Association between no-shows to scheduled clinic appointments and 30-day risk of overdose in patients prescribed methadone for opioid use disorder

Abstract

Background The introduction of synthetic opiates and non-opiate sedatives into the illicit drug market has increased overdose risk for individuals who use opiates and other drugs. The ongoing risk of overdose for patients receiving methadone as a medication for opioid use disorder in the context of this more potent and less predictable drug supply is not well characterized. Additionally, little research has explored whether commonly available clinical data (including data available even in low resource settings) can predict near-term acute overdose in patients prescribed methadone for opioid use disorder. Objective To determine whether the number of recent no-shows to scheduled clinic appointments in the past 30 days is associated with 30-day overdose risk among patients enrolled in one Opioid Treatment Program who are prescribed methadone for opioid use disorder. Methods We analyzed clinical records from 1,049 patients in an opioid treatment program (May 2020–April 2024), and for each patient-day, counted the number of no-shows to scheduled clinic appointments (not methadone administrations) in the previous 30 days. Associations between the number of standardized no-shows in the past 30 days and overdose in the subsequent 30 days were analyzed with logistic regression via generalized linear model controlling for temporal and patient-specific variables. Goodness of fit was assessed with marginal R 2 and a simulation-based approach designed for multilevel models. Results The sample included 56 overdoses with an average of 0.919 no-shows in the last 30 days (std. dev 1.37). The z-standardized number of no-shows to scheduled appointments in the past 30 days was both statistically and clinically significantly associated with risk of overdose in the next 30 days adjusting for study month and season (odds ratio 1.18 [95% CI 1.13–1.23] P < 0.001), as well as adjusting for demographics and overdoses during study period (odds ratio 1.28 [95% CI 1.22–1.34] P < 0.001), and a marginal R 2 of 0.04. Model diagnostics revealed adequate fit using a generalized additive model, with results virtually unchanged from the generalized linear model. Conclusion No-shows to scheduled clinic appointments in the past 30 days are significantly associated with overdose risk in the next 30 days with a linear relationship among patients receiving methadone in a single opioid treatment program. Population-specific acute risk prediction tools could help clinicians prioritize resources for timely intervention.