aged population.
Methods and data
This paper used workers’ compensation claims for COVID-19 combined with Labour Force Survey (LFS) estimates of the number workers, and hours of work at the workplace, between 1 April 2020 and 30 April 2022. To identify WR-C19 infections, we used accepted lost-time claims for COVID-19 from the Ontario Workplace Safety and Insurance Board (WSIB). The WSIB has a central mandate to compensate workers who have sustained a work-related injury or illness for loss of employment income (wage replacement) and the cost of healthcare and associated services related to the injury or illness. In Ontario, it is estimated that approximately 76% of the workforce is covered by the WSIB.11 Industries not specifically covered by the WSIB include much of the financial industry, as well as private employers in certain settings (eg, childcare). Across all provincial workers’ compensation agencies in Canada, COVID-19 claims were adjudicated for work relatedness. For a COVID-19 claim to be accepted, evidence was required that (1) the risk of contracting COVID-19 at work was greater than the risk of contracting COVID-19 in the community and (2) that work factors significantly contributed to the COVID-19 infection.12 While all cases of COVID-19 that are work-related may not be captured within workers’ compensation systems (eg, when the risk of contracting COVID-19 outside of work was judged to be higher than the risk at work—for example, due to living conditions or a previous household infection—however, the actual COVID-19 infection was work related), we can assume that all accepted workers’ compensation claims for COVID-19 have definitely been acquired through workplace exposure(s).
In order to estimate rates of WR-C19 infections monthly hours of workplace exposure across demographic and occupational characteristics were estimated using Statistics Canada’s LFS.13 LFS respondents are representative of 98% of non-institutionalised Canadians aged 15 years and over, excluding persons living on reserves and other Indigenous settlements, and full-time members of the Canadian Armed Forces.14 Starting in April 2020, a series of supplemental questions were added to the LFS to understand the impacts of the COVID-19 pandemic. Questions included the location where the respondent had worked the most hours in the previous week. To better estimate the population at risk of WR-C19 infection, we restricted our sample to only those people who indicated that they worked at the worksite, or outside of the home, but not in a particular location (eg, travelling from site to site or in a work vehicle). In addition, we removed from the LFS estimates of respondents who were working in industry groups with no, or incomplete, coverage from the WSIB, given not all employees in these industry groups are able to submit claims to the WSIB.15 LFS supplement questionnaires were only asked of respondents aged 15–69 years of age, so we further limited our sample of compensation claims to this age range.
Occupational exposures
We assigned occupational exposures likely to increase risk of WR-C19 infection to both workers’ compensation claims and LFS estimates using a job exposure matrix (JEM) developed by the US Council of State and Territorial Epidemiologists Occupational Health Subcommittee.16 This JEM was developed for the Standard Occupational Classification (SOC) system, based on questions used in the O*Net database. The concordance between the SOC used in the O*Net database and the National Occupational Classification (NOC) system used in the LFS and WSIB claims was facilitated through Codage Assisté des Professions et Secteurs d’activité (CAPS)-Canada,17 which provided an actual concordance across coding systems for 350 of 500 NOC codes, and Statistics Canada,18 which provided a theoretical crosswalk for the remaining 150 NOC codes. Occupational exposures included if the occupation was public facing; the primary work location and the proximity of the worker to other colleagues while at work.
Public-facing work (yes/no) was determined in the O*NET database using two questions ‘How important is performing for or working with the public to the performance of your current job?’ and ‘In your current job how important is it to work with external customers (as in retail sales) or the public in general (as in police work)?’. Occupations where working with the public is very important OR working with the public is important and the proximity of work with the public is closer than arms length were classified as public facing. All other occupations were classified as non-public facing.
Primary working location was determined using five questions ‘How often does this job require working indoors in environmentally controlled conditions?’; ‘How often does this job require working indoors in non-controlled environmental conditions (eg, warehouse without heat)?’; ‘How often does this job require working outdoors, exposed to all weather conditions?’; ‘How often does this job require working outdoors, under cover (eg, structure with roof but no walls)?’ and ‘How often does this job require working in a closed vehicle or equipment (eg, car)?’. Occupations were grouped into the following four groups: primarily indoors; mixed—often indoors; mixed—often outdoors and primarily outdoors.
Proximity to other workers was determined using three questions ‘How physically close are you to other people when you perform your job?’; ‘How important is it to work with others in a group or team in this job?’ and ‘How often do you have to have face-to-face discussions with individuals or teams in this job?’. Using these questions the following four groups of proximity were defined: very close; works nearly arms distance or closer; close—shared work space with high likelihood of both working in teams and face-to-face discussions; somewhat close—shared work space with high likelihood of working in teams OR face-to-face discussions; not close—shared work space but low likelihood of working in teams or face-to face discussions, or working in a private office or not near other people.
