Predicting population-level need for early treatment programmes in first episode psychosis: the Advanced Psychiatric MAPping Translated into Innovations for Care [PsyMaptic-A] study

Poster C68, Wednesday, October%2010, 11:30%20am%20-%201:00%20pm, Essex%20Ballroom

James Kirkbride1, Tao Ding2, Rebecca Dliwayo1, Pia Wohland3, David Osborn1, Paul French4, Gianluca Baio2, Peter Jones5; 1Division of Psychiatry, UCL, London, UK, 2Department of Statistical Science, UCL, London, UK, 3School of Earth and Environmental Sciences, University of Queensland, Brisbane, Australia, 4Greater Manchester West Mental Health NHS Foundation Trust, Manchester, UK, 5Department of Psychiatry, University of Cambridge, Cambridge, UK

Introduction Early treatment programmes for first episode psychosis [FEP] benefit patients, but require evidence-based estimates of need in the population to effectively plan services. We developed new population-level prediction models using small-area epidemiological data. Methods Data from three major FEP studies (AESOP, ELFEP, SEPEA) were combined in various Bayesian hierarchical Poisson models to estimate risk, by age, sex, their interaction, ethnicity, cannabis use and small-area deprivation, population density and social fragmentation. Informative priors were included based on systematic reviews. Model 1 contained age group, sex and ethnicity. Model 2 also included neighbourhood-level deprivation and social fragmentation. Model 3 substituted social fragmentation for population density. Posterior mean and 95% credible interval (95%CrI) were used to present predicted counts and incidence rates for the 2016 population in England. These were compared with observed national FEP data from the NHS. Analyses were performed using INLA in R. Results In 2016, 8,280 people with FEP were in contact with secondary mental healthcare services in England. All model predictions were larger than this figure, ranging from 11,326 (95%CrI: 7,731-18,771) in Model 3 to 16,252 (95%CrI: 11,076-23,920). Predicted incidence rates from Models 2 (36.4 per 100,000 person-years; 95%CrI: 25.3-55.2) and 3 (32.1; 95%CrI: 21.9-53.2) were consistent with the observed pooled rate in England from a meta-analysis (31.7; 95%CI: 24.6-40.9). Conclusions Small-area epidemiological data can be used to accurately predict incidence at a national level to inform commissioning of treatment programmes for FEP. Higher predicted FEP counts than observed in England may indicate significant unmet population need.

Topic Area: Public Policy

Back to Poster Schedule