Impact of Adjustment for Differential Testing by Age and Sex on Apparent Epidemiology of SARS-CoV-2 Infection in Ontario, Canada

Background Surveillance of communicable diseases typically relies on case counts for estimates of risk, and counts can be strongly influenced by testing rates. In the Canadian province of Ontario, testing rates varied markedly by age, sex, geography and time over the course of the SARS-CoV-2 pandemic. We applied a standardization-based approach to test-adjustment to better understand pandemic dynamics from 2020 to 2022, and to better understand when test-adjustment is necessary for accurate estimation of risk. Methods SARS-CoV-2 case counts by age, sex, public health unit and week were obtained from Ontario’s Case and Contact Management system (CCM), which includes all SARS-CoV-2 cases from March 2020 to August 2022. Complete data on testing volumes was obtained from the Ontario Laboratory Information System (OLIS). Case counts were adjusted for under-testing using a previously published standardization-based approach that estimates case numbers that would have been expected if the entire population was tested at the same rate as most-tested age and sex groups. Logistic regression was used to identify threshold testing rates beyond which test-adjustment was unnecessary. Results Testing rates varied markedly by age, sex, public health unit and pandemic wave. After adjustment for under-testing, overall case counts increased threefold. Adjusted epidemic curves suggested, in contrast to reported case counts, that the first two pandemic waves were equivalent in size, and that there were three distinct pandemic waves in 2022, due to the emergence of Omicron variants. Under-reporting was greatest in children and young males, and varied significantly across public health units, with variation explained partly by testing rates and prevalence of multigenerational households. Test adjustment resulted in little change in the epidemic curve during pandemic waves when testing rates were highest; we found that test-adjustment did not increase case counts once weekly per capita testing rates exceeded 6.3%. Conclusions Standardization-based adjustment for differential testing by age and sex, and for dynamic changes in testing over time, results in a different picture of infection risk during the SARS-CoV-2 pandemic in Ontario; test-adjusted epidemic curves are concordant with observed patterns of mortality during the pandemic and have face validity. This methodology offers an alternative to sero-epidemiology for identification of true burden of infection when reinfection, sero-reversion, and non-specificity of serological assays make sero-epidemiology challenging.