Commentary: Repairing a plane while it is flying – reflections on Rutter (2011)


  • Conflict of interest statement: This work represents my views as an individual rather than any official views of the ICD-11 International Advisory Group on Mental and Behavioral Disorders, of which I am chair, or the DSM-V task force of which I am a member. At the same time, I acknowledge with gratitude the testing of my views in multiple conversations and written exchanges with the dedicated colleagues who are members of both bodies. In addition, I am a member of the Novartis Science Board, but accept no fees for my work.

In this issue, Michael Rutter provides thoughtful observations concerning the Diagnostic and Statistical Manual of Mental Disorders, 4th edition (DSM-IV; American Psychiatric Association, 2000) and the International Classification of Diseases, 10th edition (ICD-10; World Health Organization, 1992) as well as recommendations for the current revision processes. While Rutter’s focus is on child psychiatric diagnosis and classification, many of his concerns, such as the excessive number of poorly validated diagnoses in both current manuals (Rutter, 2011), have implications for the organization of the DSM-5 and ICD-11 that go well beyond disorders of childhood.


Rutter does not bring a single reductive lens to his reading of the DSM-IV and ICD-10. His recommendations are pragmatic and rooted not only in his deep experience in the science and clinical practice of child psychiatry, but also more broadly in medicine. What ties the paper together is a desire to move beyond the once useful, but now enervated era of descriptive psychiatry. I have enormous sympathy with Rutter’s stance:

Frances (2009) concluded that: ‘it would be wise for us all to accept that descriptive psychiatry is a tired old creature’ and that a paradigm shift is essential. Neither ICD-10 nor DSM-IV is fit for purpose. Accordingly, the criterion that any proposed changes must include justifications is misleadingly one-sided. It is equally necessary to justify why an outmoded system should be retained. (Rutter, 2011)

Indeed I have expressed an even darker view of Frances’‘tired old creature’, i.e., that existing diagnostic criteria have become reified in the course of their wide use and have thus become damaging to scientific progress by imposing premature limitations on a still immature field of inquiry (Hyman, 2010). I would argue that there is a pressing need to open a clearing for new scientific approaches to neuropsychiatric disorders within the overgrown thicket of criteria promulgated by the DSM-IV and ICD-10. The challenge is to do so without damaging the existing clinical enterprise, without peremptorily scuttling the vast expanse of administrative uses to which DSM-IV and ICD-10 criteria are put, or getting ahead of the scientific evidence. Recent revisions of the DSM and ICD classifications were very conservative, thus, perhaps, inadvertently contributing to the current state of ossification. The accumulating evidence of fundamental problems with the classifications (Hyman, 2010) suggests that the current revisions must take more significant steps toward a new framework. Michael Rutter’s paper provides several penetrating avenues for change that still respect the limitations of our science. In short, psychiatry is not yet in a position to embrace a new paradigm. Like Rutter, I would very much like to see a classification based on pathophysiology as characterizes mature areas of medicine, but scientifically there is still a long way to go.

In the interim, Rutter’s (2011) recommendations would improve the clinical utility of the two major classifications while helping to make an opening to modern psychology, neuroscience, and genetics. He would do so by creating scientifically meaningful clusterings of disorders, such as a neurodevelopmental grouping; by de-emphasizing, if not abolishing, meaningless disorder subtypes; by removing minor distinctions that have given rise to a very large number of disorders; by introducing dimensional descriptions; and by disentangling distress and impairment from symptom lists.

Rutter (2011) acknowledges the well-recognized fact that the ‘modern’ DSM and ICD classifications have improved inter-rater reliability, but lack external validation. Unfortunately, this lack of validation did not impede the elaboration of a large number of diagnoses and subcategories. One unfortunate result of this historical orgy of diagnostic splitting is the high level of apparent comorbidity known to every clinician. Rutter points to the substance use disorders as a grouping in which patients can rapidly accumulate a rather startling number of diagnoses, perhaps as a result of a single diathesis. (He reasonably recognizes that a researcher might want a detailed list of all substances used, but whether each substance taken by a polysubstance user represents an independent disorder is another matter.)

Another result of the large number of diagnoses (many of which are rarely used in clinical practice) is a significant erosion of the clinical utility of the classifications. Rutter points out that within the autism spectrum disorders, neither clinicians nor researchers seem to find utility in the highly specified subtypes and often opt for the use of ‘not otherwise specified’ (NOS) diagnoses.

I am in full agreement with Rutter (2011) that most disorders represent quantitative deviation from health. The current DSM and ICD approach, which represents virtually all disorders as categories qualitatively distinct from health, are poor representations of clinical reality that result, inter alia, with thresholds that are arbitrarily set and may thus deny clinical care to many children who have not developed all of the typical symptoms (and durations) to meet criteria that were developed with adults in mind. Rutter wisely suggests that the DSM-5 and ICD-10 use a combination of dimensional and categorical diagnoses in a way that works well in other areas of medicine. For example, he points out the use of stage and grade in the practice of oncology, which superimposes a dimensional system on categorical cancer diagnoses. He appropriately recognizes that different disorders will require different conceptual approaches as the revisions attempt to wed clinical utility with improved validity.

Another important recommendation is the replacement of a separate grouping of disorders with onset specific to childhood (Rutter, 2011). In the existing classifications, this grouping served to separate many early-onset disorders likely to be seen by child psychiatrists from the remainder of the classification. A far better approach than classification by age of onset of symptoms is to create a neurodevelopmental cluster (and thus to carefully incorporate hypotheses about pathogenesis). Happily, with Michael Rutter’s help, this is almost certain to happen in DSM-5 and ICD-11. A related and critically important recommendation from the perspective of clinical utility is to describe the evolution of symptoms and signs over the life course. Rutter (2011) points out the current challenges of diagnosing adults with attention deficit/hyperactivity disorder (ADHD) because ADHD is not yet described as a developmentally based disorder that may extend over the life course.

The revision processes

One of the key points that Rutter makes is the need for harmonization between DSM-5 and ICD-11. It is hard to see the benefit of having two diagnostic manuals that result, at least to some degree, in different epidemiologies, different clinical trials inclusion criteria, and differences in the identification of patients in need of treatment. Rutter is not quite right that there have been no meetings aimed at harmonization to date, but the meetings have not had a formal scientific status, and have generally focused on administrative and procedural matters. By the time this commentary appears, a scientific meeting focused on harmonization will have occurred. The goals of a shared organization for the two classifications and of shared disorders and diagnostic criteria are important, but almost as important is Rutter’s additional advice. He would eschew abrupt, radical change that would be confusing to clinicians and highly disruptive for existing scientific data and prior regulatory decisions. As noted at the beginning of this commentary, however, he would also reject stasis as the appropriate fallback position: failures of the classification are serious and growing in number, most rapidly in genetics and brain imaging studies (Hyman, 2010). The harmony that may be most needed between the committees revising the DSM and ICD systems is a shared desire to move, responsibly, but with open minds toward classifications that invite scientific progress while achieving greater utility in the clinic.

Correspondence to

Steven E. Hyman, Office of the Provost, Harvard University, Massachusetts Hall, Cambridge, MA 02138, USA; Email: