Science and CFS

Is it good science to classify disorders by their causes? Well, yes, provided that widely accepted causal theories are currently available. Causal laws are at the heart of the scientific paradigm; and in medicine, causal theories of disease are the royal road to good clinical care.

Correspondingly, though, it is bad science to classify disorders by their causes in the absence of a widely accepted causal theory. Consequently, Sykes is right to criticise those who wish to classify CFS as a mental disorder on the grounds of presumed psychological causation, for there is no widely accepted causal theory to support this. He is wrong, though, to base his criticism on the absence (in most cases) of evidence of psychological causation, for this objection to causal attribution, like causal attribution itself, depends on the availability of a widely accepted causal theory.

The recent history of psychiatric classification provides a cautionary tale. It is well known that our current classifications of mental disorders, the DSM and ICD, are based primarily on symptoms rather than causes. This is no accident. It follows the recommendations of a report, commissioned by the World Health Organization (WHO) from the British psychiatrist Erwin Stengel (1959) in response to the very poor uptake around the world of the mental disorders section of WHO's first attempt at an international classification of diseases (chapter V of ICD–6, World Health Organization, 1948). Stengel was directly influenced by the American philosopher of science, Carl Hempel (1961). Hempel pointed out that all sciences go through a descriptive stage before developing causal theories. Correspondingly, Stengel's diagnosis of the failure of the mental disorders chapter of ICD–6 was that it was based on premature (particularly psychoanalytical) causal theories which had not gained general acceptance. What was needed, therefore, was a classification which properly reflected the descriptive stage of the development of scientific psychiatry, namely one based primarily on symptoms.

We are set for a new ‘open season’ of debates about causes v. symptoms in psychiatric classification with the launch in 2001, by WHO and the APA, of revision processes which will lead to new editions, respectively, of ICD and DSM. Currently, the basis of causal theories is likely to be biological rather than psychoanalytical. But the condition/cause distinction, implicit no less in modern debates about classification than at the time of Stengel and Hempel, is likely to remain a useful tool for clear thinking about disease classification. Certainly, it remains a useful tool for clear thinking about diagnosis in everyday clinical practice.

Clinical practice and CFS

In drawing on the experience of Westcare UK as a registered charity that provides professional psychological and other help to patients with CFS, Sykes reminds us of the extent to which such patients find themselves the butt of bad diagnostic practices. Psychological causes are attributed on the sole basis of an absence of identifiable bodily causes; causes and consequences are conflated (for example, in respect of the role of depression); and ‘all in the mind’ is used, not to access appropriate treatment, but as a label for naming and shaming.

Our response to bad diagnostic practices, however, should be good diagnostic practices based on good science. And in this instance, good diagnostic practice, to the extent that it is based on the good science of the Stengel/Hempel-inspired ICD and DSM, means keeping the condition distinct from its causes.

This distinction is reflected in the standard approach to diagnostic formulation. Here, the diagnostic possibilities (defined primarily by symptoms and signs) are listed separately from the possible aetiological factors (see Reference Gelder, Gath and MayouGelder et al, 1983). Keeping the condition and its causes distinct in this way thus allows us to consider, separately, the bodily and mental signs and symptoms, and the bodily and mental causes of those signs and symptoms, for each patient. In the present state of our knowledge, this remains important for clear thinking even in ‘organic’ psychiatry, i.e., in dealing with conditions, such as Alzheimer's disease, for which the underlying brain pathology is reasonably well understood (Reference LishmanLishman, 1978).

Sykes makes the interesting observation that if patients with CFS are told that their illness is physical rather than psychological, they are more, not less, willing to consider psychological factors in the aetiology of their condition.

I suspect that there may be some patients, if not with CFS then certainly with other conditions, for whom the reverse is true. In a study completed at Warwick University, UK, Tony Colombo (a social scientist) and his colleagues are finding that patients with schizophrenia are broadly divisible into those whose perspective is predominantly biological and those where it is predominantly psychosocial (Reference Fulford and NordenfeltFulford, 2001; Reference Colombo, Bendelow and FulfordColombo et al, 2002). The traditional diagnostic formulation, in separating the condition from its causes and considering biological, psychological and social components of each, thus allows us, as experts, to match our general knowledge appropriately to the particular perspectives of individual patients.

