Safety and Contagion in Inpatient Mental-Health Wards: An Estimate of the Contagion Effect From the Wardsonar Study in the UK

OBJECTIVES: It is widely accepted that safety incidents involving violence and self-harm in inpatient mental health wards occur in spatio-temporal clusters due to behavioral contagion. This study aims to explicitly measure the duration for which such an incident affects the probability of future incidents over time.

METHODS: We collected routine data on incident reports, staffing, and patient casemix from six wards in two NHS hospitals over a five-month period in 2022. This data was used to create a longitudinal panel dataset with time measured in hours. We applied Fixed Effect logistic regression to daytime hours (8am – 10pm) with lagged outcomes to quantify the impact of incidents that occurred up to five hours earlier on the probability of a current incident.

RESULTS: The analysis revealed significant variation in incident rates across wards, with no relationship found between daytime hours and the probability of an incident. Incidents that occurred up to four four hours prior had a statistically significant impact on the likelihood of a current incident, with an odds ratio of 1.49. Due to incomplete patient characteristic data across all wards, a sensitivity analysis was conducted on the wards with complete data, which did not change the substantive results.

CONCLUSIONS: By using real-world evidence from six wards over a five-month period, we believe this is the first study to explicitly measure the duration of behavioral contagion. Our findings suggest that the contagion effect is both practically and statistically significant, lasting up to four hours after the original event and increasing the odds of a subsequent incident by approximately 50%. This metric can inform economic evaluations aimed at assessing the cost-effectiveness of strategies designed to prevent such incidents.