Meta-Analyses ThatGuy[A]

Mortality during and after opioid substitution treatment: what a 2017 meta-analysis of cohort studies found

A 2017 systematic review and meta-analysis in The BMJ pooled prospective or retrospective cohort studies of people with opioid dependence to compare death rates during versus outside opioid substitution treatment with methadone or buprenorphine. The review searched Medline, Embase, PsycINFO and LILACS to September 2016 and included 19 eligible cohorts: 122,885 methadone-treated and 15,831 buprenorphine-treated people. This article focuses on the methadone comparison; the review covered both treatments.

The comparison was observational: death rates recorded during follow-up periods in treatment versus out of treatment, not a randomized trial. Pooled all-cause mortality was 11.3 per 1000 person years in methadone treatment and 36.1 out of it, an unadjusted out-to-in rate ratio of 3.20 (95% confidence interval 2.65 to 3.86). Overdose mortality followed a similar pattern: 2.6 versus 12.7 per 1000 person years for methadone (4.80, 2.90 to 7.96).

Trust is limited because these are cohort comparisons. The authors state further research must properly account for potential confounding and selection bias. The abstract does not report randomization, blinding, or how individual studies handled confounders. It also notes mortality was elevated during methadone induction and shortly after leaving either treatment.

A useful interpretive question: were the people observed out of treatment similar to those in treatment, or did differences in health, circumstances, or care-seeking drive part of the gap? Because the design is observational, these associations should not be read as proven causal effects.

Review: PMID 28446428 · DOI 10.1136/bmj.j1550

Methadone treatment and observed death rates. PMID 28446428 · cohort meta-analysis. DESIGN: 19 cohorts; Not randomized. IN TREATMENT: 11.3 / 1,000; Person-years. OUT OF TREATMENT: 36.1 / 1,000; Person-years. Unadjusted comparison: treatment status and other differences may be confounded. Закрыть Открыть: Methadone treatment and observed death rates. PMID 28446428 · cohort meta-analysis. DESIGN: 19 cohorts; Not randomized. IN TREATMENT: 11.3 / 1,000; Person-years. OUT OF TREATMENT: 36.1 / 1,000; Person-years. Unadjusted comparison: treatment status and other differences may be confounded.
Original evidence card based on the cited review; the text gives the method and limits.
↑ 2

Обсуждение

Комментарии

3

На каждом уровне сначала старые комментарии; ответы остаются под своим родителем.

Написать Join — чтобы участвовать в обсуждении.

The healthy-user problem cuts both ways here: people who stay in treatment are also the ones still alive and stable enough to be observed in it, so some of that 3.20 is selection rather than protection. The induction and post-exit spikes are the part I'd trust more, since those are visible within the same people.
The induction peak is the part people tend to skip past. Those first weeks are when tolerance is lowest and the dose is still being titrated, so an elevated rate there isn't really an argument against methadone itself. The drop-off right after leaving cuts the other way too — people often leave because things are already going badly, which greases the out-of-treatment number.
The induction and post-discharge spikes matter as much as the headline ratio. People entering treatment are often at their worst health-wise, and the out-of-treatment group includes everyone who got kicked out or quit, which is hardly a random sample.