The Quantitative Classification of Mental Disorders
Chapters 6-8 reflect the consensus method for determining diagnosis. Another option is to use a quantitative approach to nosology, meaning an empirically based organization of psychopathology.27,41,42 Rather than being constructed in a top-down manner, the quantitative nosology takes an independent approach in evaluating the natural organization of psychopathology. This approach has a long history, especially in child psychiatry.43,44 Krueger45,46 was among the first to use epidemiological data in this regard, thus inspiring a growing and diverse literature.
The shared assumption of this work is that psychopathology can be effectively represented by dimensional constructs. There is no conclusive evidence of categorical entities in mental health to challenge this assumption,7-10 but if such entities were to emerge, they can be easily incorporated into the quantitative nosology. Modern statistical tools, such as factor mixture models,47 allow modeling of dimensions and categories simultaneously. A common concern with dimensional classifications is whether they are applicable to clinical settings because clinical care requires categorical decisions. Indeed, cutoffs will have to be applied to these dimensions, but the choice of a cutoff can be flexible and depend on the question (e.g., low cutoff for preventive interventions, somewhat higher one for psychotherapy, higher still for pharmacotherapy, and highest for inpatient treatment). In contrast, DSM-5 offers a single cutoff—the diagnostic threshold—regardless of the clinical question. Indeed, there is a long history of clinicians’ effectively employing dimensional assessment tools, such as the Minnesota Multiphasic Personality Inventory (MMPI48) and the Child Behavior Checklist (CBCL49). Importantly, the dimensional approach resolves the issue of arbitrary thresholds and associated loss of information.50 It also resolves the issue of instability, as indicated by high test-retest reliability of dimensional psychopathology constructs and commonly used mental health scales.
The basic blueprint for research on quantitative nosology is to construct psychiatric syndromes based on empirical covariation of symptoms, grouping related symptoms together and separating unrelated ones, thus identifying unitary constructs and reducing diagnostic heterogeneity. The next step is to combine syndromes into spectra, thus incorporating comorbidity into the structure of nosology. Comorbidity conveys important information about shared risk factors, pathological processes, and illness course, and quantitative nosology formalizes this information, making it explicitly available to researchers and clinicians. Hence, if a question concerns a feature common to multiple conditions, a higher order dimension can be targeted. Alternatively, if a specific syndrome is of interest, the higher order dimension can be factored out (e.g., controlled statistically) to elucidate information unique to this syndrome. It is feasible to correct for a selected factor even in applied settings. K-correction on MMPI is an example of this51 and has been widely used by clinicians. Hierarchical organization is an important feature of the quantitative nosology. This multilevel approach (including individual symptoms, homogeneous components, disorders, subspectra, and spectra) allows for a flexible description of a patient depending on the level of specificity desired. This approach parallels the classification framework in the intelligence and personality fields.
Factor analysis is the basic statistical tool employed in the development of quantitative classification. It is a procedure designed specifically to group similar variables (symptoms, syndromes) together based on their empirical relationships. This family of techniques includes exploratory factor analysis, which searches for the optimal organization of variables, and confirmatory factor analysis, which tests the fit of hypothesized structures to the data. Both of these methods have been widely used in research on the quantitative nosology. Other techniques have been employed to investigate natural classes or hybrid models that allow for both classes and dimensions. However, simple dimensional models have shown better fit to the data than these approaches.7-10
Another potential solution to shortcomings of the current nosology are the research domain criteria (RDoC34). In promoting RDoC, the National Institute of Mental Health is lending support to a dimensional research classification system for mental disorders. The emerging system spans eight levels of analysis (from genes to behavioral tasks) and cuts across diagnostic categories. The dimensional approach can solve many problems of the current system. However, RDoC is concerned with basic biological functions (e.g., neural circuits) along with pathological behavior and seeks to link animal and human research, thus largely focusing on constructs conserved across species. RDoC holds a unique promise to advance the understanding of biological processes relevant to psychopathology, although it is not yet comprehensive in its coverage of psychopathology. RDoC begins to restructure nosology at a basic level, and translation of these advances to diagnostic practices likely lies well in the future. In contrast, quantitative nosology is driven by clinical constructs to improve on phenotypes for basic research. Overall, these two efforts approach nosology from different perspectives but are perfectly positioned to advance toward each other, ultimately producing a unified system. For example, quantitative nosology can inform RDoC with regard to pivotal self-report dimensions that should be considered in a classification system, whereas RDoC can offer biological markers useful for testing the validity of the spectra and symptom dimensions.