Bergmark's [1] contribution to the debate on treatment mechanisms is very welcome. He joins what is becoming a queue of addiction specialists lining up to register their concern at the apparent unproductivity of the hegemonic methods of conducting research in our field. Expressions of concern range from challenging the previously incontestable dominance of the randomized controlled trial [2], to questioning the evidence for purported mechanisms of change [3,4], to a more radical questioning of some of the assumptions on which addiction treatment research has been based [5–7]. Bergmark asks whether we should ‘just continue on the same old track’ or ‘re-organize the focus of addiction treatment research’. Like others in the growing queue of dissenters, I see increasingly the need for the latter. Moos [6] puts it pithily when he writes, ‘. . . we need more emphasis on empirically supported treatment processes (ESTPs) rather than on empirically supported treatments (ESTs) or empirically supported therapeutic relationships (ESTRs)’.

The fundamental problem, in my view, has been the continued adherence to the ‘drug treatment metaphor’[8] well beyond its period of usefulness. Based on a narrow biomedical model, it privileges named treatments (whether in the form of medications or psychosocial methods) in prescribed doses (milligrams or sessions), each with supposed mechanisms for effecting change (whether involving neural pathways, psychophysiological changes or cognitive–emotional processes). While that stripped-down model of change may serve well enough in the case of a well-understood medical condition treated by a physical method, it does not fit the bill in the case of forms of excessive appetitive behaviour occurring in a complex family, social, community, cultural and moral environment [9]. It simply ignores most of what is happening. As Bergmark puts it, it is indeed odd that we continue to focus on one small part of the action which accounts for so little of the variance in outcome.

Project COMBINE (Combining Medications and Behavioral Interventions) [10] illustrates the problem very well. Besides the named medications and psychosocial treatment that were the focus of perhaps the largest-ever, state-of-the-art randomized treatment trial, the interactions of patients within the wider treatment context, let alone with the extra-treatment context of their lives, is either relegated to the status of something to be controlled (for example medical management and site effects) or as something to be minimized in importance by referring to it as ‘placebo’, or is treated simply as out of the scope of this type of research (whether the patients had the support of close relatives, for example). This is in no way to criticize the authors of the study who were, as many of us appreciate, pulling out all the stops to bring off such a high-quality study. The fault lies with the prevailing treatment research ideology. It is particularly strange, but very telling, that so much attention in Project COMBINE should have been given to examining the possible differences in outcome following the named medications and psychosocial treatment when so little attention was given to the ‘medical management’ (MM) which was given to all [nine sessions in most cases and four for the combined behavioural intervention (CBI) no pills group]. MM appears to have been an excellent treatment in itself (it included a review of diagnosis and negative consequences of drinking, recommendation of abstinence, treatment plan and regular review, reviews of drinking, advice and encouragement to attend support groups for those who relapsed, etc.). As the authors stated [10], MM was: ‘. . . more intensive than that provided to alcohol-dependent patients in most healthcare settings’. The only explanation for the expectation that there would be between-treatment effects over and above MM must lie in the drug metaphor assumption that naltrexone, acamprosate and CBI were particularly active technologies with powerful mechanisms of action, whereas MM was a relatively inert offering with at best placebo effects.

The UK Alcohol Treatment Trial (UKATT), like Project MATCH and Project COMBINE, was a large multi-centre trial comparing alternative treatments (both psychosocial in this case) for alcohol problems. It also found evidence in support of the ‘dodo bird effect’ to which Bergmark [1] refers, i.e. outcomes following the two treatments could not be distinguished [11]. In UKATT, a number of process measures were used: video recordings of each treatment session [12]; sentence completion tests filled in by both clients and therapists after each session; and semistructured interviews carried out with samples of clients at 3-month and 12-month follow-ups [13]. Analyses of those data confirm that the two treatments (motivational enhancement therapy, MET; and social behaviour and network therapy, SBNT) were distinct—MET and SBNT therapists did significantly different things during sessions; both clients and therapists referred to significantly different things immediately after sessions; and clients made significantly different change attributions at follow-up. At the same time, the overlap between the two treatment groups in all those respects was considerable; many of the therapist skills on display were general ones; after sessions clients were as likely to highlight the value of non-specific factors as they were treatment-specific factors; and at follow-up clients were more likely to attribute change to non-specific factors. Those findings, although they are open to alternative interpretations, are in keeping with a contextual model of the effectiveness of treatment. Such a model emphasizes factors such as client commitment, therapist allegiance and the client–therapist alliance, and views personal change as being embedded within a complex, multi-component treatment system, itself nested within a broader life system that contains an array of inter-related factors promoting or constraining positive change [14,15]. These findings from UKATT, and those of Project COMBINE which Bergmark has looked at critically and helpfully, can be read as running counter to the prevailing medical–technical model that emphasizes specific factors related to mechanisms of change promoted by specific forms of therapy.