48 Trials and tribulations of establishing treatment effectiveness in addiction research

Abstract

Objectives Without treatment opioid addiction can incur a substantial increase in mortality and risk for serious comorbidities such as HIV and hepatitis. Opioid substitution and antagonistic therapies (OSATs) are front-line treatments for opioid addiction. The emergence of multiple OSATs renders traditional meta-analysis of direct evidence from randomised trials inadequate to provide hierarchical estimates of the best available treatment. Utilising systematic review methods, we provide the first multiple treatments comparison and network meta-analysis to combine evidence from all trials examining OSAT with the aim of distinguishing the most effective treatments for opioid addiction. Our secondary aims were to assess the quality of the evidence as well as examine how this evidence is incorporated into clinical practice guidelines. Method We searched nine databases from inception to January 1, 2014 to identify randomised controlled trials assessing the effectiveness OSATs for opioid addiction. The primary outcome was treatment retention. We also evaluated the eligibility criteria used across trials. To quantify the effect of trials’ eligibility criteria on generalizability, we applied these criteria to data from an observational study of opioid-dependent patients (n=298). We then accessed the Canadian, American, and British OSAT guidelines to evaluate how evidence is used in the recommendations. Results We identified 60 trials eligible for inclusion and among those, 28 trials testing 16 interventions in a total of 3342 participants were included in the network meta-analysis assessing treatment retention. In comparison to all other OSATs, heroin consistently ranked highest for increasing the odds of remaining in treatment. A representative number of trials exclude patients with psychiatric (60%, k=36) and physical comorbidity (51.7%, k=31), current alcohol/substance-use problems (31.7%, k=19), or patients taking psychotropic medications (48.3%, k=29). These criteria were restrictive and in some cases rendered 70% of the observational sample of 298 general methadone patients ineligible for inclusion. OSAT guidelines made strong recommendations supported by evidence with poor generalizability. Conclusions Among patients treated with OSATs, those randomised to heroin-assisted treatment or high-dose methadone were more likely to remain in therapy. There was insufficient evidence to confidently rank the remaining treatments. Trials assessing OSATs’ often exclude patients with concurrent disorders. Eligibility criteria that result in the exclusion of a substantial number of patients from randomised trials jeopardise the generalizability of treatment effect to much of the clinical population. If the excluded patients respond differently to treatment, results from these trials are likely to overestimate the true effectiveness of OSATs. North American guidelines should consider these limitations when drafting clinical recommendations.