HiTOP must meet the use requirements of the ICD before it can aspire to replace it

As described by Krueger et al1, the approach being taken by the Hierarchical Taxonomy of Psychopathology (HiTOP) consortium in attempting to elucidate the underlying dimensions of psychopathology is an important one. I agree particularly about the immediate importance of identifying connections between overt expressions of psychopathology and neural mechanisms and genomic variance, and believe that HiTOP has an important contribution to make in this regard.

At the same time, I do not believe that HiTOP can be successful as a sole approach. As with the Research Domain Criteria (RDoC) project promoted by the US National Institute of Mental Health (NIMH), it seems important not to oversell HiTOP or to pretend that it describes a classification system per se that will be capable of replacing the ICD or the DSM at any point in the immediate future. Although the NIMH has walked back its initial rhetoric2 to clarify that RDoC is actually a framework for research3, Krueger et al's paper makes the same mistake with HiTOP.

The paper is also marred by tendentious repetition of the claim that the ICD and the DSM are “consensus‐based”, “authoritative”, “political” classifications, in contrast to HiTOP, which is “empirical” and “scientific”. Such characterizations, although perhaps rhetorically useful in promoting a new approach, are actually inaccurate, as with the widely repeated and false characterization of DSM‐I and DSM‐II as psychoanalytic4, or the initial messaging about RDoC that characterized the DSM explicitly and the ICD by implication as responsible for the lack of dramatic breakthroughs in understanding the etiology of mental disorders and providing curative treatments2. This paper's similar denigration of “authoritative” as opposed to “empirical” classification systems appears to be based, thinly, on the facts that: a) the ICD‐11 and DSM‐5 (and RDoC) are institutionally sponsored; b) expert working groups developed the initial proposals for changes to the previous versions of the classifications; and c) there was an institutional demand for some degree of continuity across versions.

With regard to the first point, the development and maintenance of international classifications for health and the standardization of diagnostic procedures are core constitutional functions assigned to the World Health Organization (WHO) through international treaty by 194 member states. It is unclear why being a “consortium” without a clear formal authority structure or a responsible institution would make HiTOP inherently superior in relation to these tasks. With regard to the second point, an explicit charge of working groups for both the ICD‐11 and the DSM‐5 was to perform a rather rigorous analysis of the state of the current evidence. Krueger et al are correct, though, that the range of possibilities for transforming the classifications was to some extent limited by the adoption of a priori elements of the existing structure, such as the existence of separate groupings of mood disorders and anxiety disorders.

Most of the results presented in the paper in support of HiTOP's hierarchical dimensional models are based on a set of inter‐related techniques including taxometric analysis, latent class analysis, cluster analysis, and factor analysis. While these can be powerful and sophisticated statistical tools, they do not serve up the truth like Venus on a clamshell. They still require interpretation by human experts. The fact that HiTOP's authority structure and the specific criteria for evaluation are not transparent or explicit (at least based on this paper) does not mean that the evidence is not being synthesized and interpreted based on expert judgments.

For the WHO, a demand for explicit continuity between the ICD‐10 and the ICD‐11, at a minimum in the form of clear cross‐walking, is based on one of the ICD's main purposes – to provide a framework for the collection and reporting of health statistics – as well as on the need for longitudinal global, national and local health information. The governments of WHO member states have increasingly integrated the ICD into clinical processes and policies related to health care coverage and reimbursement, social services, and disability benefits5, and are also concerned about the continuity of health data and the continuous application of laws and policies. However, the paper suffers from a lack of familiarity with the functioning of the WHO and the purposes of the ICD‐11. Even though Krueger et al include the ICD‐11 in the sweep of their characterizations, all of the specific information in the paper about “traditional”, “authoritative” classifications is taken from the DSM‐5. This perhaps reflects the fact that only ten of the paper's 45 authors are from outside the US and none is from a developing country.

The WHO does not, in fact, “claim, through tradition and putative authority, that psychopathologies are organized into discrete diagnostic entities”. We have recently written explicitly and at great length about the better correspondence of dimensional approaches to the observed data3. The categorical nature of the ICD system is necessary for its application in global health statistics and in many instances for its use in clinical settings (e.g., eligibility, treatment selection). In most countries, provision of medical care other than routine examinations and preventive services is contingent on a qualifying diagnosis. Other relevant decisions are typically categorical (yes/no); even if the information that underlies them is dimensional, a threshold must be imposed. Inclusion of mental disorders in the ICD facilitates coordination with the classification of other disorders, as well as the search for related mechanisms of etiology, pathophysiology and comorbidity of disease processes. It also provides a basis for parity of mental disorders with other types of health conditions5. Mental disorders in the ICD‐11 must follow the same set of structural and taxonomic rules as the rest of the classification.

Within the constraints of a categorical system, the ICD‐11 has gone to considerable lengths to integrate dimensional constructs into the classification of mental disorders, which has been made possible by specific structural innovations as compared to the ICD‐10. One example that is discussed in the paper is the incorporation of a dimensional classification of personality disorders6, 7. Similarly, the ICD‐10 subtypes of schizophrenia (e.g., paranoid, hebephrenic, catatonic) have been replaced by a set of symptom ratings (e.g., positive symptoms, negative symptoms, cognitive symptoms) that may be applied to all primary psychotic disorders8. A category for anxious depression based on two correlated but distinct dimensions has been incorporated into the version of the ICD‐11 classification of mental disorders for primary care settings9. These innovations will push the ICD‐11 in the direction envisioned by HiTOP, but it is possible that they may be experienced as more complex than the purely categorical approach they are replacing, which may stimulate resistance among clinicians and health systems.

While the WHO does appear to be facing this challenge head‐on within the structural and taxonomic constraints of the ICD, there is a considerable amount that HiTOP might take on board in order to facilitate further transformations of this nature. Assuming that the correct dimensions have been identified, much work is necessary to translate group‐level research results into measures and cutoffs that are predictive at the individual level3. Although Krueger et al claim “greater relative utility and empirical accuracy of continuous conceptualizations of psychopathology”, very little work has been conducted aimed at developing tools that can be demonstrated as robustly valid as a basis for making individual health care decisions.

Any dimensional system that would seek to replace “authoritative” classifications would need to demonstrate that it is fit for purpose across the range of functions for which the world uses the ICD.

The views expressed in this commentary are those of the author and do not necessarily represent the official policies or positions of the WHO.