Problems with DSM-IV are well known and now are becoming subjects of considerable interest as various committees attempt to update it to DSM-V. Some of the more salient issues are: excessive co-occurrence among disorders; extreme heterogeneity among patients receiving the same diagnosis; arbitrary diagnostic thresholds for boundaries between pathological and normal functioning; and inadequate coverage frequently leading to NOS diagnoses. Epidemiologic and clinical studies have shown high rates of comorbidity among disorders (Rottman et al., 2009). Many diagnoses carry pejorative connotations; they never resolve, only go into remission. On top of this there is a plethora of new candidates for mental disorders such as “internet addiction,” “night eating syndrome” and “sexual dysfunction.” As one commentator observes, “Psychiatric classification has swollen into a kaleidoscope of putative disorders” (Fink & Taylor, 2008).
The transition from DSM-III to DSM-IV was a series of ad-hoc modifications. Symptoms and criteria essentially were “bolted on” to accommodate various objections and better account for clinical data. To minimize the possibility of this recurring (and to maximize the likelihood of its utility) I have two suggestions for reimagining the approach to revising DSM-IV to DSM-V. If DSM-IV is an uneasy set of compromises situated at the middle, I propose to veer slightly both to the left and to the right, from the standpoint of spatial metaphor.
A. Becoming More Empirical
The first step I would take is for DSM-V to become more empirical and adopt more of what has been characterized as a “medical model” similar to that used for diagnosing other physiological conditions such as cancer or cardiac disease. This would put DSM-V on a sounder scientific footing.
The philosophy of scientific inquiry distinguishes between two different types of approaches. The first fundamentally is empirical. It addresses issues by observing symptoms, collecting data and then developing hypotheses. One might characterize this as a “bottom up” approach. The second approach fundamentally is theoretical. It conceives of hypotheses then searches for data (most frequently) to validate them. One might characterize this as a “top down” approach. It fundamentally is conceptual or theoretical in nature.
As now set forth DSM-IV contains too much of the latter and not enough of the former. It implicates questions about scientific method that fundamentally are unanswerable. One can speculate indefinitely, comprise various committees, examine the nuances of the problem, etc. yet not arrive at any useful conclusions. Theories are useful only if they organize and account for research data. They must be internally consistent and parsimonious in the sense they neither over-explain nor under-explain the phenomena at issue. They must generate testable hypotheses that are falsifiable through empirical investigation and analysis.
DSM-IV falls short on these criteria. The main reason why is because it focuses on symptoms at the expense of etiology. Occasionally it suggests various psychodynamic or socio-psychological causes however this is the exception and not the rule. It is silent on neurochemical or neuroanatomical causes. It lacks capacity to predict future outcomes (one of the basic attributes of a coherent scientific methodology).
To address this problem I would deemphasize DSM-IV’s symptom-based approach. Psychiatric disorders are better described with a medical diagnostic model, which is more empirical than one based primarily on symptom interpretation (Zachar et al., 2007). There now is biological evidence (neurochemical and neuroanatomical correlates) for many DSM pathologies. We are in the process of developing it for most others. A medical diagnostic model would not exclude consideration of pertinent psychodynamic and socio-psychological factors to the extent they are relevant. Most importantly it would provide a basis for predicting and treating future outcomes either neuropharmacologically or through some other therapeutic means. This would restore the DSM’s ultimate purpose, which is to increase clinical utility and improve patient care (First et al., 2004).
B. Dimensional Modeling
My second recommendation to some extent is the reciprocal of the first. DSM-V should become more sensitive to lived patient experience. This would reduce its dependence on “black or white” categories in favor of a more subtle, nuanced approach. Rather than symptoms either being present or absent it should be possible to describe them using various dimensional models that better account for their nature, degree, extent and severity – something DSM-IV now lacks.
