Finding
Thus Jablensky's critiques, though accurate, are in a sense not the real issue.
Paper
Variation and validation: the example of schizophrenia
Abstract
A. Jablensky's paper1 is a clear and powerful analysis of issues related to validity and utility in the psychiatric area. We might approach the issues he raises, focusing on the example of schizophrenia, from four somewhat different perspectives: one accepting in a provisional way classical DSM schizophrenia, a second tolerating and using variation and “fuzziness” or more precisely “polytypicality”, a third considering the relation of validity and utility, and a fourth proposing that research be conducted both on the descriptive DSM‐5 account as well as on some contrasting and more etiological models of schizophrenia. Though I would agree that a disorder such as schizophrenia may ultimately need to be further analyzed, and perhaps divided or reconfigured as several different disorders, there are some empirical studies that do support the aggregation or “hanging together” of the diagnostic criteria for this and other traditional disorders. One approach to seeking objective confirmation of a common pathology underlying typical DSM‐clustering in six mental disorders was pursued by Kendler et al2 using latent class analysis. More recently, Derks et al3 used a similar approach and reported that, when they combined factor analysis and latent class analysis, 85% of the patients receiving a DSM‐IV diagnosis of schizophrenia were assigned to the Kraepelinian schizophrenia class. My inference from these studies is that classical schizophrenia is a reasonable first approximation to diagnosing and beginning to develop a treatment plan for such patients, and that, based on the widespread acceptance of the DSM and ICD accounts, the classical picture also retains reasonable clinical “utility”. But let us acknowledge the variation in several different dimensions (rather than discrete categories) in the schizophrenia area that Jablensky emphasizes. Should it be of that major a concern? Is there some way we might embrace the variation and fuzzy boundaries? The notion that the entities that are fundamental in a scientific area need to be discrete and separable is an idea that works well in some sciences such as physics and chemistry. But these types of entities are rarely found in biology, where more “polytypic” or “polythetic” concepts reflect the variation in the entities that are fundamental in that science4. And medicine and psychiatry are similarly affected by variation5. Thus Jablensky's critiques, though accurate, are in a sense not the real issue. I would suggest that we begin by accepting this polytypicality and then decide how to deal with it in psychiatry. Perhaps it would be better to deal with variation as doctors do with blood pressure and blood sugar, using a comparatively few prototypes for hypertension and diabetes, with different thresholds for different modes of treatment and employing sliding scales. In a way, the psychosis symptom severity five point scales in the DSM‐5, though not required for schizophrenia, are a start in this direction. Jablensky himself, in his fuzzy set analyses, has offered a more technical way to advance that approach6. For prodromal and early episode forms of schizophrenia, we might also consider proposals for a staging system7. A recurrent theme in Jablensky's paper is the distinction between validity and utility in psychiatry. Jablensky indicates that issues of utility (and reliability) seem to be progressing satisfactorily, though the same is not the case for validity. But it would seem that the distinction he urges between utility and validity is not quite as sharp. In fact, Jablensky himself cites Jaspers, reminding us “validity” may be a Kantian type of idea. There, Jablensky notes that, though validity is a most elusive endpoint, it may well be best approached by “progressive refinement of the utility of the diagnostic concepts and tools”. This type of approach seems also to be supported by the way the psychiatric “validator” literature has developed in the DSM context8. The set of validators that were putatively used in all of DSM‐5, and which are very likely to continue being used in DSM 5.1, include three high priority “predictive validators”. These are “diagnostic stability, course of illness, and response to treatment”. This set of validators evokes the notion of predictive validity, and also resonates with received views of clinical utility. First9 remarks that two (of four) important components of clinical utility are “implementing effective intervention” and “predicting the future” of the patient's needs and outcomes. This merging of aspects of validity with utility seems both sound, and in the spirit of philosophical pragmatism, and points the way forward. Finally, there are also more recent etiological approaches to schizophrenia, such as the work of Lewis10, and the Psychiatric Genomics Consortium's suggestions of neuronal, immunological, and epigenetic etiological pathways11, as well as the circuit‐based approach recommended by the U.S. National Institute of Mental Health's extensive Research Domain Criteria initiative12. Jablensky notes some of these, but is more skeptical of the pursuit of them than I believe warranted. For the present, and as indicated above, however, there are some very suggestive empirical reasons for retaining the DSM‐5 approach to schizophrenia and other major disorders for clinical use. From a symptom aggregational view, the DSM‐5 criteria work fairly well in providing a diagnosis and a therapeutic plan. For research, however, more etiological approaches are likely to be more fruitful. This entails that psychiatrists should be pluralistic and select whichever approach seems likely to yield progress in their area of interest. Kenneth F. Schaffner University of Pittsburgh, Pittsburgh, PA, USA
Authors
K. Schaffner
Journal
World Psychiatry