1. AER Centre for Alcohol Policy Research, Turning Point Alcohol and Drug Centre, Fitzroy, Vic 3065, Australia
    2. School of Population Health, University of Melbourne, Vic 3010, Australia
    3. Centre for Social Research on Alcohol and Drugs, Stockholm University, Stockholm 10691, Sweden. E-mail:
    Search for more papers by this author

We are in a mess, conceptually and terminologically.

We have a set of criteria for a condition with no clear relationship to need or demand for treatment

When the criteria for alcohol dependence are applied carefully to the US general population, more than half of those meeting the criteria are less than age 30 (re-calculated from [1]). Conversely, the average age at admission to treatment in the same society is approximately 37 years ([2], Table 2a). Adding abuse criteria into the new combined diagnosis will increase the age gap further [1]. This gap is interpreted commonly in terms of unmet needs for treatment. However, in my view it also reflects that the criteria for alcohol dependence are not attuned to the questions: ‘does this person need treatment, and by whose definition?’ and that these questions are not regarded as interesting in discussions in the United States in which I have participated about amending the criteria. To my mind, diagnostic criteria should be orientated to whether a case is suitable for treatment. It makes sense to measure ‘hazardous drinking’ in descriptive epidemiology, for instance, but the World Health Organization (WHO) nosologists decided that this category did not belong in ICD-10 because it was not in itself a disorder.

Related to this, we have a set of criteria which pay little attention to the threshold of severity at which the criterion should be coded as positive

Nosological discussions of severity focus on the level of ‘how many criteria are satisfied?’, not on the level of ‘is this behaviour or answer sufficiently substantial to be counted?’. The neglect of the threshold issue is functional in a couple of ways for the status quo. In the US context of diagnosis-related reimbursement for clinical services to which the DSM discussions are attuned, the lack of specification of a threshold for each criterion avoids possible second-guessing of clinical decisions by agencies which pay the medical bill. The peculiarities of a particular national health-care financing system should not be influencing so deeply how decisions about nosology are made.

The inattention to thresholds also lends itself to the ‘cookie-cutter’ approach of applying a common set of criteria across the whole broad range of psychoactive substances. With a close enough look, something resembling each of the criteria can nearly always be discerned for every substance; but this means that the substantial differences in the action of the substances and in human reactions to them are essentially excluded from the nosological system [3] and it means that the fact and implications of there being clear socio-cultural differences in what counts—in the threshold of attention for criteria [4]—are ignored.

We make nosological decisions across the broad range of psychoactive substances on the basis of presumed policy implications for particular subgroups of drugs

The primary argument O'Brien and colleagues make for restoring ‘addiction’ as a term and separating it from ‘dependence’ concerns the social handling of a particular class of drugs, the opiates. If the separation is not made, the argument goes: ‘pain patients in need of opiate medications’ will be denied or will deny themselves the medication [5].

Ironically, it was the impulse to make nosological decisions on grounds of social handling which brought us ‘dependence’ as the replacement for addiction in the first place. Looking beyond the United States, the back-story of ‘dependence’ is longer than O'Brien [6] tells us. Already in the 8th revision of the ICD in 1965, ‘drug dependence’ replaced ‘drug addiction’ as the nosological term (the ‘alcohol dependence syndrome’ came along in the next revision, 10 years later). This reflected the fact that the 13th Expert Committee on Addiction-Producing Drugs, reporting in 1964, had put forward ‘dependence’ as the term to cover its retreat from the last official attempt at a justification on pharmacological grounds for which drugs were under international control and which were not. The 1957 Committee had made a distinction between ‘addiction-producing’ and ‘habit-forming drugs’, with only the former needing international control. By 1964 this was clearly untenable, and ‘dependence’—already well established in a broader meaning than the narrow pharmacological one (e.g. [7]), and carrying its own stigmatizing baggage [8]—was the term adopted to bring all the substances back into a common frame [9].

Either the cookie-cutter approach of one set of criteria for all drugs should be abandoned, or we should not make decisions about diagnostic categories on the basis of subgroups of drugs.

