We wish to comment on Stephen et al.'s argument that trials of deep brain stimulation (DBS) are warranted in heroin addiction, because DBS needs only to be 49% as effective as methadone maintenance treatment (MMT) for it to be a cost-effective intervention [1]. We accept that threshold analyses can be informative in specifying the characteristics required for cost-effectiveness of interventions, but we have some major concerns with the way in which their analyses were performed and the conclusions drawn.

First, the analysis is biased in favour of DBS because it compares the average performance of MMT, a treatment that is often implemented indifferently, with evidence on the effectiveness of DBS in Parkinson's disease (PD) reported in clinical trials conducted in academic research settings by well-resourced research groups. MMT is often provided in ways that undermine its effectiveness significantly (e.g. by using inadequate doses of methadone to prevent drug cravings) or responding punitively to positive urines (e.g. by dose reductions or expulsion from treatment) [2]. Lintzeris showed that many treatment refractory heroin-dependent individuals will respond to MMT if it is provided appropriately [3]. History also suggests that the average effectiveness of DBS is likely to be poorer when used routinely in less than optimal conditions with less intensive psychosocial support [4].

Secondly, we doubt that good results of DBS in PD will apply to heroin-addicted individuals. PD patients are not ambivalent about controlling their movement disorders in the way that heroin-addicted individuals are about their heroin use. Heroin-addicted individuals are adept at subverting the effects of methadone and naltrexone, so we would expect them to find ways of modulating stimulation parameters to allow drug use (e.g. by misreporting craving when stimulation parameters are adjusted). This may have happened in a recent case report of DBS in a heroin-addicted patient who succeeded in using heroin for 2 weeks after undergoing DBS [5].

Thirdly, the costs of untreated heroin addiction that were used in the threshold analysis all come from the United States, where criminal justice costs comprise 57% of the total costs. By using these cost data, DBS is being advocated implicitly as a crime control measure. An ethical review of the use of psychosurgery for the treatment of psychiatric disorders recommended strongly that it should not be used for social control ([6], p. 58).

Fourthly, we do not believe that MMT is the appropriate comparator in their threshold modelling. DBS would only be used to treat addicted people who failed existing treatments (such as MMT and buprenorphine). Given this, the more appropriate comparator would be heroin maintenance treatment, which has been trialled with positive results in patients who have failed at MMT, and found to be more cost-effective than MMT in this patient population [7,8].

Many of these points are acknowledged in the discussion, but these caveats are not reflected in the abstract, which means that the results seem to support a stronger claim than we believe they warrant.

Declarations of interest