Mental health and socio-cognitive predictors of adherence to COVID-19 social distancing rules in adolescents in England

Abstract

The COVID-19 pandemic prompted governments worldwide to introduce social distancing measures, including school closures and restrictions on in-person socialising. However, adherence to social distancing was challenging for many - particularly adolescents, for whom social interaction is crucial for development. The current study aimed to identify individual-level influences on adherence to social distancing in a longitudinal sample of adolescents aged 11-20 years in England, who took part in a randomised controlled trial. At baseline, 460 participants completed detailed pre-pandemic assessments, including mental health and well-being, altruism, delayed reward discounting, rejection sensitivity, prosociality and susceptibility to prosocial and anti-social influence. Of these, 205 participants reported their compliance with COVID-19 social distancing rules and attendance at social gatherings between June and August 2020. Bayesian ordinal regression models were used to predict adherence to social distancing from predictors, controlling for age at pandemic, gender, day of assessment, and intervention group. The results indicated that higher levels of prosociality, altruism and lower susceptibility to anti-social influence were associated with higher adherence to social distancing. Pre-pandemic levels of depression, anxiety, delayed reward discounting, rejection sensitivity, conduct problems and emotional awareness were not robustly related to the outcomes. These findings have implications for understanding how adolescents comply with public health guidelines, highlighting the role of social influence and peer norms.

Fig. 1

Response distribution of each self-reported social distancing outcome variable.

Fig. 2

Ordinal regression coefficients for predicting social distancing compliance (Panel A) and attendance at social gatherings (Panel B). Note. Each panel shows the posterior probability distributions for ordinal regression coefficients relating each explanatory variable to the primary and secondary outcomes. Regression coefficients are adjusted for four covariates (age at Time 4, day of assessment, gender, and intervention group). Points indicate the distributions' median, representing each regression coefficient's central estimate. Thick and thin bars around each point indicate the 95 % and 99.5 % credible intervals, respectively. The posterior probability of each regression coefficient being small (between −.1 and .1), small-to-medium (between .1 and .3 or between −.1 and −.3), and medium-to-large (greater than .3 or less than −.3) are given in the plot. As described in Method, each effect is estimated from a separate regression model controlling for age at Time 4, day of assessment, gender and intervention group, and not from a single combined regression model with all 11 predictors in the same model.