It was a great pleasure to read the thoughtful, incisive, and positive commentaries on my classification essay (Rutter, 2011). It is good that there is such a high level of agreement on all the basic issues. Indeed, I accept the validity of all the points of disagreement or criticism. Such differences as exist all concern either variations in emphasis or calls of judgment. In this response, let me try to summarize both the agreement and differences in order to draw inferences or implications for the challenges ahead.

First, there is universal recognition that all is not well in either DSM-IV or ICD-10. There is an absurdly large number of diagnostic categories (many of which lack validity), making it impossible for anyone to remember all the criteria for each one. This has resulted in an equally ridiculous, very high rate of co-occurrence of conditions (misleadingly labeled as ‘comorbidity’, because much of the supposed comorbidity is artifactual). There is a parallel concern over the high frequency with which there is resort to the vague diagnosis of ‘not otherwise specified’ (NOS). In addition, there is an appalling unvalidated inconsistency across diagnoses in the required duration of disorders. It may be concluded that the opportunity must be taken for a more radical revision of both the American Psychiatric Association (APA) and World Health Organization (WHO) classifications.

Second, however, as I argued, and as several of the commentaries emphasized, we must be careful to ensure that in our efforts to bring about much-needed change, we do not inadvertently make things worse either through creating unnecessary discontinuity with the past or by making clinical (or research) usage more difficult. With these two points as a background, let me now turn to each of the individual commentaries.

I have struggled, and failed, to find any point in Steve Hyman’s commentary with which I disagree (Hyman, 2011). I love the title of his commentary, ‘Repairing a plane while it is flying’, which very aptly highlights the challenges we face. His final sentence is absolutely on target in arguing that both the DSM and ICD committees should ‘move responsibly, but with open minds, towards classifications that invite scientific progress while achieving greater utility in the clinic’. He and I are agreed that: ‘no rational person is yet in a position to set forth a new paradigm to replace descriptive psychiatry, except in aspirational terms’. Hyman is undoubtedly correct in arguing that there needs to be a greater attention to the criterion of validity; the requirement of reliability is reasonable but it is not enough on its own. Although the knowledge needed to base diagnosis on pathophysiology is not yet available, we must seek to make classification friendlier to scientific developments.

Let me focus next on the much trickier issues of dimensions, and of the harmonization of DSM-V and ICD-11. There really can be no serious dispute on the evidence that, as in the rest of medicine, both the risks for mental disorder and the disorders themselves operate dimensionally, with purely arbitrary cut-offs between disorder and no disorder (Rutter, 2003). It would be ridiculous not to include dimensions in DSM-V and ICD-11 and we must all work to ensure that dimensions are included. That may be somewhat easier with DSM-V than ICD-11 (but I hope not). Of course, clinicians are correct in asserting that clinical decisions have to be categorical, but we must not allow that to mean that the ICD-11 and DSM-V ignore (or, worse still, reject) scientific considerations.

The harmonization of ICD-11 and DSM-V is highly desirable but involves quite difficult issues. With respect to ICD-10 and DSM-IV the goal was to bring the two classifications as close together as possible, to avoid inadvertent differences, but to accept differences stemming from contrasting principles. Thus, the two classifications differ with respect to the need for impairment (essential for DSM but treated separately in ICD), and the handling of comorbidity. Both these differences still apply. In addition, although the APA resists accepting it, there is a huge economic pressure for ICD and DSM to be different. DSM-IV was a hugely successful earner for the APA (in millions, not thousands) and the same will apply to DSM-V. By contrast, the WHO makes no profits from ICD. It would be naïve in the extreme not to accept that this will force major differences between DSM-V and ICD-11, undesirable though these are. There is one other crucial consideration. The timing of the two revisions (DSM-V having to work for a much earlier date than ICD-11) will also get in the way of harmonization. Nevertheless, Steve Hyman is clearly correct that, regardless of bureaucratic pressures, we must endeavor to bring about as much harmonization as possible, because it is needed both clinically and scientifically.

Let me turn next to my good friend, Danny Pine, who entitles his commentary ‘There must be something left about which to argue’ (Pine, 2011). Danny implies that, if we could discuss the issues around a table, most of our apparent disagreements would disappear, and I agree. Nevertheless, in the spirit of stimulating discussion, let me concentrate on the two areas of disagreement highlighted by Danny. First, he disagrees with my suggestion that clinical and research classification should be separable. Like Danny, I am not sure how far this is a true disagreement. Of course, it is essential that the two must be closely aligned; neither of us would wish to create a gap between clinicians and researchers. After all, the two of us are both. Danny rightly points out that the publication timeline of DSM-V will make it difficult, if not impossible, to bring the two properly together. The main point that I was trying to make, probably not clearly enough, is that research classifications may reasonably restrict attention to homogeneous groups, leaving many disorders undiagnosed. By contrast, clinical classifications cannot, and should not, do that. Also, the research diagnosis of conditions based on specific number of symptoms in list ‘A’ and list ‘B’ side-steps the evidence of the huge heterogeneity in the manifestations of the disorder in question. Danny argues, in my view correctly, that both clinicians and researchers benefit when forced to discuss disorders using a shared scheme. I agree, but would simply note that the rigidity of research ‘rules’ gets in the way of this discussion.

