Does opiate substitution therapy cause short-term mortality increases in injection drug users and do these effects differ in patients prescribed buprenorphine compared to those prescribed methadone?

Problem

Opiate substitution therapy (OST) is the most evidence based treatment for injection drug use (IDU) and in the UK most is delivered in primary care. Methadone and buprenorphine are the two most widely use substitutes with trials suggesting equivalent effectiveness in relation to short term social outcomes. Trials are not powered to detect effects on mortality. Some observational evidence suggests transient mortality increases at the beginning and end of OST treatment periods. Whether these are consistently seen, whether they vary with substitute and the implications for practice are not clear.

Approach

Data were extracted from the Clinical Practice Research Datalink on 11,033 patients receiving methadone or buprenorphine as part of OST. Their prescription histories for the period 1st January 1998 to 31st July 2014 were translated into 26,546 treatment episodes where a gap of at least 28 days existed between the cessation of one prescription and the start of the next. We considered effects on all cause and drug related mortality (ACM and DRM) in relation to different periods of treatment using poisson regression taking account of clustering by patient. We adjusted analyses for gender, age, year, comorbidity and UK region. We also used propensity scores, propensity matching and prescribing preference as an instrumental variable to address residual confounding by indication.

Findings

ACM accounted for 587 deaths within the sample. In a subsample of 5935 patients with cause of death data, 87 of 345 deaths were identified as DRM. Methadone accounted for 65% of all treatment episodes and 39% of ACM occurred during periods of methadone treatment.Overall, buprenorphine treatment appeared protective for ACM compared to methadone treatment with an incident rate ratio (IRR) of 0.14 [95% CI 0.10, 0.20]. There was evidence that this benefit varied with treatment period (p=0.0022) such that the first 4 weeks on treatment (p=0.056) and the first 4 weeks off treatment (p=0.0003) had IRR of ~0.07. These apparently beneficial effects of buprenorphine were still apparent in some additional analyses to address residual confounding by indication.For DRM, there was no evident of a consistent buprenorphine beneficial effect (p=0.27) or of effect modification (p=0.16).

Consequences

We confirmed short term mortality increases at the beginning and end of episodes of treatment amongst IDU receiving OST in UK primary care. We found evidence that these effects were mainly seen amongst patients receiving methadone and that buprenorphine may be protective. Paradoxically effects were seen in relation to ACM not DRM possibly reflecting misclassification and reduced power. We provide observational evidence that buprenorphine may be safer than methadone in relation to iatrogenic mortality risk amongst IDU prescribed OST which has implications for practice. Residual confounding by indication is still possible in our data.