We also grouped occupations into three overall exposure categories: higher-risk occupations (occupations with at least two of the three exposure measures); lower-risk occupations (occupations defined as not close and not publicly facing regardless of work locations) and medium risk, which includes all other combinations of occupational exposures.16
Common occupations within each of the occupational exposure categories are listed in online supplemental table A1.
Supplemental material
Covariates
Covariates included: The wave of the pandemic (1=April 2020 to August 2020; 2=September 2020 to February 2021; 3=March 2021 to June 2021 (Alpha); 4=July 2021 to November 2021 (Delta); 5=December 2021 to April 2022 (Omicron)); industry of employment (healthcare vs non-healthcare) given the strong relationship between the healthcare industry and COVID-19 infections in particular at the start of the COVID-19 pandemic; age group in 10-year groupings and sex (male/female). Wave of the pandemic was also included as an effect modifier (stratification variable) in certain analyses.
Analysis
To estimate rates of WR-C19 infection numerator (event) information from the WSIB claims was matched with denominator (exposure) information from the LFS, by month, occupational exposure, age, sex and healthcare industry. Initial models examined crude rates of WR-C19 claims per full-time equivalent (FTE) hours of exposure, across occupational exposures and covariates. One FTE is equivalent to a 37.5-hour work week, across each of the 108.6 weeks between 1 April 2020 and 30 April 2022 (4071 hours in total). CIs around rates were estimated using a previously established approach.19 Estimates for confirmed overall COVID-19 cases among the working-aged population (2–69 years of age) in Ontario over the same five time periods were generated using publicly available data from the Ontario Data Catalogue.20 A series of negative binomial regression models examined the relationship between occupational exposures adjusted for study covariates (wave, age, sex and healthcare industry). These models used aggregated data for claims and hours of exposure, across all exposure and covariate combinations. In these models, the count of claims is the outcome, and the log of the hours of exposure is used as a model offset. Additional details about this method are provided in online supplemental material. Each occupational exposure was modelled separately, as modelling more than one exposure in the same model resulted in sparse data, given the relationships between exposures. Subsequent models then examined potential effect modification across waves of the pandemic. All analyses were completed using PROC GENMOD in SAS V.9.4.
Results
Table 1 presents overall rates of COVID-19 claims per hours of work across occupational exposures, industry group, age, sex and COVID-19 wave. During the total study time period, there were 43 208 accepted lost-time claims for COVID-19. Through the course of the study time period, COVID-19 claims increased from approximately 10.11 claims per 1000 FTEs in the first phase of the pandemic, to 17.65 per 1000 FTEs in the second wave (September 2020 to February 2021), dropping to a low of 1.98 per 1000 FTEs in July 2021 to November 2021, increasing to a high of 26.41 per 1000 FTEs during the Omicron wave (December 2021 to April 2022). Occupational exposures with the highest rates of COVID-19 were public-facing occupations (29.47 claims per 1000 FTEs) and occupations working in very close proximity (26.62 per 1000 FTEs). Across all variables, rates were highest among healthcare industry workers (93.84 per 1000 FTEs).
COVID-19 workers’ compensation claims, full-time equivalent (FTE) hours at the workplace and rate of COVID-19 claims per 1000 FTEs
At the bottom of table 1 are rates of overall COVID-19 infections among the working-aged population. A comparison of rates of WR-C19 and overall COVID-19 infection is also provided in figure 1. Rates of overall COVID-19 infection increased from 173 cases per day in the first time period to 1468 cases per day in the third wave of the pandemic. They then dropped to 337 cases per day within the fourth (Delta) wave, increasing to 3189 cases per day during the Omicron wave. Trends in total COVID-19 claims and rates of WR-C19 claims differed over the study time period, with rates of workers’ compensation claims declining between waves 2 and 3 and declining to one-fifth of the rate of the first wave during wave 4. In comparison, rates per day of overall COVID-19 among the working-aged population in Ontario were twice as high in the fourth wave compared with the first wave. Finally, rates of WR-C19 claims during the omicron wave were just over 2.5 times as high as they were in wave 1, compared with rates of overall COVID-19 cases among the working population, which were more than 18 times higher than the first wave.
