Mapping the intersection of demographics, behavior, and government response to the COVID-19 pandemic: an observational cohort study

Abstract

Background: During the early phase of the COVID-19 pandemic, the province of Ontario enacted restrictions and recommendations that changed over time. These measures were effective in reducing COVID-19-related illness and deaths, but adherence to these non-pharmaceutical interventions may be modified by individual factors including demographics and health status which shape exposure risk behaviors.

Methods: A total of 348 participants completed baseline questionnaires (to assess demographics, pre-pandemic exposure risk, and health status), weekly illness reports, and monthly social distancing behavior questionnaires to evaluate exposure risk over time in response to changing levels of government restrictions. Exposure risk behaviors were calculated using seven categories: attendance at social events, receiving care (hospital, etc.), visiting or volunteering at care facilities, public transportation use, hours working outside of the home, hours volunteering outside of the home, and handwashing frequency. The impact of individual and environmental factors on exposure risk over time was evaluated by a Poisson family generalized linear mixed model.

Results: Participants across all age groups and health statuses adapted their behaviors in response to evolving regulations, but older individuals and those with pre-existing conditions had the largest change in behavior. These individuals also had the most severe symptoms when they developed COVID-19 or other influenza-like illnesses. Participants who were older or had pre-existing health conditions had lower levels of exposure risk overall, and this was largely driven by a lower prevalence and frequency of in-person work. Female participants also had lower levels of exposure risk overall, consistent with an increased frequency of handwashing in this group. Unexpectedly, we found no effect of vaccination on total exposure risk.

Conclusions: Participant behavior was generally responsive to government-imposed restrictions, with increased stringency coinciding with decreased exposure risk among participants. Demographic-associated differences in exposure risk behaviors appear to be driven by systemic factors (i.e., a return to in-person work) to a greater extent than personal choices (i.e., social gatherings). These findings emphasize the interplay between demographic factors and government interventions in shaping individual behaviors over the course of the pandemic. Understanding these dynamics is crucial for informing interventions and mitigating the impact of future pandemics.

Keywords: COVID-19; Exposure risk; Social distancing.

Fig. 1

Age and health status impact the severity of reported illnesses. a Proportion of illness reports categorized by illness type and b proportion of age groups reporting each illness type. Both c increased age and d pre-existing conditions were associated with greater illness severity and e–f days of missed activities. Significance assessed by mixed linear model (Kenward-Rogers D.F. and Tukey-adjusted post hoc multiple comparisons) with age group or health status as a fixed effect and participant ID as a random effect

Fig. 2

Pandemic declaration was associated with decreased exposure risk. Exposure risk post-pandemic declaration was decreased compared to pre-pandemic declaration (a) across all age groups and b the magnitude of this decrease was greatest in younger individuals (≤ 39 vs. 40–59, p = 0.025; ≤ 39 vs. ≥ 60, p < 0.0001) and least in older individuals (40–59 vs. ≥ 60, p < 0.0001). c Exposure risk was also decreased post-pandemic declaration in individuals with pre-existing conditions and d the magnitude of this decrease was greater in healthy individuals (p = 0.0009). a, c Significance assessed by mixed linear model (Kenward-Rogers D.F. and Tukey-adjusted post hoc multiple comparisons) with timepoint (pre/post) and age group or health status as a fixed effect and participant ID as a random effect. b, d Significance assessed by linear model (Tukey-adjusted post hoc multiple comparisons) with timepoint (pre/post) and age group or health status as a fixed effect

Fig. 3

Government restrictions and other factors impacting exposure risk over the course of the pandemic. a Monthly mean Oxford Stringency Index (blue) plotted in relation to monthly mean participant exposure risk (red) (error represented as SEM) demonstrating exposure risk decreases as government restrictions increase. Shaded areas indicate when provincial “state of emergency” orders were in place (when only essential travel and essential services were allowed to remain open). Dashed lines indicate noteworthy events in the province of Ontario during this period. b Estimated effects (mean ± SE) of demographic and pandemic-related factors on exposure risk, as determined by a Poisson family generalized linear mixed model. All fixed effects and interactions included in the model are shown. Additionally, the model included participant ID as a random intercept, and pandemic week as a random slope within participant ID. Estimates represent the change in exposure risk score per unit increase in each predictor variable