“Throwing out the baby with the bathwater”? Conceptual and methodological limitations of the HiTOP approach

More sophisticated explorations of the higher‐order dimensional and hierarchical structure of psychopathology have become an exciting complementary way towards developing an improved classification of mental disorders and reducing artefactual comorbidity.

The impressive work of the Hierarchical Taxonomy of Psychopathology (HiTOP) Consortium with their mission paper1 provides evidence for considerable advances as compared to previous suggestions, and underscores the potential of such approaches not only for improved future classificatory models with increased utility for research and practice, but also for the development of improved psychometric assessment instruments for psychopathology. However, as impressive such an approach might appear at first sight, there is a need of pointing out several limitations that caution against the use of this model.

On the conceptual level, we emphasize first of all that comorbidity is not “a problem”, but a clinical characteristic of patients meaningful for treatment and management2. The belief that people suffer from only one underlying condition is implausible and misleading. The value of the HiTOP Consortium approach might be in reducing a certain degree of what has been called “artefactual” comorbidity, due to overlapping criteria in our current classification systems.

Second, the suggested hierarchical structural model has a serious limitation: it is based almost exclusively on traditional assessment instruments (dimensional scales, interviews) from cross‐sectional studies. Leaving aside the vast array of inherent general psychometric problems, we highlight that such scales merely reflect a subjective‐verbal “snapshot” picture of the level of symptom‐distress that a person reports at the time of investigation. As essential such a snapshot might be for a first “impressionistic” step of a syndromal diagnosis, it certainly does not allow to decide on a diagnosis relevant for treatment without taking into account the patient's history (e.g., depressive syndromes cannot be equated with diagnoses of major depression or even of any affective or any mental disorder).

Third, the HiTOP approach does not grab appropriately the nature of mental disorders as dysfunctions – up to now insufficiently understood – of basic psychological processes as well as associated “perturbations” in brain functions at the cell and systems level3. The former are centrally involved in the behavioral, cognitive‐affective and somatic symptom processes currently used to define mental disorders. The latter “perturbations” can be best described as various types of fluctuating dysfunctions in complex structural and functional neural circuits involved in information processing and emotion regulation.

The identification of common causal pathways is of core relevance for an improved diagnostic system. They allow identifying the factors and mechanisms responsible for the onset, progression and maintenance of mental disorders. Proposed models based on such mechanisms provide guidance for improved research strategies and the derivation of improved interventions, targeted to interrupt the causal pathways3.

Promising examples come from psychosis research. In a clinical staging framework, the at‐risk or symptomatic state of a patient can be evaluated to derive tailored interventions spanning from primary selective prevention in asymptomatic subgroups (stage 0) and high‐risk subjects (stage 1), over early treatment in first episode (stage 2) or relapsing psychotic patients (stage 3), to maintenance treatment in unremitting patients (stage 4)4.

Such frameworks also exist for other facets of psychopathology such as anxiety, depression or substance use, providing specific guidance on early targeted interventions. The “symptom progression ‐ comorbidity development” model3, 5 emphasizes the early signs and symptoms of fear and anxiety in the development of psychopathology and a staging based on “comorbid” escalations from circumscribed manifestations in childhood to more complex diagnostic constellations (multiple anxiety disorders, comorbid depression and substance disorders) later in adolescence or adulthood. Besides a range of vulnerability factors at various levels and in different developmental periods, the initial psychopathology itself entails a causal cascade (e.g., increasing demoralization and inactivity due to avoidance promoting depression)6. This model has direct implications for therapeutic and preventive interventions.

Therefore, the first caveat of higher‐order taxonomies such as the one suggested by the HiTOP Consortium is that they are at best a complementary piece of descriptive evidence that might prove useful in reducing artefactual comorbidity. But they do not reflect the true dynamic developmental nature of mental disorders and might even be an obstacle for developing improved targeted causal interventions.

Regarding methodological constraints, we do not refer here to the numerous mathematical and statistical limitations of the higher‐order dimensional and hierarchical approaches that call for caution7, 8. Beyond these, the strongest evidence against such models comes from prospective‐longitudinal investigations, revealing the instability of the assumed higher‐order structure and spectra over time7. Along the developmental axis, the structure of higher‐order dimensions changes significantly, both within factors and across spectra. The assumption that this instability might be due to a limited reliability of assessments is implausible and would actually also argue against such higher‐order models in general.

Furthermore, the statement that dimensional measures are advantageous over categorical data is trivial. They simply provide more information and are thus preferable in any approach9. Assuming that hierarchical structural models based on dimensional data may lead per se to an improved classification of mental disorders and “solve the problem of comorbidity” is like “throwing out the baby with the bathwater” and obscures important issues, given the underlying assumptions and the lack of developmental considerations. This does not invalidate the additional utility and the potential of such approaches, but suggests that these models are at best complementary to other principles and sources of evidence.

Undoubtedly, as compared to previous simpler models, the HiTOP model has increased in breadth and specificity (e.g., spectra for thought disorder and detachment). However, the extensions also cause new inconsistencies, such as enhancing the “distance” between internalizing and externalizing dimensions, although externalizing disorders might involve preceding internalizing pathways (and vice versa). Moreover, as attractive and impressive the visual depiction of a new taxonomy of psychopathology may be, using new words for old ones might increase the risk that already established research findings lack consideration in the future.

Further, “somatoform” diagnoses (dismissed in DSM‐5) are reintroduced without explaining the rationale. This particular cluster also serves as an example for the difficulty – even cross‐sectionally – to find a coherent general structure of psychopathology. Somatoform syndromes are differentially (i.e., by gender and age group) associated with a broad range of conditions which are spread out in the HiTOP model (anxiety, psychosis, hypomania, post‐traumatic stress disorder, and many other diagnoses not mentioned in the framework)7, which complicates the implementation of the model.

To conclude, higher‐order dimensional and hierarchical models of psychopathology such as the ambitious HiTOP model are at best a complementary way towards developing an improved classification of mental disorders for research and practice. Their potential value lies in reducing artefactual comorbidity and deriving improved cross‐sectional psychometric assessment instruments.

However, HiTOP provides little specific guidance towards our ultimate goal, namely, a classification of mental disorders based on causal factors and mechanisms involved in the first development of psychopathology and its progression over time. Its inherent weakness remains the overemphasis on cross‐sectional psychopathology and the neglect of dynamic developmental pathways and differential diagnostic issues relevant to treatment and management.