Rates of WR-C19 infection per 1000 full-time equivalents, and cases of general COVID-19 infections per day among the working-aged population in Ontario, 1 April 2020 to 30 April 2022. WR-C19, work-related COVID-19.
Table 2 presents adjusted rate ratios (RRs) for WR-C19 claims across occupational exposure groups, adjusted for age, sex, wave of the pandemic and industry (healthcare vs on-healthcare). All occupational exposures were associated with increased risk of WR-C19 infection, with risk ratios ranging from 1.31 (95% CI 1.14 to 1.52) for public-facing occupations to 2.90 (95% CI 2.47 to 3.39) for occupations working in very close proximity of others. A graded relationship between risk for WR-C19 infection was observed across the combined risk grouping, with occupations with a medium risk having a 30% elevated risk (RR 1.30, 95% CI 1.09 to 1.55), and occupations with a high risk having an 146% higher risk (RR 2.46, 95% CI 2.10 to 2.88) of COVID-19 claims compared with occupations with a low risk.
Adjusted* rate ratio (RR) estimates for occupational exposures and risk of COVID-19 workers’ compensation claim
Table 3 presents RRs for occupational exposures and WR-C19 claims stratified by wave of the pandemic. The relationship between occupational exposure and risk of WR-C19 claim changed throughout the pandemic, being markedly stronger in the first and fifth waves of the pandemic (with the exception of public-facing occupations). The relationship between occupational exposures and risk of WR-C19 infection was generally weakest in the third wave of the pandemic (March 2021 to June 2021). It should be noted that the RRs in table 3 are specific to different waves of the pandemic, and do not represent the overall risk of exposure categories throughout the pandemic. That is, changes in relative risk over time may reflect both changes in absolute risk in the exposed group, or the unexposed (reference) group, or both, across waves. As an alternative, we also present occupational exposure risk relative to the unexposed group in the first wave of the pandemic. These results are provided in online supplemental material to this paper for comparison in online supplemental table A2.
Adjusted* estimates for occupational exposures and risk of COVID-19 workers’ compensation claim
Discussion
The current paper used novel data to better understand the association between occupational exposures and risk of WR-C19 infection, and the extent to which this risk varies over time. We observed that public-facing occupations, proximity to other workers and location of work were each associated with increased risk of workers’ compensation claims for COVID-19 after adjustment for age, sex, wave of the pandemic and healthcare industry. The relationship between occupational exposures changed across different waves of the pandemic, being weakest during the time period March 2021 to June 2021 for most exposures, and strongest at the start and end of our time period.
Our finding of the changing relationships between occupational exposures and COVID-19 across different waves of the pandemic, extends the previous work of others. Rhodes et al examined odds of new COVID-19 infections (identified using PCR tests) in the UK among essential workers (healthcare, social and education, and other essential workers) compared with other workers over four phases of the pandemic (April 2020–Sepember 2020; October 2020–February 2021; March 2021–May 2021 and June 2021–October 2021). They observed that the odds of new infections among healthcare workers decreased throughout the pandemic, being lower than non-essential workers in the third and fourth phases of their study. Conversely odds for new infections increased between the first and second phases among social and educational workers and other non-essential workers, remaining elevated into the third wave for other essential workers, and for both the third and fourth waves among social and educational workers.4 A Dutch study examined the relationships between occupational exposures, using a JEM approach21 similar to our study, across three time periods: 26 February 2020–24 August 2020; 25 August 2020–21 June 2021 and 22 June 2021–12 December 2021. In that study, occupational exposure was most strongly associated with positive PCR tests in the earliest phase of the pandemic, gradually reducing over time. For example, workers who are unable to socially distance at work had an odds of over 4.0 compared with those who did not work with others during their first time period, reducing to below 1.50 in the second time period, and between 1.00 and 1.25, depending on vaccination status in the third time period. Reductions in odds followed a similar pattern for workers who worked mainly indoors.6 Similar trends in decreasing risk across occupational groups among Danish labour force participants have also been reported.22 Our results are consistent with these studies as we observed a decline in risk ratios across work exposures across the first three time periods we examined, with continued reductions into the fourth time period for location of work. However, we also observed increases in risk across all occupational exposures in the fifth period of observation, which coincided with the Omicron COVID-19 wave. This time period was not included in the study by Rhodes et al,4 and only partially included in the study by van der Feltz et al.6
