Dr Sellman's excellent synthesis and common-sense conclusions regarding the addiction treatment field will doubtless become required reading in many addiction training centres, and deservedly so [1]. However, Professor Sellman makes one point that appears to us to misrepresent some of the evidence and reflects a line of thinking which, if broadly accepted, has some potential to reverse some of the important, and very hard-won, progress we have made in improving the quality of addiction treatment. The premise that ‘Different psychotherapies appear to produce similar results’, reminiscent of the ‘Dodo Bird verdict’[2], overlooks some of the recent evidence and results in some misconceptions. Following Professor Sellman's lead, we will limit ourselves to 10 points, as follows.

  • 1As our research improves, so do our treatments

While the general equivalence of effect sizes of psychotherapy may have been accurate 20 years ago, this is no longer the case. Greater methodological rigor, with emphasis on treatment integrity and focus upon treatment specificity, has led to the development of a range of empirically validated therapies (EVTs) [3]. While we acknowledge that there is still ample room for improvement, both our methodologies and our effect sizes have advanced over the years. Take, for example, the development of Contingency Management (CM) [4], which has yielded consistently large and robust effect sizes with a range of populations and settings [5,6]. While the relative cost of CM has limited its utility in practice, less expensive versions have been developed [7].

  • 2‘Psychotherapies have similar effect sizes’ does not translate to ‘. . . so it doesn't matter what you do’

Even if we accept the premise that rigorous randomized efficacy trials of well-defined, well-supervised behavioral therapies tend to yield similar effect sizes, it does not then follow that ‘anything you do works’. There is ample evidence that EVTs tend to improve outcomes when transferred to regular settings; that is, are more effective than most ‘treatment as usual’.

  • 3Empirically based therapies are still rare in clinical practice

Similarly, the argument that ‘all therapies are alike, so it doesn't matter what you do’ also assumes that clinicians are in fact providing EVTs. There are now data indicating that this is not the case: audiotapes of ‘standard treatment’ sessions from the community settings in the United States, where clinicians professed to using a range of EVTs, indicated that interventions associated with therapies such as CBT, Twelve Step facilitation and human immunodeficiency virus (HIV) risk reduction were so rare as to be almost undetectable [8]. These session tapes also indicated that (i) clinicians also overestimated the amount of EVTs they were delivering consistently [9], and (ii) clinician-initiated discourse that was unrelated to the patients problems (i.e. ‘chat’) was seen much more frequently than EVTs [10].

  • 4Important differences among therapies emerge when we look for them systematically

Recent data from well-controlled trials emphasizes specific effects of defined treatments. These have yielded data indicating that, for example, behavioral family therapy indeed has positive effects on the adult parents who are treated, but which also extend to their children [11], that cognitive behavioural therapy (CBT) skill training results in durable benefits [12,13], and that use of contingency management to reduce smoking in pregnant women has demonstrable positive effects on infant's birth weight [14].

  • 5Why are we always picking on poor psychotherapy?

Before we lay the woes of comparative effectiveness research at the feet of behavioral therapies alone, we should remember that there are also few meaningful effect sizes among pharmacotherapies such as antidepressants. The effect sizes of our most powerful behavioral therapies tend to be larger than even those of antidepressants [6].

  • 6What treatment would you select for a loved one?

The argument that ‘all therapies are alike, so it doesn't matter what you do’ also falls short when one considers which treatment one would select if a friend or relative developed addiction problems. In such cases, one becomes acutely aware that a range of different therapies are required, where a brief motivational approach would be indicated if addiction problems are just beginning to emerge, an intensive CBT or contingency management approach would be needed if the problems were more severe. Further, one would hope for an experienced, well-trained and supervised clinician who had demonstrated mastery of the approach delivered.

  • 7The working alliance may be a necessary part of treatment, but not sufficient to produce lasting effects

Sellman's review also implies the key ‘active ingredient’ of psychotherapy is a good working alliance. The therapeutic alliance may be necessary, but it is by no means a sufficient ingredient for good outcome. It well may be the case that when one is delivering unstructured supportive counseling a positive alliance is critical. Although positive treatment outcomes cannot be expected, even with EVTs, when alliances are poor there is evidence that EVTs may be less dependent than unstructured supportive therapies to exert their effects [16].

  • 8Perhaps the alliance is not everything

Some novel therapies appear to be effective even in the absence of a therapeutic alliance. Consider, for example, the evidence regarding the efficacy of computer-assisted therapies, which appear to have efficacy [17] and durability [12], even in the absence of a traditional therapeutic alliance.

  • 9Curing ‘non-specificitis’

A corollary of the premise that ‘all therapies work, so it doesn't matter what you do’ is ‘all therapies are the same because they do the same things’. Systematic evaluation of mediation (i.e. that specific therapies have specific effects, and that those effects account for therapeutic benefit) are rare [18]. True mediational analyses are methodologically challenging, as they require psychometrically sound assessment of putative mediators (a woefully understudied area) and measurement of change in the mediator prior to evaluation of outcome [19]. One possible example is our work on our computer-assisted version of CBT4CBT, where we have demonstrated that the computer program actually enhances skills in the targeted domains, that such skill acquisition is treatment-specific and associated with improved outcomes.

  • 10Let us wait for the data to come in

Another caution before we accept any conclusion regarding equality of psychotherapies is that there are, in fact, few head-to-head, direct comparisons of EVTs. We have comparatively few studies in the field that directly contrast two active behavioral treatments for addictions with hypothetically different mechanisms of action. Project MATCH, the perennial example used to make this point, was not in fact designed to compare main effects of the three treatments evaluated. In contrast with medication trials, where drug companies have a financial incentive to contrast their drugs with existing treatments, there is no such driving force among behavioral therapies. Studies with strong a priori hypotheses about treatment-specific effects are possible but are infrequently undertaken, and we emphasize the need for a new generation of research that looks at treatment-specific effects as well as combined treatment effects.

Declaration of interest



Support was provided by National Institute on Drug Abuse grants P50-DA09241, R37-DA 015969, U10 DA13038, K05-DA00457 (to K.M.C.), K05-DA00089 (to B.J.R.) and the VISN 1 Mental Illness Educational, Research, and Clinical Center (MIRECC).