The End of the Dependence Syndrome as We Know It?

The American Psychiatric Association (APA) plans to abandon the diagnoses of substance abuse and dependence in favour of a single dimension of substance-related disorders in DSM-V, combining all but one of the items and concepts of its diagnostic predecessors. It is argued here that statistical considerations have been prioritized over conceptual issues. Further, there is a risk that this will serve to increase stigma for people suffering from dependence. Dependence has been a helpful unifying heuristic for clinicians, scientists and sufferers for more than 30 years, and has strong empirical support. While the term ‘substance abuse’ has outlived its usefulness and lacks empirical support, consideration should be given to retaining dependence as a diagnostic term in DSM-V with specifiers of mild, moderate and severe. It is true that there were some problems with DSM-IV. The diagnostic criteria for ‘alcohol abuse’ had limited reliability and validity in practice and the term was considered by many to be pejorative, hence the choice of ‘harmful use’ in ICD-10 (although it, too, lacks adequate validity) [1]. However, as Professor O'Brien acknowledges in his paper, the diagnosis of dependence shows excellent inter-rater reliability [2], and the dependence syndrome has a wealth of evidence of construct, concurrent and predictive validity [3,4]. Nevertheless, O'Brien tells us that although the term ‘dependence’ was adopted in DSM-III-R by a narrow margin in preference to ‘addiction’, concerns remained about its meaning and implications for clinical practice. The main consideration in the DSM-V working group's decision to jettison dependence from the classification system in favour of ‘substance-related disorders’ or ‘use disorder’ appears to be linked to the concerns of physicians managing chronic pain and various physical problems with dependence-producing drugs. It is fair to say that dependence should be defined by more than simply the symptoms related to neuroadaptation. However, there appears to be another subtext, namely that physicians are uncomfortable that patients being treated, for example, for cancer-related pain, could potentially have the same dependence diagnosis as people with illicit heroin dependence, and hence suffer the same social stigma. However, by endorsing this distinction, the DSM-V working group has succeeded in adding to the stigma already experienced by people who become dependent on substances which are not being prescribed or taken for medical purposes. Surely this is a missed opportunity to work on the negative attitudes and stereotypes which lead to the stigma felt by alcoholics and illicit drug addicts. If, as is being argued by O'Brien and others, dependence is a valid and reliable construct, it should not matter if cancer patients on opiates and illicit heroin users meet the same criteria for dependence. Neither group should have to tolerate stigma. Equally, the clinical decision of whether or not or how much to prescribe opioids to either group should, as now, be based on clinical need, including the various co-occurring diagnoses, not on the dependence alone. So what of the new DSM-V proposals? This seems to be a triumph of statistical over conceptual considerations. Using latent class analysis and item–response theory applied mainly to general population data on the abuse and dependence items of DSM-IV, a rationale has been developed to have a single scale containing all the old dependence criteria, and all except one of the old abuse criteria [5]. An additional criterion of craving has been added, which seems fair enough and in line with ICD-10. Now meeting two to three (instead of one) criteria equates to a ‘moderate’ use disorder and meeting four (instead of two) or more equals ‘severe’. There seems to be a lack of clarity in these proposals about the underlying concepts and an over-reliance on empirical general population data. In some ways the biaxial concept [6] is more important than the dependence syndrome [7] in helping our understanding. The biaxial concept proposed that social, physical and psychological problems (or disabilities) related to drinking are conceptually different from dependence symptoms. Dependence is a special kind of problem related to drinking which essentially represents an altered psychobiological state that drives drinking behaviour [7]. As such, one would expect dependence symptoms to be related more conceptually and empirically to each other than to other kinds of problems, which they are [8–10]. Also, other kinds of problems occur with vastly different frequencies; are often unrelated to each other; more so than dependence symptoms, they are subject to cultural norms as to what represents ‘a problem’ and are more affected by response biases of both the sufferer and the observer than with dependence [10]. Hence, it is found that measures of dependence are much more internally consistent than measures of problems [8,9]. Also, while problems are correlated with dependence, the correlations tend to be modest, as there are some people who experience dependence with few or no problems and vice versa. Further, in the biaxial concept problems and dependence exist on separate continua of severity, which is tricky for categorical classification systems such as DSM-V. Therefore, the biaxial concept in theory and practice is not a good justification for lumping problems and dependence together. All this seems to have been somewhat overlooked. In summary, the American Psychiatric Association (APA) should think very carefully before abandoning dependence. As a heuristic it has produced a wealth of data and helped greatly to improve our understanding and treatment of clinical phenomena over more than 30 years. I find it conceptually and linguistically helpful in communicating with other scientists and clinicians and achieving a common understanding. I also find dependence much less stigmatizing than most other terms such as ‘abuse’ and ‘disorder’ when talking to patients and trying to explain complex clinical and phenomenological issues to lay people: one can be dependent without necessarily having set out to become so. After all, we know that approximately 50% of the risk for developing alcohol dependence is inherited genetically [11]. Instead of pandering to an ill-informed stigmatizing stance of some parts of the medical profession, the APA should take the lead in seeking to reduce stigma for people who are dependent on substances for whatever reason. Retaining dependence in DSM-V and explaining why to non-psychiatrists would be a good start. How about dropping abuse, as it has proved of limited value, and having mild, moderate and severe dependence, with substance-related problems as associated but not defining phenomena? I think most of the field could live with that. CD is Professor of Addiction Psychiatry at King's College London and is in receipt of research grants to study alcohol dependence.