Key Takeaways
Extended Takeaways
- The 2 candidate thresholds serve different clinical purposes: the weeks 1–2 cutoff showed higher sensitivity than specificity (0.747 vs 0.688), whereas the weeks 1–3 cutoff showed higher specificity than sensitivity (0.810 vs 0.649), which may help clinicians match lower-intensity versus more resource-intensive responses to risk.
- Both early-response definitions were better at ruling out later sustained opioid use than confirming it: negative predictive value was 86% for the weeks 1–2 threshold and 83% for the weeks 1–3 threshold, while positive predictive value was 52% and 61%, respectively.
- Early nonresponse was linked not only to later sustained use but also to less short-term treatment engagement, with approximately 1.11 fewer retention weeks using the weeks 1–2 threshold and 1.46 fewer retention weeks using the weeks 1–3 threshold.
- The association with ongoing opioid use was clinically substantial on continuous measures, as early nonresponse predicted approximately 3.12 fewer opioid-free weeks with the weeks 1–2 definition and approximately 3.70 fewer opioid-free weeks with the weeks 1–3 definition during weeks 5–12.
- A single opioid-use day across weeks 1–3 did not uniformly indicate high risk: among participants with 1 day of use, 73.2% used during the first 2 weeks, and outcome differences were not significant by whether that day occurred earlier or in week 3.
- These thresholds were derived from self-reported opioid use frequency and showed similar results in bootstrapped and out-of-bag analyses, supporting the practicality of using brief weekly patient report rather than relying solely on urine toxicology to track early buprenorphine response.
