Study population and setting
The study objective was to describe high-risk occupations for COVID-19, and extracted data from government case reports in Japan (n=405), Singapore (n=149), Hong Kong (n=87), Taiwan (n=25), Thailand (n=16), and Vietnam (n=8) from January 23 to March 14, 2020. These countries were selected for: 1) their proximity to mainland China where the outbreak first occurred, 2) the report of cases due to travel in these countries beginning in mid-January, and 3) local transmission occurring in late January to early February. They identified 2002 total confirmed cases, but excluded imported cases, leading to 690 cases of local transmission included (predominantly from Japan).
Summary of Main Findings
The investigators found that 103 of 690 (15%) total local transmission cases were possibly due to work-related transmission, and of those, 22% (n=23) had healthcare worker occupations. Other occupations represented among these work-related transmission cases were drivers and transport workers (n=19, 18%), services and sales (n=19, 18%), domestic workers (n=9, 9%), and public safety workers (n=7, 7%). More incident cases occurred during later stages of the epidemic (defined as the 11th to 40th day of the 40-day time period) among health professionals, whereas shop salespersons and receptionists were particularly affected earlier in the epidemic. They also found that the number of work-related cases reported each day remained stable through the follow-up period, despite an increased number of cases of local transmission overall.
The study makes use of extant confirmed case reports, and had review by two occupational health physicians in order to determine potential work-related exposure. They conducted sensitivity analyses with and without Japan included, to compensate for the different reporting system the country uses and its large representation in the data. The use of a 40-day period allowed for trends to also be explored.
The study relied on extant reports that may not have had pertinent details available (e.g., missing occupational history and other potential sociodemographic confounders) and therefore, there may be an underreporting of work-related cases, or confounding due to the overrepresentation of certain sociodemographic groups in certain occupations and other concerns of selection bias. There also may have been variation in reporting both between and within countries as guidelines changed during the 40-day time period, which would also result in non-differential information bias. The study identified a range of occupational groups that they determined were at “high risk,” however they did not statistically assess whether these workers were at increased risk of infection compared to non-workers. Additionally, they did not report how many people within each profession then became cases which did not allow for risk comparisons between professions, and overall their absolute numbers remain low in the occupational groups, which likely limited their power to detect such an association.
This study is one of the few comparisons of multiple industries in an occupational health assessment of COVID-19, and uses confirmed cases from extant reports to show this risk in multiple countries.