The changes in terminology proposed in the draft version of DSM are significant, but the semantic alterations are cosmetic compared with the more substantive changes proposed in the underlying criteria for the core syndrome within the broader category of substance use disorders. My comments on Charles O'Brien's [1] essay will therefore focus on the substance of the changes, rather than the semantics [2]. The following points summarize my concerns and recommendations.

Problems versus core syndrome

Whatever the core syndrome is called, there are good reasons to maintain the current bidimensional approach implicit in both DSM-IV and ICD-10. That approach differentiates between a core biobehavioral disorder (inter-related cognitive, behavioral and physiological dependence symptoms) and the medical and psychiatric problems that are associated with it [3–5]. Although dependence and problems tend to be correlated, the problems experienced by substance users should not be part of the diagnosis of the core syndrome, just as unemployment should not be part of the diagnosis of lung cancer.

The abuse category

The term ‘substance abuse’ has grown in popularity during the last 50 years, in part because it seems to fill a terminological gap between dependence and less harmful and compulsive substance use. Because of the potential for substance-related social and occupational problems to be influenced by economic and cultural factors, the tenth revision of the International Classification of Diseases (ICD-10) fills this gap with the ‘harmful use’ diagnosis, which emphasizes medical and psychiatric problems linked directly to the use of psychoactive substances [3,4,6]. In contrast, the abuse category in DSM-IV was based on a residual set of four items that have no real conceptual basis [6] (i.e. two dependence symptoms, the ‘hazardous use’ item and a criterion defining social and occupational problems). Research suggests that the ‘hazardous use’ criterion in particular contributes disproportionately to the prevalence rate of alcohol abuse [7,8] and is susceptible to cultural and socio-economic bias [9]. To incorporate this item and the ‘failure of role obligations’ symptom into the core syndrome is to: (a) confuse apples and oranges (i.e. problems and core symptoms) and (b) transfer the rate inflation and cultural bias observed in the DSM-IV abuse category to the DSM-V ‘addiction’ category.

Blind empiricism

One of the arguments for aggregating the DSM-IV abuse criteria with the DSM-IV dependence criteria is that they all show up on the same dimension when evaluated according to statistical clustering techniques. There are problems with this type of ‘blind empiricism’. First, the data on which these analyses are based were obtained from structured psychiatric interviews that were never validated at the item level. These interviews were designed to measure diagnostic classification at the disorder or syndrome level, not at the symptom level, so the analyses themselves could be subject to measurement error. Some items, such as those measuring tolerance and withdrawal, are very poorly worded and subject to misinterpretation, especially by young adults who are frequent binge drinkers, thus explaining why some epidemiological surveys have produced what appears to be an epidemic of alcohol dependence among college students and teenagers [8] who do not have a history of chronic drinking. The problem is not necessarily with the criteria, but rather with the operational definitions of the items developed to measure them in psychiatric interviews. If the criteria are not revised (re-worded) to prevent poor operationalization of these symptoms (e.g. rapid initial tolerance confused for an acquired tolerance developed over long periods of very heavy daily drinking; hangover symptoms confused for withdrawal tremors, etc.), there is a real risk that the proposed criteria will further inflate already inaccurate prevalence estimates, especially as these two ‘physiological dependence’ criteria are now flagged for further specification.

Severity specifiers

The threshold of two symptoms to meet criteria for ‘moderate addiction’ seems so low that it will grossly inflate estimates of a ‘treatable’ psychiatric disorder. This is not a trivial matter, as these data are cited frequently by National Institutes of Health (NIH) directors and the Substance Abuse and Mental Health Services Administration (SAMHSA) to support funding decisions, and are also interpreted as an indicator of resource needs.

Continuum idea implies progression

Perhaps the most serious problem with the elimination of the abuse category and the labeling of the core syndrome as a continuum of ‘addiction’ is that there is only one dimension worth noting from a psychiatric perspective. With no other alternative than acute intoxication, individuals who use psychoactive substances infrequently without major indications of dependence/addiction are unlikely to have their substance-related problem diagnosed in emergency rooms, hospital wards and out-patient clinics. To the extent that there is now strong evidence that early intervention is effective with non-dependent substance users [10], it would be remiss not to recognize the need for these interventions in the American Psychiatric Association diagnostic system. Epidemiologically, some estimates suggest that up to 50% of the problems associated with alcohol misuse are experienced by individuals who do not meet criteria for alcohol dependence [11]. This ‘prevention paradox’ implies the need for diagnostic categories that recognize problematic and hazardous substance use so that public health measures now shown to be effective can be applied and billed to third-party payers.

The following recommendations follow from the points raised above: (i) focus the core syndrome (dependence/addiction) diagnosis on a more carefully worded set of operational definitions, especially reformulations of tolerance, withdrawal and salience of substance-seeking behavior; (ii) eliminate highly redundant criteria from the core syndrome list, and balance it with equal numbers of cognitive, behavioral and physiological criteria; (iii) create a harmful use category using criteria similar to those in ICD-10, which emphasizes psychological and medical harm, such as accidents and injuries; (iv) eliminate the hazardous use item from the diagnosis of ‘addiction’ in DSM-V because of its socio-economic bias and its tendency to inflate prevalence estimates; and (v) conduct further research during the DSM-V field trials on the semantic meaning of the terms ‘dependence’ and ‘addiction’. The choice of terminology, like the choice of criteria, should be based on empirical data, not speculation.

Declaration of interest



The author is grateful to John Higgins-Biddle, Roger Meyer, Judith and Ed Bernstein and a number of other people who made contributions along the way.