Riverine suicidal behaviour: analysis of CCTV logs, search & rescue, and footfall data

Introduction: There has been a rise in incidents of suicidal behaviour along a stretch of river and three bridges which cross it. Public suicides can attract unwanted, potentially harmful media attention and can be distressing for those who may witness the event.
Aim: The aim of this research is to better understand riverine crisis behaviour in an urban area and identify opportunities for improved cross agency data coding.
Methods: Data from 2017-2019 were analysed including CCTV logs (n=2,466) describing incidents of suicidal behaviour (88% ‘welfare concern’, 12% ‘attempted suicide’); incident reports (n=1,285) from a search & rescue charity (52% ‘cause for concern’, 40% ‘person over railings’, 7% ‘person in water’, 1% body recoveries); and footfall data from a nearby bridge. Incidents were explored by gender, age, location, drug/alcohol usage and returning clients. Trends over time and the relationship between footfall and suicidal behaviour were investigated. Topic modelling was used to characterise incidents based on written accounts.
Results: Most incidents were males (60%) aged 16-25 (25%). Across years, there was no identifiable pattern across seasons or months. For all years combined, incidents were highest in summer months, peaking at the weekend, in the early hours of Sunday morning. The relationship between incidents and footfall around one bridge was not significant. Topic modelling was used to separate incidents into 4 different groups. “Middle-aged males”: predominantly males (77%), most common age range 36-45 (15%), varying severity of incidents. “Young males”: mostly male (75%), ages 26-35 (13%), roughly half incidents ‘cause for concern’ or ‘person over railings’, highest proportion of alcohol use (13%). Group 3 “females”: mostly female (60%), majority ‘person over railings’ (61%), ages 36-45 (22%), largest proportion of returning clients (23%). Group 4 “male youth”: mostly male (58%), ages 16-25 (13%), majority ‘cause for concern’ incidents (58%), highest proportion of body recoveries (8%).
Conclusion: We propose that a unified coding system for labelling incidents would be beneficial across services. Insights from this study can inform suicide prevention approaches targeting these 4 groups, which could play a critical role in preventing suicides by jumping or drowning.