Dimensions fit the data, but can clinicians fit the dimensions?

Krueger et al's paper1 is impressive and erudite. One might say it is too erudite, because the average clinician will find it difficult to anchor his or her clinical practice to the attachments offered. But the arguments put forward are scientifically incontrovertible; the data for almost, if not all, psychiatric disorders indicate that their dimensional description is nearer to truth than a categorical one.

The key section in this paper to most readers in practice is “clinical translation”, and here the work group is going to have to work extra hard. To what extent can the dimensional system be adapted, transformed, or forced, depending on your starting point, into clinical decision‐making?

There is an interesting historical parallel here. In the UK, in the late 1950s and early 1960s, there was what is commonly called the Platt‐Pickering debate, played out in the columns of The Lancet. This pitted the cerebral (dimensional) champion, G. Pickering, in one corner, against the clinical (categorical) pugilist, R. Platt, in the other. Although there were no apparent knockout blows, the debate was a riveting spectacle, illustrated by rapier‐like thrusts and counter‐punches by two austere but slightly irritable protagonists, always polite but each showing incredulity at the apparent stupidity of the other.

Their debate was over the classification of high blood pressure. Was it best regarded as a continuous variable2 or better described as two categorical populations, a larger one with normal blood pressure, and a smaller one with hypertension3? Pickering made the case that blood pressure is a continuously distributed characteristic with no clean separation between abnormal and normal. Platt insisted that those with very high blood pressure were a discrete group who represented the disease, hypertension, and that this fact could potentially be explained by genetic characteristics; he proposed a Mendelian dominant gene. This genetic theory was not supported and the Pickering power‐house swept away the old arguments: “The new view, for which we and our colleagues are largely responsible, is that essential hypertension represents a quantitative and not a qualitative deviation from the norm”4.

This resonates strongly with the current debate about dimensions in psychiatry. In the Platt‐Pickering debate, the clinicians – and, dare one say, The Lancet itself5 – were on Platt's side. After all, if he was right, it would make their job so much easier. Clean categorical diagnosis is always better than a dimensional fudge. What the Hierarchical Taxonomy of Psychopathology (HiTOP) investigators need to do is to show the clinician that there is genuine clinical value in the dimensional approach; that it is not a fudge. We have some clues. Thus, in the case of personality disorder, shortly to be a dimensional diagnosis in ICD‐11 and regarded as a genuine paradigm shift6, it is important to know that the more severe the disorder the greater its persistence and its impact on long‐term social functioning7.

But this only describes prognosis. Can a dimensional diagnosis help treatment? Again we have some encouraging findings. Sub‐clinical depression is not a formal diagnosis but it causes a lot of suffering. It is easily accommodated on a dimensional continuum and could be a suitable condition for treatment, and in a recent meta‐analysis there is some evidence, not yet strong, that psychological treatments are effective here8. Would this apply to drug treatment too? Probably not, and, for this to be appropriate, a higher point on the dimension would probably have to be chosen9.

Clinicians are naturally conservative when it comes to diagnosis and classification, and change is always resisted at first. But if it can be shown that there is definite advantage in a dimensional approach, that it can lead to better and more fine‐tuning of management, then it may win approval. It would probably be necessary to have parallel systems at first to allow comparisons to be made between categorical and dimensional approaches.

But there will be continuing concerns in clinical practice if there are not clear indications for decision‐making offered by the diagnostic system. Krueger and his colleagues rightly note that the recent elegant Research Domain Criteria proposals, whatever their value in identifying neurobiological constructs, do not help such decision‐making. Although the HiTOP team may go further and succeed in their aim of “connecting structure and process” in explaining psychopathology, the clinician at the coal‐face can only look on with bemusement at any system which, however well grounded in empirical science, still does not provide answers to key questions. When is apparent pathology within the range of normality? At what stage in a dimensional system of a major diagnosis is coercion justified in management? When is it right to regard co‐occurrence of disorders as comorbidity or instead as part of the same spectrum (e.g., anxiety‐depression)?

These are not academic talking points. Therapeutic advancement often happens by serendipity, but we also need to have a classification system that helps empirical science to focus on specific aspects of efficacy. So, instead of psychiatry using the current pot‐pourri of general interventions into heterogeneous populations giving equivocal results, we could look forward to “focused diagnosis‐specific gain”. The possible value of quinine in malaria was discovered by chance but, because malaria was a clearly identified disease, it was possible, even in the mid‐1860s, to show that all the cinchona alkaloids – quinine, quinidine, cinchonine and cinchonidine – were equally effective in treating the disease. Remember, at this time in history, malaria was identified by the same procedures that we use in psychiatry today.

The HiTOP investigators may feel it is far outside their remit to enter the therapeutic and other intervention arenas, but they need to be aware of their importance. The oldest and most successful classification in psychiatry has been the dimensional one of intellectual disability based on IQ. Although this has been rightly modified in several ways to take account of adaptive functioning, for more than a century this classification has allowed appropriate placement, support and management to take place for people in each of the dimensional groups.

What about the long‐term outcome of the Platt‐Pickering debate, which Pickering was generally assumed to have won? Currently the most common diagnosis in cardiology in the ICD‐10 classification is essential hypertension, so the Platt supporters may now claim some sort of victory. So, in 60 years hence, will it be seen that dimensions have triumphed or will psychiatric classification be essentially the same as now? If Krueger and his colleagues can come forward with more clinical meat to add to their helping of science, things will certainly change.