Estimating prevalence of opiate and crack cocaine use and injecting in England using mixed-effects capture-recapture models

Abstract

Multiple social and health problems are associated with opiate and crack cocaine use and injecting, but the prevalence of these behaviours is unknown. We linked three sources of administrative data on individuals engaging in opiate and/or crack cocaine use (OCU) community treatment; arrests, prison, or probation; and drug-related deaths. We used probabilistic linkage to increase the number of matching records and capture-recapture analysis to estimate the unknown number of individuals not observed in any source. We fitted random effects models to data stratified by source, local authority (LA) area, age, sex, drug type, and injecting. Two-way interactions between sources and explanatory variables were included as fixed effects. We included random effects at the LA level, with random intercepts and random coefficients. The total count of opiate and/ or crack cocaine use in the financial year 2018/19 was estimated at 336,531 [95% confidence interval (CI): 311,932-360,845], corresponding to 9.44 (95% CI: 8.75-10.12) per 1,000 population in England. Three-quarters were men, and the estimate of people currently injecting drugs was 67,715 (95% CI: 64,769-70,526). Our study used random effects models to borrow strength across local authorities and produce robust prevalence estimates of OCU for England