Does co-prescription of benzodiazepines, z-drugs or gabapentinoids increase mortality risk in injection drug users prescribed opiate substitution therapy?
Deaths amongst injection drug users (IDU) are increasing and a substantial proportion of these deaths are amongst IDU in primary care treatment with opiate substitution therapy (OST). Despite advice against co-prescription in clinical guidelines many IDU on OST are co-prescribed benzodiazepines (BZD), z-drugs (zaleplon, zolpidem and zopiclone) and gabapentinoids. Prescribing data linked to mortality data in Scotland provided inconclusive evidence of increased mortality associated with BZD co-prescription. More recent evidence from London suggested that BZD co-prescription was associated with better treatment outcomes and lower mortality. None of this evidence provided strong support for any causal conclusions and confounding by indication remained a strong possibility.
Data were extracted from the Clinical Practice Research Datalink concerning 12,171 patients receiving OST between 1st January 1998 and 31st July 2014. In these patients we identified episodes of co-prescription of BZD, z-drugs and gabapentinoids. We assessed the effects of these exposures on both all cause mortality (ACM) and drug related poisoning (DRP) through linkage to mortality data from the ONS. Poisson regression taking account of clustering. by patients was used to analyse the data. We adjusted analyses for gender, age, year, comorbidity and UK region. We also used propensity scores and propensity matching to address residual confounding by indication. We took evidence of an effect specific to DRP, evidence of a dose response effect and lack of attenuation on adjustment to provide stronger support for a causal relation
Of 31,659 OST treatment episodes identified around 60% involved co-prescription, mainly of BZD. In 39,670 person years of follow-up there were 708 deaths. ONS linkage allowed acertainment of DRP in 54% of patients. Overall rates of ACM and DRP were 1.78 and 0.55 per 100 person years respectively. BZD co-prescription had little effect on ACM but substantially increased risk of DRP with evidence of a dose response effect (mortality rates per 100 person years 0.44 off treatment, 0.79 normal dose, 1.65 high dose, p<0.0001). Adjustment had little effect on these estimates. In contrast z-drug and gabapentinoid co-prescription was associated with increased risk of both ACM and DRP with no evidence of a dose response effect and, in the case of gabapentinoids, substantial attenuation on adjustment.
We provide the strongest observational evidence to date that co-prescription of BZD causes increased risk of death in IDU on OST. Co-prescription of z-drugs and gabapentinoids is also associated with increased mortality risk however evidence that this association is causal was not strong and it may reflect residual confounding by indication. GPs prescribing OST to IDU should adhere to current clinical guidelines advising against co-prescription of BZDs. Guidelines also advise against co-prescription of z-drugs and gabapentinoids. Evidence that this causes harm is less clear but a cautious approach seems warranted till better evidence is available.