Defining Dependence: Describing Symptom Clusters Versus Central Constructs

The paper by Li et al. [1] reminds us that the publication of ‘Alcohol dependence: provisional description of a clinical syndrome’ by Edwards & Gross [2] was a watershed event for understanding alcohol and drug dependence [3]. We may forget that prior to this publication, definitions of dependence were vague and idiosyncratic. For example, at that time the existing World Health Organization definition was ‘a compulsion to take the drug . . . in order to experience its psychic effects and sometimes to avoid the discomfort of its absence’[4]. The existing Diagnostic and Statistical Manual (DSM) definition for substance dependence was tolerance and withdrawal and for alcohol dependence was ‘pathological use . . . that causes impairment in social or occupational functioning’[5]. The National Council of Alcoholism definition was ‘tolerance and physical dependency; or pathological organ changes’[6]. The proposal of specific, operationally defined criteria by Edwards & Gross was consistent with what was happening in the rest of psychiatry [7]. The major advantage of such operationalization was that it produced a ‘conjecture capable of refutation’[7]. The many studies since have shown this ‘provisional’ description was, at least for alcohol, sedatives and opiates, amazingly accurate [8]. In a follow-up article in 1986 Edwards stated ‘it is legitimate to go beyond the question as to whether a dimensional syndrome exists, to an exploration of theoretical questions . . .’[9]. Unfortunately, consensus on the symptoms of the dependence syndrome has not led to any greater consensus on the theoretical basis of dependence [10]. There are a plethora of biological and psychological explanations and all have problems [10]. What ‘theoretical question’ needs to be addressed to fix this? I believe we need to produce a very concrete, very clear, statement of the central construct(s) of dependence. If a syndrome does not have essential element(s) then it may not be a single disorder. Central constructs can be derived by induction from a good psychometric description of the phenotype. In fact, improved descriptions of phenotypes have often led to the discovery of etiologies in medicine. Central constructs can also be derived from a known or hypothesized biological or psychological cause for the disorder. This is less common in medicine and psychology. Typically it is the interplay between advances in defining the phenotype and advances in basic psychology or neuroscience that suggests central constructs. Defining central construct(s) that are concrete, are well accepted, cannot be misunderstood and have adequate coverage is not easy. Most important is that the constructs be essential; i.e. if they do not occur, then one cannot use the term dependence. Of course, constructs can manifest in different ways in different people and different times, but nevertheless the central constructs should be present. The constructs may not necessarily be sufficient. For example, certain environments may be necessary, e.g. those that allow drug access. What are the essential construct(s) of dependence? Although Edwards & Gross did not state it explicitly in their first article, a few years later Edwards stated it is when ‘the use of a given psychoactive drug, or class of drugs, is given a much higher priority than other behaviors’[11]. The most recent DSM-IV says it is when ‘the individual continues use of the substance despite significant substance related problems’. Interestingly, the International Classification of Diseases (ICD-10) does not cite a central construct but rather appears to state the criteria are the disorder [12]. Other central constructs cited include physical dependence, impaired control of drug use, psychoactivity, the drug serving as a strong reinforcer, etc. [10]. Some may say all these constructs are really the same thing, but are they? For example, if one smokes 20 cigarettes/day without becoming sick, has withdrawal when he tries to quit, has tried to quit 10 times and cannot do it, but had no medical problems from smoking, he/she meets DSM criteria for dependence but not the central construct of DSM. Clearly, we need to be very precise in defining constructs. If we cannot agree on central constructs, then perhaps we should ask articles on the psychometrics of dependence to either define clearly their a priori central constructs or discuss what their findings suggest as central constructs. In summary, Edwards & Gross [2] made an important breakthrough in developing consensus about dependence symptomatology. As Griffiths et al. proposed [7], we now need to move beyond studying how symptoms cluster and on to the, perhaps harder, task of agreeing on the central constructs of dependence. Preparation of this commentary was funded by Senior Scientist Award DA-00490 from the US National Institute on Drug Abuse. I thank Tim Baker for his helpful comments.