Haasen et al (Reference Haasen, Verthein and Degkwitz2007) report highly significant findings from their trial of heroin plus methadone maintenance. A small problem is that the heroin plus methadone group were, to a large extent, self-selected, with only 2.3% failing to initiate treatment in this group v. 28.8% in the methadone only arms. They state that this ‘limiting effect… is minimised’ by randomisation and intention-to-treat analysis. Intention-to-treat analysis makes their already significant findings even more impressive, but randomisation is limited by the unavoidable self-selection in a trial which is necessarily not masked. The paper goes on to say that ‘retention was higher in the heroin group, with 67.2% completing the 12-month treatment compared with 40% of the methadone group’, but later this is given as 56.3% for the methadone only group when the 28.8% who did not initiate treatment were excluded. The retention rate would rise again if the drop-out (‘discontinued’) rate was calculated using the same reduced denominator, and therefore retention rates would possibly differ insignificantly. Taking this into consideration would also explain the almost equal numbers of ‘discontinued’ participants in the two main arms of the trial.

The findings of this aspect of the trial are not surprising and without doubt it would be difficult to devise a control with the reinforcing power of heroin. Injectable methadone, financial incentives or pleasurable activities might approximate a substitute and produce more accurate retention figures. With the high cost of freeze-dried heroin, as used in the UK, adding these incentives might attract funding for a suitably modified study conducted here. Given that high retention rates are today's centrally defined most desirable outcome in the UK, this sort of study might be even more attractive here.