Opioid dependence is a chronic, relapsing-remitting condition increasing morbidity, mortality and social problems. Opiate substitution reduces problems. Methadone and buprenorphine are the commonest substitutes. Both are effective.
NICE recently considered their use in the UK and our team modelled their cost-effectiveness. Both agents were cost-effective compared to no substitute but methadone dominated buprenorphine. Shortly after, an editorial argued buprenorphine should replace methadone because of its superior safety. The model for NICE did not consider the mortality. We thought it important to undertake further modelling to explore whether taking this into account would significantly alter the ICERS of the agents.
The starting point was the original model (one year time horizon). We used our previous systematic reviews to provide parameter estimates and data on HRQoL from a value of health panel of the general public. We sought additional data on risks of death from observational studies. We estimated QALY differences using probabilities of death in the different treatments and the proportions retained on treatment. Because of the relapsing-remitting nature of this condition and highly varying treatment pathways of patients, we assumed those surviving one year after starting treatment would have similar future life courses, regardless of original therapy. Deaths were assumed to occur at six months. The analysis took a health service perspective. Benefits were discounted at 3.5%.
In the original model ICERs for methadone and buprenorphine compared to no substitute, were £4K/QALY and £27K/QALY. Methadone dominated buprenorphine. These reduced to £6K/QALY and £11K/QALY when mortality was incorporated. Methadone still dominated buprenorphine. Sensitivity analyses showed buprenorphine compared to methadone could reach conventional UK levels for cost-effectiveness by doubling the RR of death of methadone compared to no substitute.
Despite its significantly higher risk of death from overdose, methadone is more effective and cost-effective than buprenorphine due to better retention rates. A striking finding was how much better either substitute was compared to no substitute. Higher drop-out with buprenorphine increases mortality and counteracts the gains from its superior safety. According to accepted UK thresholds, both types of therapy are cost-effective compared to no substitute. Methadone remains more effective and cost-effective when directly compared. This modelling shows the importance of basing policies on evidence. The editorial showed how easily pathophysiological reasoning can mislead us.