Revisiting the X:BOT Naltrexone Clinical Trial Using a Comprehensive Survival Analysis.

OBJECTIVES:This paper illustrates survival models for analysis of trials of substance use treatment programs. It uses public release data from a study of extended-release naltrexone (XR-NTX), relative to buprenorphine-naloxone (BUP-NX). METHODS:We used publicly available data from the X:BOT trial (n = 570), which compared XR-NTX to BUP-NX on 2 efficacy outcomes (opioid relapse, use of nonprescribed opioids; positive opioid urine test) and 1 safety outcome (overdose). Intention-to-treat (ITT) and per-protocol approaches were implemented using survival models that included treatment-by-time interactions. RESULTS:Consistent with the original trial findings, 72% of XR-NTX and 94% of BUP-NX subjects initiated treatment; the ITT hazard ratio for XR-NTX relative to BUP-NX was 1.40 (95% confidence interval: 1.13, 1.73; P < 0.01) for opioid relapse and 1.31 (1.07, 1.60; P = 0.01) for positive urine test. Using treatment-by-time interactions, we examined the time-dependent effect of XR-NTX and found an elevated ITT overdose hazard ratio of 2.4 (1.1, 5.3; P = 0.03) overall and 3.8 (1.2, 11.6; P = 0.02) during the study treatment phase. This result (28 overdoses overall; 17 overdoses during the study treatment phase) contrasts with the previous analysis, which reported minimal differences in overdose between XR-NTX and BUP-NX. CONCLUSIONS:An advantage of using time-dependent Cox models is its ability to isolate effects during specific periods. In general, our survival analyses concur with the conclusions of Lee et al (2018) for the efficacy outcomes, which demonstrated superiority of BUP-NX. In contrast to the original report, our analysis indicates a greater risk of overdose for XR-NTX, predominantly during the study treatment phase. Further investigation of this finding is a pressing research priority.