Statistical Analysis

To achieve the objectives of this study, descriptive statistics were used to characterize the 982 individuals who received methadone within prison over the entire observation period (Table 1). More extensive analyses were conducted, however, for the subset of 645 individuals who received methadone within prison and were released to the community (Table 2). Time from prison release to linkage to community methadone was first explored using Kaplan-Meier estimates and is presented as a survival curve (Figure 1). Cox proportional hazards models were then used to estimate the effect of methadone receipt at release, on time to linkage to methadone in the community after release (Table 3). We present results of both univariate and multivariate hazard models. We utilized a backward elimination strategy to build the multivariate model, and variables were retained if they demonstrated statistical significance (p<0.05). Follow-up time was censored at the earliest of the following: date of linkage to methadone in the community after release, date of death, or January 30, 2018. Data analyses were conducted using SPSS statistical software, version 25.

Kaplan-Meier curves of linkage to methadone in community after release (n = 638)*

* Excluded individuals with missing dependent variable values; MMT (methadone maintenance treatment); ^a on MMT at the time of release.

Characteristics of all people who were prescribed methadone during incarceration (N = 982).

Characteristics of those released from prison who were prescribed methadone during incarceration (n = 645).

Cox regression hazard ratios for time to linkage to methadone maintenance treatment in community after release (n = 645).

Legend: 95% CI: 95% confidence interval; MMT: methadone maintenance treatment.