We lump discrete phenomena into a single category

Here O'Brien moves in two directions at once. On one hand, he wishes to redivide ‘physical dependence’ from ‘psychological dependence’, and give them different nosological names. On the other hand, however, he is part of the current move to lump together the criteria for ‘abuse’ and ‘dependence’ into a single nosological category.

The various criteria as they are currently defined indeed overlap a great deal in clinical samples, and correlate highly in general population samples. Those in treatment are typically at the quite extreme end of the spectrum of drug-using behaviours [10], and are doing and suffering just about everything, as we have known for alcohol since Jellinek's early Alcoholics Anonymous (AA) study [11]. The high correlation in general population samples reflects, to a considerable extent, the high number of ‘negative matches’ between the criteria; the large number of respondents who are not using the substance at all, or using it sparingly, will contribute heavily to the correlations, if the usual statistics valuing negative and positive matches equally are used [12]. Another factor behind the high correlations is the deliberate blurring of conceptual domains in the criteria. ‘Continued use despite knowledge of a persistent problem caused or exacerbated by the use’, for instance, includes within a single criterion a domain of drug use, a domain of cognition and a domain of adverse consequences.

Even if a high overlap were to remain when these factors were removed, there is still substantial justification, in my view, for moving towards a way of recording substance use disorders in dimensional terms—such dimensions as amount and pattern of use; impairment of major social roles and adverse social consequences; and experience of loss of control/compulsion/craving—in a health-based system, we can count on adverse health consequences being recorded elsewhere in the nosology. It would be clinically useful in making a treatment plan, as well as useful for analysis and policy, to code dimensions separately in such a way.

To end up with a single ‘substance use disorder’, internally differentiated only in terms of the substance involved, means that there will no longer be a ‘junk category’, as ‘abuse’ has been de-facto, which can be used where systems of funders need to be satisfied with some code. It means that ‘drug use disorder’, whether or not it is called ‘addiction’, becomes the junk category, without a more substantively meaningful category differentiated from it.

We have built the nosology too much around a single society

This is not the fault only of the United States; in a US context, the divergences elsewhere may not be apparent. It is the duty of those more on the periphery to speak truth to power. The problem is epitomized by the growing dominance internationally of a nosological system, the DSM, designed by and for a particular society. Addiction, I have argued [13], can be seen as a ‘culture-bound syndrome’, in the terminology of anthropologically orientated psychiatry. Jellinek certainly came to see it in this way in his later work describing different ‘species’ of alcoholism [14], after experiencing at WHO the very different things that ‘Anglo-Saxon’, French and Finnish psychiatrists meant by alcoholism. The arguments of O'Brien and his colleagues are all in terms of presumptions about stigma and particular labels in English in American society. If such arguments belong in nosological discussions at all, they should have a broader linguistic and societal base, as Caetano exemplifies for Spanish and Portuguese in his commentary in this series [15].

Where do we go from here? I deal here only with immediate steps.

For a nondescript assortment, use a nondescript term

If we end up with the proposed miscellany of 11 criteria, it is confusing to call it ‘addiction’—‘use disorder’ is a suitably non-specific term. If people want ‘addiction’ back or to retain ‘dependence’, it should be used for a pared-down set of criteria which is focused on the terrain of the experience of loss of control, compulsion and craving. The choice of the term does not matter much; if a new, non-stigmatized term is used, it will soon become stigmatized. The stigmatization is from the concept and its social context, not from the term; as Lemert remarked, in societies such as the United States, lack of self-control ‘is one of the most vivid and isolating distinctions which can be made in a culture which attributes morality, success and respectability to the power of a disciplined will’[16].

Tackle stigma directly, not by seeking euphemisms

In my view, it is a mistake to change the diagnostic system to improve the lot of pain patients and pain doctors; again, changing the label will accomplish little, and the argument that they are different actually reinforces the stigma on those left behind (as in O'Brien's terminology of ‘abuse or aberrant behaviour’) [6]. Doing something about the stigma surrounding the field is an important and challenging issue [17,18], but changing the labels in the diagnostic system is not an effective long-term solution to stigma.

Declaration of interest