UCSF

Individual behaviors are critical for preventing the spread of coronavirus disease 2019 (COVID-19) infection. Given that both protective and risky behaviors influence risk of infection, it is critical that we understand how such behaviors cluster together and in whom. Using a data-driven approach, we identified clusters of COVID-19-related protective and risky behaviors and examined associations with socio-demographic, pandemic, and mental health factors. Data came from a cross-sectional online U.S. nationwide study of 832 adults with high levels of pre-pandemic trauma. Latent class analysis was performed with ten protective (e.g., washing hands, wearing masks) and eight risky (e.g., attending indoor restaurants, taking a flight) behaviors for COVID-19. Then, we examined distributions of socio-demographic and pandemic factors across behavior classes using ANOVA or Chi-square tests, and associations between mental health factors (depressive, anxiety, posttraumatic stress symptoms) and behavior classes using multinomial logistic regression. We identified four classes, including three classes with relatively low risky but high (28.8%), moderate (33.5%) and minimal (25.5%) protective behaviors and one high risky behaviors class with associated moderate protective behaviors (12.1%). Age, sexual orientation, political preference, and most pandemic factors differed significantly across behavior classes. Anxiety and posttraumatic stress symptoms, but not depression, were higher in the High Risk, but also Highly and Moderately Protective classes, relative to Minimally Protective. Prevention and intervention efforts should examine constellations of protective and risky behaviors to comprehensively understand risk, and consider current anxiety and posttraumatic stress symptoms as potential risk indicators.