Protective Behaviors Against COVID-19 by Individual Vaccination Status in 12 Countries During the Pandemic

This cross-sectional study examines associations of protective behaviors against COVID-19 among individuals in 12 countries who had received 0, 1, or 2 COVID-19 vaccine doses.


Introduction
The eagerly awaited onset of vaccination programs against COVID-19, along with people's urge to return to "normal life," has prompted concerns that individuals who were vaccinated would reduce their protective behaviors faster than recommended. 1,2 Whereas worries about risk compensation 3 have proven unfounded in some preventive medicine contexts (eg, human papillomavirus vaccination 4 ), signs of riskier conduct among the treated have appeared in others (eg, individuals who use HIV preexposure prophylaxis 5 ). Based on representative samples of 12 countries in various stages of immunization programs, this study compared the self-reported protective behaviors of individuals who had received 0, 1, or 2 COVID-19 vaccine doses. We assessed the period February 23 to June 1, 2021, when, barring anecdotal exceptions, 6 governments had yet to exempt individuals who had been vaccinated from COVID-19 protective-behavior policies.

Methods
This cross-sectional study was approved by the ethical review boards of Imperial College London (ICL) and Columbia University. Data on COVID-19 protective behaviors, vaccination status, and demographic characteristics were obtained from the ICL-YouGov Global Survey. Respondents completed written informed consent documents online. This study is reported following the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline.
Nationally representative samples were collected online in 15 countries, in biweekly waves.
Countries with less than 5% of the population having received 2 doses of a COVID-19 vaccine by June 1 (Australia, Japan, and South Korea) were excluded, leaving 80 305 respondents from Canada, Denmark, France, Germany, Israel, Italy, Norway, Singapore, Spain, Sweden, the United Kingdom and the US (Table).
The ICL-YouGov Global Survey consists of repeated cross sections rather than longitudinal data, so weights from coarsened and exact matching were used to balance respondents who had not received a vaccine dose to those who had received 1 dose and respondents who had received 1 vaccine dose to those who had received 2 doses. Matching variables included calendar-week, age, sex, household size, employment status, employment in health care, trust in COVID-19 vaccines, having self-reported chronic diseases, having self-reported mental health conditions, and children in the household. We matched by country to generate a larger matched sample and by subnational region to provide more precise matching, although this consequently led to a smaller sample. The associations of vaccination status with COVID-19 protective behaviors, including physical distancing and mask wearing, were estimated using bivariate linear regressions with inverse probability of treatment weighting (eMethods in the Supplement). Protective behaviors were self-reported on a 5-point scale, with higher scores indicating more frequent behavior. Separate regression models were estimated for each dependent variable, and for each independent variable (binary indicators of first and second vaccine dose). Analyses were conducted using Stata statistical software version 14.1    compared with respondents who had not received a vaccine for country match (β = −0.02, 95% CI, −0.08 to 0.04]) or region match (β = −0.02; 95% CI, −0.08 to 0.04). There was a significant difference comparing individuals who had received 2 doses with those who had received 1 dose for country match (β = −0.08; 95% CI, −0.16 to −0.01) but not region match (β = −0.03; 95% CI, −0.13 to 0.06) (Figure). Respondents who had received 2 doses physically distanced less than respondents who had not received a vaccine according to country match (β = −0.07; 95% CI, −0.13 to <−0.01) and region match (β = −0.12; 95% CI, −0.21 to −0.02]) in an aggregated sample, matching respondents who had received 2 doses to those who had received none. In the US, there were similar differences in physical distancing among respondents who had received 2 vaccine doses vs those who had received 1 dose before (β = −0.22; 95% CI, −0.38 to −0.06]) and after (β = −0.33; 95% CI, −0.63 to −0.02]) the May 13 announcement of a vaccination-conditional protective-behavior policy (country match). There were no significant differences across groups for mask use using the aggregated sample.

Discussion
Despite occasional significant results, all of small magnitude, overall, this cross-sectional study found no substantial reduction in physical distancing or mask use associated with receipt of COVID-19 vaccine doses. This suggests that until early June, people generally did not engage in concerning levels of risk compensation as they acquired immunity. There was no discontinuity in behavior soon after US President Joe Biden announced that US adults who had been fully vaccinated need not wear masks or physically distance. We encourage investigation of the between-country variation identified in this study and reexamination of these results as time passes and as protective-behavior policies remain in place or alter in specificity and strength. Study limitations include the lack of longitudinal data and reliance on self-reported behavior.