Impact of community mask mandates on SARS-CoV-2 transmission in Ontario after adjustment for differential testing by age and sex

Abstract

Mask use for prevention of respiratory infectious disease transmission is not new but has proven controversial during the SARS-CoV-2 pandemic. In Ontario, Canada, irregular regional introduction of community mask mandates in 2020 created a quasi-experiment useful for evaluating the impact of such mandates; however, Ontario SARS-CoV-2 case counts were likely biased by testing focused on long-term care facilities and healthcare workers. We developed a regression-based method that allowed us to adjust cases for under-testing by age and gender. We evaluated mask mandate effects using count-based regression models with either unadjusted cases, or testing-adjusted case counts, as dependent variables. Models were used to estimate mask mandate effectiveness, and the fraction of SARS-CoV-2 cases, severe outcomes, and costs, averted by mask mandates. Models using unadjusted cases as dependent variables identified modest protective effects of mask mandates (range 31-42%), with variable statistical significance. Mask mandate effectiveness in models predicting test-adjusted case counts was higher, ranging from 49% (95% CI 44-53%) to 76% (95% CI 57-86%). The prevented fraction associated with mask mandates was 46% (95% CI 41-51%), with 290,000 clinical cases, 3,008 deaths, and loss of 29,038 quality-adjusted life years averted from 2020 June to December, representing $CDN 610 million in economic wealth. Under-testing in younger individuals biases estimates of SARS-CoV-2 infection risk and obscures the impact of public health preventive measures. After adjustment for under-testing, mask mandates emerged as highly effective. Community masking saved substantial numbers of lives, and prevented economic costs, during the SARS-CoV-2 pandemic in Ontario, Canada.

Fig. 1.

Trends in testing for SARS-CoV-2 in Ontario, Canada, 2020. Weekly per capita tests are presented on the Y-axis in (A) and (B). In A), test report date is presented on the X-axis. Testing increased at an average rate of 4.3% (95% CI 4.2 to 4.4%) from March to December 2020. B) presents weekly per capita tests by age group (X-axis) and sex. Females testing rates are plotted in the upper black curve; the lower blue curve represents testing in males.

Fig. 2.

Ontario SARS-CoV-2 epidemic curve in 2020 with and without test adjustment. Weekly case counts are presented on the Y-axis; test report dates are presented on the X-axis. The lower red curve shows reported case counts without adjustment for under-testing. The upper blue curve shows expected case counts if all age and sex groups were tested with the same intensity as females aged 80 and over; the gray shaded area represents upper and lower confidence bounds for test-adjusted cases, estimated as described in the text.

Fig. 3.

Timing of indoor mask mandates in Ontario, Canada, 2020. Inset shows the entire province, while main body of the map is restricted to southern Ontario. Intensity of color represents the lag since introduction of indoor mask mandates in the Wellington–Dufferin–Guelph public health unit on 2020 June 12. Northwestern health unit (inset, left side of map) and the Chatham–Kent health unit were the last public health units to introduce mask mandates, with Chatham–Kent's mandate introduced on 2020 September 14.

Fig. 4.

Model-based estimation of mask mandate impact. Test-adjusted weekly case counts (circles) were used to fit a negative binomial regression model that included mask mandate effects (predictions presented in the lower blue curve; dashed curves are 95% confidence intervals). Predictions from a negative binomial model with mask mandate effects set to zero, but all other covariates identical, is shown as an upper orange curve (dashed lines are 95% confidence intervals). The gap between the two modeled curves is the estimated fraction of cases prevented by indoor mask mandates.