Danny’s second disagreement concerned my supposed suggestion that all anxiety disorder subcategories would be eliminated from DSM-V. As Danny went on to admit later in his commentary, actually I did not argue for the elimination of all anxiety subcategories, but rather for their reduction and better validation by empirical evidence. Possibly, given that this is Danny’s main research interest, even seated around a table, we might still have differences, although possibly not. However, what is certain is that we would both want such discussion to be based on research findings, and we would both accept that there is not, and cannot be, any single unambiguously correct answer.

Eric Taylor’s commentary ‘The language of diagnosis’, in typically thoughtful fashion, points to some of the unresolved controversies, starting with the complex and uncertain connections between symptoms and impairment (Taylor, 2011). He rightly notes that the parallels with internal medicine are problematic in the absence in psychopathology of a demonstrated pathophysiology. I entirely support his plea that we should be careful to include both symptoms and impairment in our clinical thinking, to be rigorous about both, and sensitive in the ways in which they are communicated. In connection with communication, he notes that medical terms, such as psychopathy, often carry unwanted additional meanings as understood by patients, caregivers and teachers. I agree. It was for this reason that in our own work with adoptees from Romanian institutions we altered the terms (Rutter & Sonuga-Barke, 2010). We need to retain what is good about diagnostic concepts but be careful to avoid language that may be stigmatizing. The same applies to his comments about problems that carry a high probabilistic risk for later conditions but with many false positives. As I pointed out, this is a major limitation inherent in the term ‘prodromal schizophrenia’. The implication that schizophrenia will inevitably develop (as meant by the term prodrome) is simply wrong.

The warning that we need to ensure that primary care categories are sufficiently discriminating (and therefore narrow) is well taken. But, I have yet to be persuaded that the diagnosis of attention deficit/hyperactivity disorder (ADHD) can be validly made by generalist clinicians on the basis of a 10- to 15-minute consultation. Nevertheless, I do agree that it is highly desirable to be able to diagnose ADHD in primary care. Moreover, I worry that any attempt to expect that primary care professionals could diagnose ADHD on the basis of a brief assessment may do more harm than good in encouraging overuse of mediation.

Yi Zheng supports the use of multiaxial and dimensional approaches, with the former particularly important in bringing the social environment into diagnostic considerations (Zheng, 2011). The importance of recognizing subthreshold problems is underlined and the need for some diagnoses to undergo further testing is emphasized. Concern is expressed that putting disorders needing further testing into a single group may impede their treatment. In my view, all depends on how the disorders are conceptualized and expressed and on where they are placed in the classification. ICD-10 has no such grouping so they are likely to be ignored. DSM-IV has such a grouping but it is in an appendix. So far as I am aware, there has been very little research to test these disorders and they are often not accepted as justifying treatment. Hence the pleas for a place in the main classification.

Luis Rohde helpfully focuses on how they apply to the high proportion of young people living in developing countries (Rohde, 2011). He applauds the development of an ICD-11 primary care classification applicable to children and hopes that DSM-V will follow suit. Like Yi Zheng, he notes the value of multiaxial classification. While he welcomes the deleting of a special section for disorders with an onset in childhood or adult life, he wisely worries that this would be beneficial only if there truly is an overt developmental perspective throughout the whole of the classification.

Finally, Yanki Yazgan, using the tantalizing title of ‘The naïve and sentimental diagnostician’, argues that a ‘fast and frugal’ prototype approach is both necessary and appropriate at the primary care level, because it is used to decide which patients need what sort of more specialized services (Yazgan, 2011). It may be that that would be sufficient at the primary care level were it not for the fact that, in many cases, ongoing care will have to be provided at the primary level. That requires what is described as a ‘sentimental’ approach – but explained as meaning ‘reflective’ or ‘pensive’.

It is also argued that impairment is essential for diagnosis because it brings the disorder to the attention of some professional. I agree that is often the case, but it will not be so if biomarkers can be developed to the point of providing, at an individual level, a valid indicator of a condition that is already present but which is not yet manifest through either symptoms or impairment. Despite that, if safe and effective preventive measures are available, services should be provided. Of course, these tricky diagnostic issues will only hit home when science has developed beyond its current point. In my essay, I was seeking to look ahead to challenges and not just proffer solutions for the present.

I am extremely gratified by the constructive questioning of some points in my essay. They enable me to conclude by reiterating that, like everyone else, I have no ‘correct’ answer to all the major classification questions. Rather, I have sought to outline my views on the principles that are needed in order to come to some acceptable solution. Necessarily, it will involve judgments when the scientific evidence does not all point in the same direction and inevitably it will require compromises when judgments have to take into account practical utility and varied societal circumstances, as well as science. What is really encouraging is the extent to which people are increasingly recognizing the complexity but also the need to pool ideas and thinking in a problem-solving mode so that the revised ICD and DSM classifications are moving ahead in the right direction, and hopefully in harmony.

Correspondence to

  1. Top of page
  2. Correspondence to
  3. References

Michael Rutter, Box 80, MRC Social Genetic & Developmental Psychiatry Research Centre, Institute of Psychiatry, De Crespigny Park, Denmark Hill, London, SE5 8AF, UK; Email:


  1. Top of page
  2. Correspondence to
  3. References