Importance: Optimizing the public health response to reduce coronavirus disease 2019 (COVID-19) burden necessitates characterizing population-level heterogeneity of COVID-19 risks. However, heterogeneity in severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) testing may introduce biased estimates depending on analytic design. Objective: Characterizing individual, environmental, and social determinants of SARSCoV-2 testing and COVID-19 diagnosis. Design: We conducted cross-sectional analyses among 14.7 million people comparing individual, environmental, and social determinants among individuals who were tested versus not yet tested. Among those diagnosed, we used three analytic designs to compare predictors of: 1) individuals testing positive versus negative; 2) symptomatic individuals testing positive versus negative; and 3) individuals testing positive versus individuals not testing positive (i.e. testing negative or not being tested). Analyses included tests conducted between March 1 and June 20, 2020. Setting: Ontario, Canada. Participants: All individuals with ≥1 healthcare system contact since March 2012, excluding individuals deceased before, or born after, March 1, 2020, or residing in a long-term care facility. Exposures: Individual-level characteristics (age, sex, underlying health conditions, prior healthcare use), area-based environmental (air pollution) exposures, and area-based social determinants of health (income, education, housing, marital status, race/ethnicity, and recent immigration). Main Outcomes and Measures: Odds of SARS-CoV-2 test, and of COVID-19 diagnosis. Results: Of a total of 14,695,579 individuals, 758,691 had been tested, of whom 25,030 (3.3%) tested positive. The further the odds of testing from the null, the more variability observed in the odds of diagnosis across analytic design, particularly among individual factors. There was less variability in testing by social determinants across analytic design. Residing in areas with highest household density (adjusted odds ratio: 2.08; 95%CI: 1.95-1.21), lowest educational attainment (adjusted odds ratio: 1.52; 95%CI: 1.44-1.60), and highest proportion of recent immigrants (adjusted odds ratio: 1.12; 95%CI: 1.07-1.16) were consistently related to increased odds of COVID-19 across analytic designs. Conclusions and Relevance: Where testing is limited, risk factors may be better estimated using population comparators rather than test-negative comparators. Optimizing COVID-19 responses necessitates investment and sufficient coverage of structural interventions tailored to heterogeneity in social determinants of risk, including household crowding and systemic racism.
ASJC Scopus subject areas