Beyond one-size-fits-all: Latent patient classes at methadone treatment entry and their clinical implications - Evidence from a large real-world cohort.

BACKGROUND: Opioid use disorder is characterised by substantial interindividual variability shaped by individual trajectories and social determinants of health, yet medication-assisted treatment continues to apply uniform clinical protocols across a heterogeneous population. This study aimed to identify latent patient profiles at treatment admission, examine their association with clinical and treatment characteristics, and assess their predictive utility for treatment dropout. METHODS: Using a retrospective observational design, records from 12,999 OUD patients initiating methadone treatment between 2015 and 2023 were analysed. Latent class analysis was applied, followed by bivariate analyses and Cox proportional hazards models to examine class-level differences in outcomes. RESULTS: Four clinically distinct profiles emerged: Socially Stable High-Frequency Users (65.6%), Vulnerable Mothers with Late-Onset (8.8%), Justice-Referred Early-Onset Users (13.2%), and Stable Health-Referred Low-Frequency Users (12.4%). Classes differed significantly in methadone dosage, dispensing conditions, psychiatric comorbidity, infectious disease burden, and treatment retention. Justice-Referred Early-Onset Users exhibited the highest early dropout risk; Vulnerable Mothers with Late-Onset showed a risk of early dropout that did not differ significantly from that of the highest-risk group. CONCLUSIONS: These findings provide empirical grounds for moving beyond uniform methadone treatment models toward stratified, needs-based protocols. Concrete policy changes, including formalised penitentiary-to-community care continuity, revision of dose-setting criteria for justice-referred patients, and maternal-informed treatment pathways that address the specific psychiatric and social vulnerabilities of women with caregiving responsibilities, could substantially reduce attrition across the profiles with the greatest clinical complexity.