Difficulties for practice and multiple continua need more recognition: Commentary on Morris et al. “Should we promote alcohol problems as a continuum? Implications for policy and practice”

The authors put forward a compelling argument for continuum models of alcohol use. They argue that by shifting the focus in alcohol use disorders and harms away from discrete categories and towards a continuum approach we would be more accurately reflecting the realities of alcohol-related harms. Furthermore, this could help to reduce both stigma and also ‘neurobiological othering,’ which, they argue, is encouraged by industry bodies hoping to shift focus away from the harms resulting from consumption of their product, and instead focus attention on alcohol use disorders. This continuum approach is the approach of the social scientist looking at human behaviour and the emergence of problems with it—noticing the distributions of alcohol consumption and of the occurrence of problems from it are much more continuous distributions. As the authors note, this was the approach of the 1990 Institute of Medicine report, building on a tradition established by Genevieve Knupfer (both a sociologist and a psychiatrist), who measured ten dimensions of ‘alcohol problems’ in a general population (Knupfer, 1967). Anyone approaching alcohol from a public health perspective of diminishing problems in the population as a whole is going to end up with continuous curve distributions, not with a yes/no dichotomies.

We believe that the authors make some extremely important points, particularly those that focus on the reduction of stigma. However, we also believe that two of the limitations that they note towards the end of the piece will get in the way of this model being effectively implemented. Firstly, the assumption that alcohol problems could be appropriately represented on one continuum, and secondly not taking account of how such a model will work when facing practical realities on the ground.

On the first point, the authors note the heterogenous nature of alcohol problems in the first line of their abstract. They refer to Hasin and colleagues’ work (Hasin et al., 2013) that finds that alcohol use disorders (AUD) can be considered unidimensional; however, Hasin et al. cite works that found that AUD could be considered one factor OR two correlated factors (abuse and dependence). Morris and colleagues also point to Skogen et al.’s work (Skogen et al., 2019) on the AUDIT as evidence of unidimensionality; however, this was a study of the AUDIT, a ten-item screening measure, not the DSM criteria. Furthermore, other work has found AUDs to be multi-dimensional with a dependence and an abuse factor (Nelson et al., 1999) and even finding three factors in the AUDIT (Gecaite-Stonciene et al., 2021). These factors are often themselves characterised in terms of alternative dimensions; thus ICD-11 defines ‘alcohol dependence’ as consisting of any two of three criteria: impaired control over alcohol use; alcohol use taking priority over responsibilities, interests or self-care; and withdrawal or tolerance symptoms (Saunders et al., 2019). Even in the era when ‘alcoholism’ was being most fervently presented as a unitary concept, the scholar most strongly identified with it, E.M. Jellinek, had moved on to a multidimensional approach. After his experience at the World Health Organisation (WHO) of how differently psychiatrists in different societies characterised ‘alcoholism,’ he decided that it needed to be subdivided into four or five different ‘species,’ with the different conceptions identified by Greek letters (Jellinek, 1960; Room, 1984).

Turning to the second point, we agree that continuum models should be embraced by researchers and wherever else possible. We also agree it’s important from a health promotion and policy perspective to be very clear that alcohol problems are on a continuum (or continuums). However, like Morris and colleagues, we are concerned about how this would work with those who are in time-poor situations where they are dealing with multiple problems, not just alcohol, at a time. These are people who often need to make quick decisions about what needs to be done next and often these choices are themselves dichotomous—treat or don’t treat, refer or don’t refer, etc. Both DSM-5 and ICD-11 are categorical in nature because that is how medicine thinks in its pragmatic approach to how it deals with patients.

While Morris and colleagues do acknowledge these issues as potential limitations, it is worth fleshing out what could occur as this shift plays out. One potential issue is that while researchers and policy-makers move towards a continuum model, practitioners, due to the realities noted above, will continue to operate out of necessity for the most part in a more categorical way. This could then lead to a disconnect between research and practice. Ignoring the way in which people in face-to-face situations will make their decisions risks a situation where researchers and practitioners do their work in different ways, lessening communication between these groups. We would urge that all those working in this space are always mindful of the practical limitations that those on the front-lines face. Accepting that ‘alcohol use disorders’ continues as an overall clinical category, but with a variety of subcategories tied more specifically to verified paths of clinical intervention, could help. Furthermore, it is also important that there be recognition of a more multidimensional ‘alcohol-related problems’ construct, reflecting the more sociological/epidemiological tradition—including ‘problems’ of relationships and collectivities, different to the individual level at which clinical thinking operates.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

SC has been funded by a veski Inspiring Women Fellowship Grant.