Background The COVID-19 pandemic has necessitated the examination of behavioural factors contributing to its spread. This study aimed to estimate the total causal effect of a set of behaviours on the risk of COVID-19 infection.

Methods A population-based case-cohort study was conducted in a district of Istanbul, Türkiye. Cases were recruited from all PCR-confirmed COVID-19 patients in the district. Controls were randomly selected in a 1:1 ratio from all GP-registered residents in the district. Data were collected through telephone interviews. The outcome variable was PCR-confirmed COVID-19 and exposure variables were self-reported practice behaviours in the last two weeks. Causal assumptions are elicited using a directed acyclic graph. Odds ratios (ORs) are calculated for each behaviour using Inverse Probability of Treatment Weighting (IPTW). Data analysis is programmed in Python. Statistical significance set at p<0.05.

Results The study comprised 346 participants (170 cases, 176 controls) with a mean age of 42.8±16.0, including 141 (40.8%) males and 205 (59.2%) females. Attending mass gatherings significantly increased COVID-19 risk (OR=19.4, p<0.001), as did using public transportation (OR=1.7, p=0.016) and contact with COVID-19 patients (OR=5.1, p<0.001). Mask usage reduced the risk of infection when it is worn all the time outside the home (OR=0.5, p=0.0041) and in an appropriate way (OR=0.5, p=0.016). Handwashing also decreased the infection risk (OR=0.6, p<0.001).

Conclusion The findings highlight the importance of preventive behaviours in mitigating COVID-19 transmission. The study reinforces the necessity for continued public health messaging and interventions that promote adherence to effective preventive behaviours. Also, the substantial impact of attending mass gatherings on infection rates suggests that policies aimed at regulating such events could be particularly effective in controlling outbreaks. The study's strengths include mandatory reporting of COVID-19 cases and random selection of controls. However, reliance on self-reported data was a limitation. Future research should aim to incorporate more objective measures of behaviour and explore the potential for longitudinal designs.