Non-linear association of induction dose with early dropout from methadone maintenance treatment: A retrospective cohort study.

BACKGROUND: Conservative induction dose in methadone maintenance treatment (MMT) may lead to early dropout. Existing evidence mostly focuses on linear associations, leaving non-linear dose-response relationships and critical inflection points unexplored. METHODS: This retrospective cohort study analyzed 5083 first-time patients across 11 MMT clinics in Guangdong Province (2006-2021). The mean methadone dose in the first week defined the induction dose, and dropout within 30 days was the outcome. Restricted Cubic Spline (RCS) and threshold analysis explored non-linear and inflection points. RESULTS: After confounder adjustment, a significant negative association existed between dose and dropout (OR=0.900, 95% CI: 0.876-0.924), confirmed with weighting. A non-linear relationship was identified (Pnon-linear <0.001) with an inflection point at 40.71mg/day. Below this point, each 5mg/day dose increase reduced dropout risk by 17% (OR=0.83, 95% CI: 0.78-0.87). Above 40.71mg/day, the risk reduction was 6% (OR=0.94, 95% CI: 0.91-0.98). Sensitivity analysis (excluding 48 high doses) lowered the inflection point to 33.57mg/day. Subgroup analysis indicated interactions of the induction dose effect with both drug-use frequencies (P = 0.043) and the year entering MMT (P = 0.027). Those with higher drug frequency showed a smaller reduction in dropout risk after increasing the induction dose. CONCLUSION: The study identified a non-linear association between induction dose and early dropout, and there is a threshold effect inflection point. Based on current guidelines, there is a need to appropriately increase the starting dose of MMT, with further consideration of methadone safety and individual differences required in clinical practice.