RESPONSE TO COMMENTARIES

We thank Borges [1], Caetano [2], Hasin [3] and Teesson et al. [4] for their comments. Our principal points—that consequence-related criteria are accessory to substance use disorders (SUDs), and introduce systematic biases to diagnosis—were meant to help frame a future research agenda, but not present the fruits of a completed one, as Borges [1] and Teesson et al. [4] desired. Our brief discussion of possible alternatives to consequences was meant to highlight novel candidate criteria that deserve further study (e.g. [5–7]).

We do not support a bi-axial approach to SUD diagnosis, such that consequences form a separate illness category. To the contrary, we agree with Hasin [3] on this point, and have long called for the use of a single criterion set to diagnose SUDs [8]. Our primary concern is that the items in this single criterion array should minimize, to the extent possible, systematic biases due to age, gender, culture and socio-economic status, as these degrade diagnostic validity.

If a symptom loads on a unidimensional SUD factor, this is not by itself good evidence that it is valid. If this were the case we could include many non-substance externalizing behaviors as SUD symptoms [9], but this does not make sense conceptually or practically. Diagnostic validity must also be based on conceptual coherence and evidence that a criterion performs similarly in various demographic groups. There are already considerable data showing the bias produced by consequence-related criteria. Substance-related medical problems are developmentally biased [10,11]. Both ‘hazardous use’ (e.g. intoxicated driving) and ‘legal problems’ show a great deal of demographic and policy-related bias [12–14], which compromises their validity. Retaining hazardous use confers on driving under the influence (DUI) policy the ability to strongly influence the prevalence of a mental disorder, even though the two should not be conflated [15].

Caetano [2] and Hasin [3] note that non-consequence symptoms can be influenced by context, and we strongly agree. For example, craving is influenced by cues [3]. However, our use of the term ‘context’ was meant to emphasize systematic biases that are unrelated to basic addictive processes. In contrast, craving is probably a good example of an addictive phenomenon that occurs regularly among most heavy drug users, across drug, country, gender culture, race and age. The same cannot be said for consequences. Similarly, tolerance and withdrawal are moderated by genetics [3], but this is a far cry from the numerous moderating influences on, for example, intoxicated driving. Indeed, the fact that tolerance and withdrawal occur in non-human animals indicates that these addiction constructs can be measured relatively independently of human cultural influences.

As noted in our commentary and reaffirmed by Teesson et al. [4], symptoms defined by ‘continued use despite’ consequences can indicate compulsive drug use. However, the way in which these symptoms are defined is problematic: compulsive use is implied rather than directly described, the compound nature of the criteria makes it easy for interviewers and respondents to focus on consequences rather than compulsion, and the criteria are conditioned on problem recognition and the attribution of a causal role to drug use. More generally, compulsive drug use can occur in the absence of consequences. There is a need for better assessment of compulsive use without consequences in our diagnostic systems.

Hasin [3] states that DSM-5 provides a ‘standard set of reliable, valid, evidence-based criteria’. Regarding reliability, current evidence is limited [16], but existing data are cause for concern [17]. Regarding validity, research has identified important limitations of both consequence-related and non-consequence-related DSM-IV SUD symptoms [8], yet none of the symptoms that were retained in DSM-5 were revised in any way. We continue to believe that diagnostic systems should become better aligned with modern neurobehavioral theories of addiction, although Caetano [2] is correct that this is a difficult and complex undertaking. We should attempt to develop diagnostic systems that better distinguish addiction from imprudence, culture, and public policy.