Newman rightly draws attention to the effectiveness of appropriately delivered methadone treatment for many people with heroin addiction worldwide over the past half-century. Our economic evaluation ^{Reference Byford, Barrett, Metrebian, Groshkova, Cary and Charles1} and the preceding report on the main findings from the RIOTT trial ^{Reference Strang, Metrebian, Lintzeris, Potts, Carnwath and Mayet2} should not be considered an attack on the value of oral methadone to the majority who show substantial benefit from this treatment. ^{3,Reference Strang, Babor, Caulkins, Fischer, Foxcroft and Humphreys4} Rather the RIOTT trial needs to be recognised for what it was - an investigation of effectiveness and cost-effectiveness of alternative treatments in a subgroup of the treatment population with severe and chronic addiction who were not responding to oral methadone maintenance treatment.

It is also appropriate to inject a note of caution about the potential influence of expectations on trial participants. This limitation is inherent in any trial where the patient has a preference for which treatment arm they may be assigned to, and Newman is right that this has the potential to be a pronounced influence in the addiction treatment field. In fact, aware of this potential, we gathered some data from patients on their expectations and experiences of treatment within the trial, and this has recently been reported separately. ^{Reference Groshkova, Metrebian, Hallam, Charles, Martin and Forzisi5}

Newman notes the modest sample size in this trial (total of 127 participants). This is a particular challenge in a field where treatment is intensive and expensive, and in countries which do not have a tradition of funding large treatment trials in the addictions field. We would nevertheless point out that the sample size was calculated in advance by the applicants for the original research award and was judged to be adequate to detect the expected effect size as defined in the protocol. ^{Reference Strang, Metrebian, Lintzeris, Potts, Carnwath and Mayet2}

Newman highlights a further limitation of sample size in this highly variable population, using the example of criminal activity. Although the oral methadone group reported committing a much higher number of crimes than the injectable methadone group, the latter group spent more nights in prison. However, the total number of participants spending any time in prison (n = 6; 5%) is extremely small relative to the number reporting any criminal activity (n = 50; 42%), so it would be inappropriate to try and come to any comparative conclusions.

In conclusion, we acknowledge the limitations of research in this complex subgroup with chronic heroin addiction and also the evidence of benefit from oral methadone in the broader population of people addicted to the drug. However, we consider the important findings reported in the paper are that, for this subgroup doing persistently badly on oral methadone treatment, it is important for clinicians to work with their patients to explore alternative options, such as injectable treatments, which may achieve health benefits not being achieved in the expected manner with the orthodox first-line treatment, and which may achieve this health benefit in a more cost-effective manner. Such personalisation of treatment plans is important but is currently being hindered by the cost implications of providing injectable alternatives and a previous lack of evidence of cost-effectiveness.