Commentary on Derringer et al. (2013): More evidence for a unidimensional framework for substance use disorders


The recent publication of the 5th edition of the Diagnostic and Statistical Manual of the American Psychiatric Association (DSM-5) [1] has brought considerable attention to the decision to merge the abuse and dependence criteria in the previous edition [2] into a single unidimensional disorder. It has seemed to this reader that the critics have predominated in the media, but the science behind the decision is both compelling and consistent. With few exceptions, studies examining whether there are one, two or three disorders have concluded that the data most clearly fit a model of a single disorder, varying only in the severity of the disorder [3]. In DSM-5, substance use disorders are designated mild, moderate or severe, depending upon the number of criteria met. Increases in alcohol consumption and the number of diagnostic criteria met are correlated strongly with substance-related consequences and harms, providing convergent validity [4, 5], has found consistent support. In the current issue, Derringer and colleagues [6] examined the psychometric qualities of ever use and DSM-IV [2] dependence across several substances in both clinically over-sampled and community (epidemiological) groups. Their analysis consisted of first determining whether the groups' diagnostic symptoms differed by more than severity (i.e. number of criteria), then whether diagnoses for different substances had unique qualities or instead represented a single, multi-substance dependence factor. Finally, they determined whether there were differences between sexes and ethnic groups. Overall, their results provide strong support for unidimensional diagnosis across substances and, importantly, reassurance that clinical and community-dwelling groups differ only in severity, not in type or pattern of diagnostic criteria, thus facilitating the interpretation of studies and public health policy development. In addition, they found that the psychometric qualities of diagnosis did not vary across substances, but instead supported a single, multi-substance dependence model. This finding is in accord with accumulating evidence that most of the genetic diathesis for multiple substance use disorders and psychiatric disorders is common, rather than specific to each [7]. Finally, they found no differences between men and women, or among different ethnic groups. However, as the authors' note, there were too few Hispanics for reliable analysis, and other groups such as Asians were not represented. The relationship of consumption patterns and associated harms, but not necessarily diagnosis, varies among different racial and ethnic groups in the United States [8], and among different countries [9]. Although there is evidence of cross-national unidimensionality of alcohol use disorder [5], additional research across ethnic groups and countries is needed. Unfortunately, measures of consumption patterns are too imprecise, as most rely upon retrospective self-report. Although such surveys may provide general categories of use (abstinent, moderate, heavy), it will be difficult to make progress until more precise, real-time measures are used, because of the complexity of drinking patterns [10]. Mathematical modeling of more precise measurements of drinking behavior could, potentially, provide categories of consumption patterns that can then be tied to genetics, environmental factors, harms and treatment approaches. This could, in turn, help us to understand the path from common diatheses to specific disorders.

Declaration of interests

I am founder and sole owner of a company that promotes evidence-based treatments for substance use disorders. I serve or have served on several government and other advisory committees (none commercial), and have given invited presentations on treatment of SUDs.