A0737
Title: Multi-parameter estimation of prevalence (MPEP) models to estimate the prevalence of opioid dependence
Authors: Hayley Jones - University of Bristol (United Kingdom) [presenting]
Matthew Hickman - University of Bristol (United Kingdom)
Andreas Markoulidakis - University of Bristol (United Kingdom)
Abstract: Estimates of the number of people dependent on opioids are critical for planning public health services and evaluating interventions to reduce drug-related harms. So-called indirect methods are needed since population surveys underestimate the extent of stigmatized behaviors. The MPEP approach is developed to estimate prevalence or population size. MPEP utilizes routinely collected linked administrative data on opioid agonist therapy (OAT) prescriptions and adverse events such as opioid-related deaths or non-fatal overdoses. The known population of people with opioid dependence is identified as everyone who recently received OAT, and the size of the additional unknown population is estimated through modelling of adverse event rate data, critically including events occurring outside of the known population. In a joint model fitted in a Bayesian framework, simultaneous regressions are fitted to event rates and (latent) prevalence. Importantly, the impact of OAT on event risk is accounted for since everyone in the unknown group is not receiving OAT, by definition. Joint modelling of two or more types of adverse events allows checking the consistency of evidence. The MPEP approach, its assumptions and limitations, and the recent application are described to estimate the prevalence of opioid dependence in Scotland from 2014/15 to 2019/20.
