[Commentary] DRUG TREATMENT—NOT AS USELESS AS IT SEEMS

The results of the Combining Medications and Behavioral Interventions for Alcoholism (COMBINE) study [1] must surely have disappointed a great many addiction therapists who had looked forward to confirmation of the value of combining psychosocial therapy with pharmacotherapy using naltrexone or acamprosate or both. Much previous literature [2-4] had indicated significant beneficial effects of both drugs, and had suggested that they would be still more effective when combined with each other and with some form of behavioral or psychotherapy. Instead, the COMBINE study results appeared to show that naltrexone and behavioral therapy were only slightly more effective than placebo; that acamprosate was not superior to placebo; and that combined therapies were not significantly better than single therapies. The range of reactions to these reported findings has probably been very wide, but two broad types of reaction can be identified. The first, which is illustrated by Bergmark's paper in the present issue of Addiction[5] is a seemingly discouraged conclusion that pharmacotherapies have no useful role in the treatment of alcohol dependence, and that the treatment outcome depends much more on factors affecting the relationship between therapist and patient. This rather sweeping rejection of pharmacotherapy leads to some puzzling statements. Bergmark states that the COMBINE results provide somewhat weak support for the preferential use of either drug, but does not relate this conclusion to the finding that naltrexone or behavioral therapy combined with medical management gave a significantly better outcome than placebo plus medical management, whereas acamprosate did not. He also states that the improvement seen in the group treated with combined placebo/behavioural therapy ‘must be mind-mediated and not of a neurochemical character’. This appears to reject all the modern psychobiological research into the neurochemical bases of affective and cognitive functions, and the mechanisms of action of drugs such as the antidepressants and antipsychotics. The second type of reaction, illustrated by Mason's response (Dr B. J. Mason, personal communication) is an analytical approach to explain the apparent contradiction between the COMBINE study results and those of a number of other fairly large-scale, well-designed, randomized controlled trials, both single-centre and multi-centre in scope. She points out two factors that could well explain the discrepancy. The first is that the statistical analysis carried out in the COMBINE study was designed specifically to test for interaction effects between medications and behavioral treatments, and did not include the no pill/behavioral treatment group. The latter group had a poorer outcome than all the other groups [1, Table 5]. When she re-analysed the results by analysis of variance including this group, all the other treatments were significantly better but did not differ from each other. Therefore it cannot be concluded that acamprosate was less effective. The second factor is that the placebo response rate in the COMBINE study was probably the highest of all randomized controlled trials conducted to date for treatment of alcohol dependence. She suggests that this reflects the fact that the patient selection and advance explanation processes were extremely intensive and time-demanding, and that this probably resulted in selection of a very highly motivated subject group, with greatly elevated expectations of improvement, even in those receiving placebo. The latter point is, in some respects, reminiscent of Bergmark's conclusion that the improvement was ‘mind-mediated’, but differs in not excluding a pharmacological effect of the drugs. The most useful indicator of the latter would probably be the difference in cumulative relapse rate during the year following the end of treatment. The importance of patients' expectations in short-term outcomes has been known for many years. For example, Armstrong [6] reported that patients who expressed willingness to take disulfiram had better outcomes than those who did not, even if they did not continue use of the drug for very long. He related the better outcome to personality characteristics of the patients who were willing to take the drug, or ‘to a strong positive relationship to a treatment person’. All the genetic studies in recent years have made it abundantly clear that the heritable risk of alcohol dependence is determined by a large number of genes rather than a single one, and it would seem highly improbable that all who become alcohol-dependent have exactly the same pattern of multiple genetic risk factors. A priori, therefore, it would seem most unlikely that one single treatment modality would be uniformly effective in all cases. There is still good reason to expect that a combination of modalities will prove to be of greatest benefit to the largest number of patients. I am greatly indebted to Dr Barbara J. Mason for allowing me to quote her as-yet unpublished re-analysis of the COMBINE study results.

Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.