HOW TO MEASURE OUTCOME IN CLINICAL TRIALS OF SUBSTANCE ABUSE TREATMENT

Abstract

Donovan and colleagues 1 move from a restrictive title (treatment of illicit drug dependence) to general conclusions about substance abuse treatment outcomes, and from a single common indicator to a range of trial specific measures. The reason for identifying a single indicator is not evident, nor is the proposed choice of drug-taking behaviour. The statement ‘drug-taking behavior is the primary target for interventions’ cannot be supported in the light of recent developments 2. Drug-taking behaviour is presented as the main indicator, but ‘how to specify and to measure it must be determined for each trial separately’, and ‘the most appropriate outcome will vary as a function of salient variables inherent in the clinical trial’. A decision process should be developed to guide the selection of primary and secondary outcomes and the methods to assess them. Needless to say, the intended comparability of trial outcomes will remain difficult. Two details of limitations concern the treatment of nicotine dependence and the measurement of drug-taking. The paper assumes that all tobacco use is harmful and therefore treatment should aim at complete abstinence. This is in contrast with developments to reduce harm from tobacco smoking by effective replacement therapies 3 or by reducing the number of cigarettes smoked 4. The measurement of drug-taking behaviour is restricted to the extent of substance use. Route of administration and situational factors, relevant for negative consequences of use, are not considered. Of an initial list of 21 domains, Tiffany and colleagues 5 propose those responding to five criteria serving as a guideline, and identified five candidates for inclusion as primary outcomes in treatment studies. Among those are self-efficacy, social network and psychosocial functioning, but these are instrumental outcomes rather than end-points of outcome. Measures of social integration would be more appropriate as end-points. There are doubts about craving as a candidate primary outcome for all clinical trials. A recent review concludes: ‘Despite the apparent importance of craving to our understanding of addiction, no consensus regarding definition, measurement, and interpretation has been reached’6. Also, the paper does not consider indirect behavioural, physiological and reaction-time measures designed to assess craving 7. Imaging studies do not show a consistent connection between cue-induced brain stimuli eliciting craving and consecutive relapse 8. Moreover, awareness of danger from relapse can mobilize coping which reduces the odds for relapse 9. Successful resistance to craving might be a better candidate to measure treatment efficacy. Instruments measuring quality of life might be complemented by asking for alternative sources of contentment and satisfaction, which constitute an essential protective factor against developing substance dependence by ‘balancing the motivational system’, according to present theory of addiction 10. Both papers provide a rich overview of relevant work being done in the past. This may have impeded an attempt to rethink the problem anew. An alternative model could start from the diagnostic criteria. As usual in medicine, substance abuse and dependence are characterized by symptoms described in ICD-10 and DSM-IV. The goals of treatment might be defined as a reduction of the number and severity of symptoms of a given disorder. Why not apply this model? Select end-point symptoms where appropriate measurements are available. Some will need consensus-building efforts, but comparability of outcomes is feasible, and future treatment research might focus on which methods have a specific impact on which symptoms. This would include testing empirically the applicability of generic dependence symptoms for all substances of abuse 11. None.