Conceptual models and CFS

The biopsychosocial model, as it is usually called, is sometimes thought to have made the body/mind distinction in medicine otiose, and with it the biological and psychosocial clinical perspectives just discussed. As Sykes spells out, this is plainly nonsense. The ‘biopsycho-‘ part of the model directly depends on the body/mind distinction. The ‘-social’ part adds the social context, which has been relatively neglected by philosophers of mind.

Sykes, then, writing from the perspective of his background in philosophy, is right to remind us that we are stuck with the ‘mental illness’/‘bodily illness’ distinction, at any rate until someone comes up with a solution to that mother of all philosophical problems, the mind–body problem.

As already noted, there is no problem with adopting the causal distinction in principle. However, in practice, it is subject to all the difficulties we should expect, if we follow Hempel, in a science which is at a descriptive stage of development. Where there are competing causal theories, how should the ‘medical specialist’ (step 7 in Sykes's argument above) choose between mental and bodily theories? Similarly, in a multi-causation model, how should the medical specialist decide what is the ‘predominant’ cause, as Sykes points out would be necessary?

The ‘predominant’ cause, at the present stage of development of psychiatric science, tends to be interpreted according to the theoretical orientation of the person making the judgement in a particular case (Reference Tyrer and SteinbergTyrer & Steinberg, 1993: chapter 5). Schizophrenia, for example, tends to be regarded as a brain disease by ‘biological’ psychiatrists, as a psychological disorder by psychologists, or as a product of adverse social factors by social scientists and anthropologists (as in cross-cultural psychiatry, for example). All three groups acknowledge the relevance of all three kinds of causal factor, but each group regards its ‘own’ factor as the most important. Nor is it likely that such differences of emphasis will be easily resolved. Even such relatively clear-cut causal attributions as the ‘cause of death’, are subject to widely differing interpretations (Reference Lindahl and JohanssonLindahl & Johansson, 1994).

Therefore, unless causal attributions are made, whether by medical specialists or others, on the basis of a well-established causal theory, they will remain highly subjective and hence vulnerable to just those stigmatising abuses against which Sykes has campaigned so vigorously.

Anyway, the required distinction between mental disorder and bodily disorder can be drawn relatively straightforwardly at the level of the condition itself, i.e. in terms of symptoms, rather than at the level of causes. Moreover, if drawn at the symptomatic level, the distinction is entirely consistent with Sykes's proposal for CFS. Thus, where physical medicine is concerned with symptoms involving bodily functions (bodily sensations, such as pain, nausea, paralysis and blindness), psychiatry is concerned with symptoms involving the ‘higher’ mental functions, such as emotion, desire, volition, belief and motivation (Reference FulfordFulford, 1989: chapter 5). Characterised as it is by physical exhaustion, bodily pain and so forth, CFS is, consistently with Sykes's proposal, at least as much bodily as mental at the symptomatic level.

Taking care with causes

In psychiatry, then, we have scientific, clinical and conceptual reasons for taking care with causes. I do not want to be misunderstood here. The point is not that we should be barred, in principle, from distinguishing mental from physical disorders on the basis of causes: we do, and we should (in the case of dissociative disorders, for example). Neither is it that distinguishing mental from physical disorders at the level of the conditions themselves (i.e. at the level of symptoms) will always be unproblematic – on the contrary, the ‘appetitive’ disorders, for example, provide teasing examples at the border between the mental and the physical (e.g. anorexia). Rather, the point is that at the present stage of our knowledge, following the Stengel/Hempel line, we should be chary of making causal attributions do all, or even much of the work of distinguishing between mental and physical disorders.

Those who classify CFS as a mental illness on grounds of presumed psychological causation do, I believe, try to make causal attributions do too much work. However, in making the absence of evidence for psychological causation the grounds of his objection to this, Sykes, too, tries to make psychological causation do too much work. For the proposal and objection both depend, in equal and opposite ways, on a widely accepted theory of psychological causation, and we lack this at the present stage of our knowledge.

Better, therefore, to stick with the Stengel/Hempel line: to define CFS descriptively, by its (mainly bodily) symptoms, and consider biological, psychological and social causal factors separately on a case-by-case basis. This is a more complicated approach, certainly, but it is true to the descriptive stage of the development of psychiatric science; it provides a framework for clear thinking clinically (as in the traditional diagnostic formulation); and it offers a robust conceptual model for countering the abuses to which CFS and psychiatric patients alike are subject, as Sykes has done so much to show.