Using a predicting Risky Opioid-Benzodiazepine Trajectory e-Care Tool (PROTeCT) to identify high-risk regions

点击翻译按钮获取中文摘要

Prescription opioid overdose deaths have increased markedly in the past two decades, with a third of these fatalities involving concurrent benzodiazepine (BZD) use. Despite clinical guidelines and FDA black-box warnings cautioning against concurrent opioid (OPI) and BZD use (hereafter OPI-BZD use), the number of patients using OPI-BZD increased by 41% between 2002 and 2014. OPI-BZD use increases the risk of overdose and other adverse outcomes, especially among older adults. However, little is known about the thresholds of duration and dose or patterns of OPI-BZD use profiles most associated with the risk of overdose and other adverse outcomes among older adults. Prior studies have defined OPI-BZD use with arbitrary thresholds (e.g., ≥1 day overlapped supply) and focused only on duration or dose alone rather than combinations of duration and dose of concurrent use. Applying arbitrary and broad thresholds without evidence and validation to all patients imposes challenges in tailoring clinical care, and leads to ineffective therapies and interventions involving OPI- BZD use. Alternatively, advanced group-based trajectory models (GBTMs) can be used to better characterize OPI-BZD use in clinical practice. GBTMs have the ability to account for dynamic medication use, identify subgroups with similar changes over time, and simultaneously examine dose and duration thresholds or other patterns most relevant to outcomes to better aid clinical decision-making. The parent R21 aims at developing an innovative, real-time “Predicting Risky Opioid-Benzodiazepine Trajectory e-Care Tool (PROTeCT)” for efficiently identifying and predicting subgroups of older adults with distinct and potentially unsafe patterns of OPI-BZD use. Because Medicare and Medicaid enrollees are experiencing an increased number of chronic pain conditions, mental health/substance use disorders, and prescription OPI use, Medicare and Medicaid are ideal settings for developing the PROTeCT tool. Using national Medicare claims and Arizona (AZ) and Florida (FL) Medicaid data from 2013-2016, Aim 1 focuses on identifying distinct trajectories of OPI-BZD use. We will also identify predictors associated with specific trajectories or patterns. In Aim 2, we will identify the distinct trajectories or patterns of OPI-BZD use that are the most closely associated with two separate outcomes (i.e., overdoses; falls and fractures). Finally, we propose to develop a prototype of a real- time PROTeCT platform capable of prospectively and iteratively predicting patients with unsafe patterns of OPI- BZD use by prospectively analyzing more recent data using 2017-2019 AZ and FL Medicaid data. In order to help Dr. Lo-Ciganic to successfully complete this R21 after her recent critical life event, this supplement requests funds to mainly support a specialized program coordinator’s effort and include two additional investigators with health informatics expertise for Aim 3. The infrastructure of our findings and tool may be generalizable to Medicare, Medicaid programs in other states, or other healthcare data systems with similar data structures.