This project will use linked administrative data to understand risk for mortality and other adverse outcomes during and after opioid agonist treatment (OAT). It will use standard biostatistical approaches and sophisticated machine learning techniques.
School of Public Health and Community Medicine, UNSW
Monash Addiction Research Centre, Monash University
Adelaide Medical School, University of Adelaide
North America is in the midst of an opioid use epidemic, and opioid use is also increasing dramatically in Australia. OAT is an effective treatment for opioid dependence, but there are important questions regarding risk of adverse clinical outcomes, including death, that are yet to be answered. Who is most at risk of adverse outcomes? What patient, provider and treatment setting actors may influence this risk? Answers to these and other questions are critical to inform the massive scale-up of OAT internationally that will be required to respond to the opioid epidemic.
Aim 1: Determine the magnitude of risk for specific adverse clinical outcomes (e.g. mortality, hospitalization and ED presentation, and unplanned treatment cessation) during and after OAT with methadone and buprenorphine
Aim 2: Identify patient, treatment setting, and provider risk factors associated with adverse clinical outcomes during and after OAT with methadone and buprenorphine
Aim 3: Develop a risk prediction model to identify patients at greatest risk of adverse clinical outcomes during and after OAT
The study will use a population cohort of OAT patients (n≈45,000) treated between 2001 (when buprenorphine became available for OAT in New South Wales) and 2016. These data will be probabilistically linked63 to State-wide hospitalization, emergency department, incarceration and mortality data. Linkage will be undertaken by dedicated data linkage institutions and data custodians using best practice protocols that protect individual privacy and confidentiality, with extensive clerical review to maximize linkage sensitivity and specificity. We will examine incidence (Aim 1) and risk (Aim 2) for specific adverse clinical outcomes during OAT, with a special focus on the period of OAT induction, as well as the remainder of time in OAT and the 4 weeks immediately following cessation of OAT. Adverse clinical outcomes to be examined will include all-cause and cause-specific (drug, self-harm/suicide, and injury-related) emergency department visits, hospitalisation and mortality and unplanned treatment cessation. Then, we will develop a risk prediction model to identify patients at greatest risk of adverse outcomes during OAT (Aim 3). While Aims 1 and 2 are focused on understanding the magnitude of risk for an outcome associated with a specific factor, Aim 3 is focused on maximizing the predictive ability of the model to enable the real-time identification of individual patients at risk of adverse clinical outcomes during OAT.
All data for this project has been linked and we are currently in the process of undertaking the various analyses outlined in the objectives. The results from the project so far are summarised below.
Larney S; Hickman M; Fiellin DA; Dobbins T; Nielsen S; Jones NR; Mattick RP; Ali R; Degenhardt L, 2018, 'Using routinely collected data to understand and predict adverse outcomes in opioid agonist treatment: Protocol for the Opioid Agonist Treatment Safety (OATS) Study', BMJ Open, vol. 8, http://dx.doi.org/10.1136/bmjopen-2018-025204
Jones NR; Shanahan M; Dobbins T; Degenhardt L; Montebello M; Gisev N; Larney S, 2019, 'Reductions in emergency department presentations associated with opioid agonist treatment vary by geographic location: A retrospective study in New South Wales, Australia', Drug and Alcohol Review, vol. 38, pp. 690 - 698, http://dx.doi.org/10.1111/dar.12976