The Mask-Wearing Bias In The Estimates Of Vaccine Efficacy

Abstract

In the United States mask wearing is positively correlated with vaccine acceptance. This correlation introduces an important bias into real-world estimates of vaccine efficacy. I derive the formulae for vaccine efficacy that correct for this phenomenon and show that such biases explain some of the differences between higher estimates of vaccine efficacy observed in the US studies, on one hand, and lower estimates from Israel and Pfizer trials, on the other hand. Control for such biases is important for currently-debated public health decisions regarding COVID19 vaccine booster doses.

Effectiveness of COVID-19 vaccines against Omicron or Delta symptomatic infection and severe outcomes

“…For example, if vaccinated individuals have more exposure to SARS-CoV-2, our VE estimates are likely underestimated. 21 Last, changes in testing patterns, including increased use of rapid antigen tests (which are not captured in our data) and decreased PCR testing availability, may have impacted our estimates, but the direction of any resulting bias is uncertain.…”

“…Finally, a study from South Africa estimated VE against infection at 33% in the Omicron period compared to 77% in the pre-Omicron period. 19 Direct comparisons to other jurisdictions are challenging 20 due to differences in study methodology, outcome definitions (i.e., symptomatic infection vs. any infection), vaccination policies (i.e., homologous vs. heterologous vaccine schedules, third dose eligibility criteria, product-specific policies), population age structures, and public health measures that were in place during the study period (e.g., vaccine certificates, mask mandates 21 ). Despite this, the general trends across the studies are similar, demonstrating substantially lower VE against Omicron infection than for previous SARS-CoV-2 variants.…”

“…Although our 2-dose estimates against severe Omicron-related outcomes were challenging to compare given instability and wide confidence intervals, they do suggest less waning against hospitalization than has been observed elsewhere and this finding is aligned with surveillance data in Ontario. 39 40 Direct comparisons to other jurisdictions are challenging 41 due to differences in study methodology, outcome definitions (i.e., identification of Omicron and Delta based on laboratory confirmation vs. time-based criteria), intervals following latest dose selected to monitor VE, vaccination policies (i.e., homologous vs. heterologous vaccine schedules, third dose eligibility criteria, product-specific policies [use of mRNA vs. viral vector vaccines, a preferential recommendation in Ontario of BNT162b2 for young adults]), 42 43 population age structures, public health measures implemented during the study period (e.g., vaccine certificates, mask mandates 44 ), testing patterns, and use of antivirals or other therapies. Further, Ontario has experienced a lower cumulative incidence of reported infections and has attained higher vaccine coverage than other countries that have estimated VE against Omicron to date, 45 and thus has a potentially dissimilar distribution of infection-induced versus vaccine-induced immunity.…”

Effectiveness of COVID-19 Vaccines Over Time Prior to Omicron Emergence in Ontario, Canada: Test-Negative Design Study

“…If these individuals subsequently acquired breakthrough infections, VE at shorter intervals since second dose would be low, which might underestimate the extent of waning VE. Similarly, if vaccinated individuals were more likely to wear masks and adhere to public health guidelines, this may overestimate VE and make waning immunity appear minimal [ 37 ]. Second, we were unable to identify some high-risk individuals who were prioritized to receive their primary doses (eg, residents of congregate settings other than long-term care homes, health care workers, and caregivers).…”

Weak or No Correlation Between Recent COVID-19 Data and Vaccination Rates in France

“…In addition, the decrease of the population's vigilance (especially, those vaccinated) with regard to barrier and hygiene measures, could explain -at least partially -the growth of SARS-CoV-2 infections. A study has found in the United States that adherence to mask wearing and acceptance of vaccines are correlated, which makes the vaccine efficacy estimation from real-world data biased by a significant amount [8]. Therefore, the lack of correlation with vaccine coverage suggests that herd immunity threshold would not be reached without respecting barrier gestures.…”