Mortality rates amongst patients on opiate substitution treatment - comparing those on methadone or buprenorphine medications

Talk Code: 

The problem

Opiate substitution treatment (OST) is the commonest intervention offered to injection drug users (IDU) and in the UK is mainly prescribed by GPs. Some evidence suggests an increased risk of mortality at the beginning and end of periods of OST and that for net mortality reduction treatment should be prolonged. It is not clear if this phenomenon is generally true or whether it varies according other considerations such as choice of substitute, other drugs or comorbidity.

The approach

Using the Clinical Practice Research Datalink (CPRD), data for 11,593 patients undergoing OST between 1998 and 2014 were investigated. Prescription histories for these patients were converted to a series of treatment episodes. These episodes were analysed using poisson regression. Due to the effect of the censoring of episodes by death on the estimation of between patient variability, random effect methods were not used. Instead standard poisson regression was used with adjustment for over dispersion. The risk of mortality was assessed for periods within episodes covering weeks 1 and 2, weeks 3 and 4 and the remainder of weeks on and off treatment. OST regimes were classified as methadone only and all other combinations including those who switched from methadone to buprenorphine or dihydrocodeine. The interaction between regime and risk period was explored. Adjustment was made for time varying covariates including age of the patient, year of drug administration and a comorbidity index. Adjustment was also made for gender.


A total of 786,263 prescriptions provided information on 26,239 episodes representing a mean of 2.3 episodes per patient. Of these, 15,656 episodes were for methadone only; 4813 for buprenorphine only; 2948 for dihydrocodeine only with the remainder (2822) representing a combination of medications. In all, 6010 patients (10,813 episodes) were treated with methadone only throughout their treatment history recorded in CPRD. A total of 374 deaths were observed. Using the remainder of weeks on treatment as a reference group, the mortality rate ratios for the ‘other' regime were 2.15 [95% CI 0.45-10.19] and 1.07 [0.10-11.00] for weeks 1-2 and 3-4 on treatment and 17.61 [9.12-34.01], 7.37 [2.94-18.46] and 3.34 [2.02-5.50] for weeks 1-2, 3-4 and the remainder off treatment. Methadone only patients tended to have ratios 1.87 times higher (including the reference group). There was no evidence of an interaction between regime and risk period (p=0.47).


These preliminary findings confirm increased mortality risk at the beginning and end of periods of OST and suggest that this may vary with choice of substitute. We are continuing to explore these data in particular we are investigating strategies to address the issue of confounding by indication. This evidence has important implications for all GPs providing care to IDU.


  • Colin Steer
  • John Macleod
  • Rosie Cornish
  • Matt Hickman