Our study has strengths and limitations that should be considered when interpreting our results. Strengths include data sources that allowed us to make two important improvements on previous analyses examining work exposures and risk of COVID-19. First, we were able to restrict our population at risk of WR-C19 infection to only those labour market participants who were spending the majority of their week outside of the home. This addressed limitations of previous work where the entire labour force, including the non-trivial proportion of participants who were working remotely, were considered at risk of WR-C19 infection. Second, we were able to restrict COVID-19 cases to only those which were definitely work related, through the use of workers’ compensation claims. Based on our study, trends in rates of work-related and overall COVID-19 infections differed across waves of the pandemic. While this divergence in rates could reflect under-reporting to the WSIB, we note that this divergence between overall cases and work-related cases has also been observed on Ontario using workplace outbreaks as a source of WR-C19 infections, with the proportion of COVID-19 infections due to workplace outbreaks reducing from 23% of all cases between April and August 2020, to 11% of cases in September to December 2020 and January to March 2021.9 Workplace outbreaks are collected and assessed independently of workers’ compensation claims by the Ontario Public Health Authority (Public Health Ontario). The divergence between WR-C19 cases and overall COVID-19 cases is unlikely due to increases in denial of workers’ compensation claims for COVID-19, which were highest in the first wave of the pandemic. Conversely, this divergence may be due to the high level of workplace infection control activities in Canadian workplaces as measured in late 2020.23 Importantly, had we used overall infections in this study, given the changing proportion of WR-C19 among the overall cases in the working population, this may have led to bias in the estimates for occupational exposures across specific waves of the pandemic, if non-WR-C19 infections were not randomly distributed across occupational exposure groups.
Limitations of our study include the use of a JEM which was developed using information collected prior to the COVID-19 pandemic. As such, we do not know if occupational exposures, such as proximity to other workers, reflected the situation during the COVID-19 pandemic, where infection control strategies such as staggered shift schedules and physical distancing may have been in place. In addition, occupational exposures do not take into account specific mitigation strategies, which are likely to differ across workplace size and industries.23 The use of workers’ compensation claims, while allowing us to identify COVID-19 cases that were definitely work related, likely do not capture all WR-C19 infections, in the same way as workers’ compensation claims in general do not capture all work-related injuries and illnesses.24 We did examine if proportions of denied versus accepted claims different across occupational exposure groups or for workers in the healthcare industry. We observed limited variation in the proportion of claims which were denied, and no clear relationship with increasing occupational exposure and denial rates (results not shown but available from authors on request). It should also be noted that the capture of WR-C19 infections, in particular asymptomatic infections may differ across industry groups, given increased rates of testing in industries such as healthcare.10 Finally, we were unable to adjust our models for factors such as race and immigrant status, which likely differ across occupational exposures, but are not captured in workers’ compensation claim data.25
In conclusion, we observed that occupational exposures such as public-facing occupations, proximity to other workers and work location were independently associated with risk of WR-C19 infections in Ontario, after adjustment for age, sex, wave of the pandemic and working in the healthcare industry. However, while the direction of risk was consistent, the magnitude of the risk differed across waves of the pandemic. We also observed that, in Ontario at least, the proportion of WR-C19 infections changed across waves of the pandemic, which has potential implications for the use of overall COVID-19 cases to understand risk associated with occupational exposures. Preparation for future pandemics should consider more accurate reporting of WR-C19 infections and the potential dynamic nature of occupational exposures.
Data availability statement
Data may be obtained from a third party and are not publicly available.
Ethics statements
Patient consent for publication
Ethics approval
This study involves human participants and this study was approved by the University of Toronto, Health Sciences Research Ethics Board (Protocol# 39267). This study used administrative data from workers’ compensation.
Supplementary materials
Supplementary Data
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
Footnotes
Contributors PMS, FS and CM developed the study idea. QL performed the analysis with assistance from VL, FS and PMS. All authors discussed interpretation of the study findings. PMS wrote the first draft of the manuscript. All authors provided substantive comments and suggestions. All authors have reviewed the final version of the manuscript and have approved it for submission. PS is guarantor for the paper.
Funding This research did not receive any specific grant from funding agencies in the public, commercial or not-for-profit sectors. All authors worked for the Institute for Work and Health while this project was completed. The Institute for Work and Health is supported through funding from the Ontario Ministry of Labour, Immigration, Training and Skills Development (MLITSD).
Disclaimer The analyses, conclusions, opinions and statements expressed herein are solely those of the authors and do not reflect those of the MLTSD; no endorsement is intended or should be inferred.
Competing interests PMS is an editor at Occupational and Environmental Medicine. He was not involved in any part of the peer-review process for this manuscript.
Provenance and peer review Not commissioned; externally peer reviewed.
Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.