Risk of SARS-CoV-2 among Ontario Healthcare Workers: A comparison of test-negative and cohort study designs in a heterogeneous reason-for-testing population

Objective: Differential reason-for-testing may bias test-negative estimates. This study aimed to estimate healthcare worker SARS-CoV-2 risk, with adjustment for healthcare seeking behaviour and unmeasured reason-for-testing. Methods: 1.2-million workers in Ontario, Canada were followed for SARS-CoV-2 PCR tests from February 2020 to December 2021. Hazard ratios (HR) were used to estimate SARS-CoV-2 risk in healthcare workers versus non-healthcare workers based on the overall and test-negative sub-cohort. Unmeasured reason-for-testing was examined through probabilistic-bias-analyses. Results: In the test-negative sub-cohort, healthcare and non-healthcare workers had similar risk of SARS-CoV-2. However, healthcare workers had an increased risk in the symptomatic-adjusted (HR: 1.15, 95% CI 1.03-1.40) and asymptomatic-adjusted (HR: 2.83, 95% CI: 1.08-9.07) models. Conclusions: Future test-negative studies should account for potential bias from varying symptomatic and asymptomatic testing groups and may consider using probabilistic-bias-analysis methods when reason-for-testing data is missing.