The Paradox of Polythetic Descriptive Diagnosis
Characteristically, the DSM provides checklists for diagnosing mental disorders in a descriptive and polythetic way. Each disorder is defined in terms of inclusion and exclusion criteria, based on which a professional can decide from which disorder a person is suffering. This decision-making process is based on a list of descriptive features of each disorder; with an inventory of symptoms the diagnostician explores whether a given individual meets a sufficient number of the inclusion criteria as well as certain specifications that would exclude the diagnosis of a particular disorder. In other words, the specific complaints with which the individual presents are translated in psychiatric jargon, and situated relative to a taxonomic classification of possible disorders, and the clinician must evaluate whether the presenting symptoms are signs of one or more disorders. For example, for major depressive disorder the DSM-5 lists nine relevant 
clinical characteristics. Major depressive disorder can be diagnosed if at least five criteria are present, with the proviso that at least one of the symptoms includes either depressed mood or loss of interest or pleasure. For each DSM-5 disorder, the same template of inclusion and exclusion criteria is used. Some disorders exclude one another; for example, a diagnosis of major depressive disorder excludes schizoaffective disorder, while other disorders might be combined, pointing to comorbidity, like major depressive disorder with comorbid generalized anxiety disorder.
Since the checklist-based approach was adopted, the DSM follows a polythetic system: it sketches broad profiles for most disorders, which implies that people with the same disorder might actually have completely different symptom profiles. Within the logic of polythetic diagnosis it is assumed that such variation is not cause for concern due to the obvious “family resemblance” between different symptom profiles: all symptoms more or less point to the same underlying category (Berrios 1999, 2012, p. 102). The strength of polythetic diagnosis is that it covers variability: patients with different symptomatic profiles can be classified into one category. Such grouping is unproblematic as long as all patients within the same category still share key features. The weakness of poly- thetic diagnosis, by contrast, is that such diversity is subsumed under a general nominator and as a result the more specific features that characterize individual cases fade into an abyss of irrelevance. Moreover, binary polythetic diagnosis only establishes whether symptoms are present or not. This obfuscates the extent to which the given list of descriptors actually grasps the nature of the individual’s complaint. The DSM-5 partly solved this problem by adding dimensional severity scales for clusters of symptoms for some disorders. For example, this is the case for autism spectrum disorder and alcohol use disorder, but not for major depressive disorder or schizophrenia. Most importantly, a detailed network analysis in which it was examined to which extent symptoms contribute to broader disorder categories shows that individual DSM symptoms differ substantially in their association with the diagnostic category they should belong to (Boschloo et al. 2015). Each symptom has a unique valence and value, and therefore is not interchangeable with other symptoms. This finding contradicts the idea that diagnoses can validly be made by just counting the number of symptoms.
Just as its two predecessors, the DSM-5 continues to use the poly- thetic classification principle in an as rigorous way as ever. As such, it is more than likely that two individuals with the same diagnosis have only a few, and occasionally no, traits in common. For instance, in the case of schizophrenia it might be that one individual obtains the diagnosis due to having hallucinations and disorganized speech, and another because of delusions and negative symptoms (i.e., diminished emotional expression or avolition) (see DSM-5, p. 99). In major depressive disorder, it is possible that one individual is diagnosed due to the presence of depressed mood, weight loss, insomnia, psychomotor agitation, and fatigue, while another is given the same diagnosis due to the presence of diminished interest in most daily activities, recurrent thoughts of death, diminished ability to concentrate, feelings of worthlessness, and hypersomnia (see: DSM-5, pp. 160-161): one disorder, two entirely different symptom profiles. Indeed, for the majority of DSM diagnoses patients with entirely different symptom profiles might be classified in the same disorder category. Such rigorous application of the polythetic classification principle might function as a source of unreliability. Moreover, Lilienfeld and Marino (1999, p. 400) suggest that notwithstanding the use of diagnostic checklists in the DSM, disorder categories function as “Roschian concepts.” This means that based on the disorder criteria, diagnosticians try to agree on “an ideal mental prototype that embodies the central features of the category.” Yet as the polythetic principle is applied it is quite likely that different diagnosticians consider different features as central to a disorder. This might bring them to building a different diagnosis for the same individual.
Given the fact that since the DSM-III the American Psychiatric Association defines all mental disorders in purely descriptive terms, such an extreme polythetic approach is rather paradoxical. In the DSM-III it was argued that for most mental disorders the etiology was actually unknown, which gave rise to the well-known descriptive approach: “the definitions of the disorders generally consist of description of the clinical features of the disorders. [ ... ] This descriptive approach is also used in the division of the mental disorders into diagnostic classes. All of the disorders without known etiology or pathophysiological process are grouped together on the basis of shared clinical features” (DSM-III, p. 7). Following this logic, one would expect that disorder descriptions build on a common set of mandatory symptoms and that variation is only possible in terms of additional features. Indeed, if a disorder is nothing more than the sum of specific symptoms, irrespective of its underlying processes and etiology, one would expect that in two individuals with the same diagnosis, common symptoms prevail. However, this is not the case in the DSM. The examples of schizophrenia and major depressive disorder show that two individuals with the same diagnosis might not actually share any symptoms at all. This brings us to the paradox: if disorders are defined based on the presence of a specific number of descriptive features only, how is it possible for two individuals to have the same disorder if they have none of the clinical features in common? This paradox has important consequences in terms of the validity of a DSM diagnosis. Diagnostic categories can only be considered as valid if the constituent cases are sufficiently alike and mutually coherent, referring to a similar underlying clinical reality. Extreme polythetic classification weakens diagnostic coherency, thus undermining diagnostic validity.