Smyth et al (Reference Smyth, Barry and Lane2005) reported outcomes of abstinence-oriented in-patient treatment for opiate users at 2–3 years and found that 23% of patients were abstinent for the preceding 30 days according to self-report without methadone maintenance. At the start of the treatment 49% had injected heroin. There was, however, a group of patients who were truly abstinent: those who had died.

Of the 109 patients who had been located out of the original 149, 5 had died. The total expected number of deaths from the original sample would therefore be closer to 7, but would perhaps be even higher if we assume that those lost to follow-up led more ‘chaotic’ lifestyles. The authors rightly note that abstinence-oriented treatment is associated with accidental overdose (Reference Strang, McCambridge and BestStrang et al, 2003).

In Glasgow, before the advent of supervised consumption, rates of methadone-related overdose were around 2.5 per 100 treatment-years. This rate fell to less than 0.5 per 100 treatment-years (Advisory Committee on the Misuse of Drugs, 2000) after the supervised consumption of methadone was introduced. Supervised methadone consumption is known to be effective in reducing the risk of overdose and there is a dose-related effect in reducing mortality, with doses over 75 mg being more effective than doses below 55 mg (Reference van Ameijden, Langendam and Coutinhovan Ameijden et al, 1999). Methadone also reduces the risk of injecting; this in turn reduces viral transmission, which is the other significant risk of increased mortality among drug users (Reference Dolan, Wodak and HallDolan et al, 1998).

However, the attitude of treatment agencies towards extended maintenance is changing in the direction of delineated treatment episodes (National Treatment Agency for Substance Misuse, 2005). In these days of crack cocaine, the belief that methadone treatment works (Reference Gossop, Marsden and StewartGossop et al, 2003) and saves money (Reference Godfrey, Stewart and GossopGodfrey et al, 2004) has diminished. This is despite evidence for interventions such as contingency management and cognitive–behavioural therapy using substitute prescribing (Reference Rowan-Szal, Bartholomew and ChathamRowan-Szal et al, 2004).

Of course, abstinence should be a potential goal of drug treatment. Deciding those patients for whom abstinence-oriented treatment is appropriate, and the risk of such treatment, is more difficult. There is no reliable evidence for matching patients to optimal treatments in addiction. However, those who inject, isolated users and alcohol/benzodiazepine co-users are all over-represented in the morgue (Reference Warner-Smith, Darke and LynskeyWarner-Smith et al, 2001). Risk awareness might well be a reasonable first step and for many abstinence might be more dangerous than desirable.