Gambling venues and their association with incidence of help-seeking gambling disorder: A Bayesian spatial analysis

Introduction and background: While the association between physical accessibility to gambling venues and problematic gambling has been demonstrated in previous research, there are no published data from UK NHS-led gambling services exploring such relationships. The aim of this study was to identify associations between referrals to an NHS gambling service and local access to gambling venues, while adjusting for area characteristics. Methods: Data for referrals per Lower layer Super Output Areas (UK LSOAs) to the NHS Southern Gambling Service (SGS) were merged with large national datasets for gambling accessibility and area characteristics (deprivation, health assets accessibility, rurality). We conducted Bayesian hierarchical Poisson regression using Integrated Nested Laplace Approximation (INLA) to estimate associations between referrals and area-level predictors. Results: Self-referrals to the gambling treatment service (excluding those who gambled exclusively online) were associated with a higher number of local gambling venues (β = 0.070, 95%CrI: -0.003 to 0.143), and a shorter driving time to the nearest venue (β = -0.025, 95%CrI: -0.045 to -0.006). The residual spatial autocorrelation (φ = 0.45) was moderate. The addition of deprivation rank, access to health assets and rurality added complexity to the model but led to poorer fitting. Conclusion: Gambling venue density plus proximity were positively associated with the number of self-referrals to a large regional gambling treatment service, likely due to those venues contributing to the occurrence of help-seeking for gambling difficulties. The approaches used herein could be scaled up nationally to inform strategic planning of healthcare provision for gambling disorder, by predicting rates of likely referrals based on geographical and other contextual data.