To illustrate this transition I would like to consider personality disorders. These are particularly topical in addressing DSM revisions because the diagnostic criteria are far more indeterminate than they are for more serious mental illnesses such as schizophrenia. Unlike psychotic conditions no laboratory markers have been found that specifically correlate with any DSM-defined syndrome for PDs. Epidemiologic and clinical studies show high rates of comorbidity (within and across axes) and short-term diagnostic instability, which is exacerbated by the lack of disorder-specific treatment (Kupfer et al. 2002). For these reasons PDs have become a litmus test for other diagnostic approaches; it is not an exaggeration that what happens with PDs well could end up pervading the rest of the DSM (Rounsaville et al., 2002).
DSM-IV adopts what might be called a taxonomical or “symptom”-based approach to diagnosing PDs. This “categorical” approach comprises a list of “objective” signs or indicators that should be used by clinicians and investigators to achieve a reliable rating. It is based on the concept that PDs present qualitatively distinct clinical symptoms. Examples are the time constraints for certain pathologies or aggregating the number of symptoms that must be present in order to reach a particular diagnosis.
This type of categorical model has several disadvantages. Symptoms are discrete clinical distinctions frequently with arbitrary boundaries. Categorical analysis hypothesizes that syndromes represent distinct etiologies (Widiger & Samuel, 2005). Symptoms are coded as present or absent without making any more subtle distinctions. There are no clear boundaries between categorical diagnoses. This results in symptom heterogeneity and loss of information as to possible subthreshhold traits.
The DSM-IV schematic for PDs establishes 10 categorical disorders. Embedded in these 10 disorders are 79 descriptive criteria (not counting ancillary criteria such as exclusionary criteria). To thoroughly differentially diagnosis a specific patient one would have to consider the applicability of each of these 79 criteria within each of the 10 DSM-IV PDs. After signs or symptoms associated with a diagnosis are evaluated, one must apply a monothetic or polythetic algorithm to determine if it is accurate (Trull & Widiger, 2008).
A “dimensional approach” on the other hand utilizes common “higher-order” factors to characterize personality pathology (Trull et al., 2007). Although they have several different meanings these “dimensions” are fundamentally are domains or traits of personality. They are systematically collected descriptions of symptoms thought to be relevant to various personality disorders (or more generally, personality pathology).(2) One theory posits four elements or facets that can be used to provide a comprehensive description of abnormal personality (Krueger et al., 2007). This four-factor model includes: emotional dysregulation, dissocial behavior, inhibitedness, and compulsivity (Trull & Widiger, 2008).
The primary advantage of the dimensional model is its sensitivity. PDs are among the most difficult of psychiatric illnesses to conceptualize. A dimensional model makes it possible to evaluate an individual on a scale of personality and evaluated for potential psychopathology given her/his relative suppressions or elevations on different scale scores. Such an approach presents a more comprehensive description of patient functioning and more accurately represents the organization of personality and personal pathology (Huprich & Bornstein, 2007).
Another advantage of the dimensional approach is that it activates and better accommodates the use of first-person reports. DSM-IV does not consider the patient’s subjective experience of a disorder. First-person reports differ from third-person observations in that they can indicate the somatic and psychological processes that underlie the symptoms. They also can help clinicians understand their patient’s experiences, creating an “empathetic bridge” (Flanagan, et al., 2007).
Finally dimensional analysis also facilitates “clinical staging,” which tracks the time course of a pathology and where the patient is situated along this continuum. This enables clinicians to select treatment that is more time-sensitive to the earlier stages. Such interventions have a higher probability of being more effective and less deleterious than ones delivered later during the course of the illness. While this might not increase the likelihood of cure it should reduce mortality and disability. It is a prevention-oriented strategy for understanding the pathogenesis and evaluation of interventions (McGorry, 2007).
(1) Here are some examples of problems associated with this latter approach as applied to DSM-IV (Kendler et al., 2008): (1) “What is the definition of a mental disorder?”; (2) “Given a symptom cluster, how does one decide if it is a genuine psychiatric condition?”; (3) “What should the criteria be for revising diagnostic criteria?”; (4) “Should disorders be classified by how symptoms resemble each other, etiological relationships, or other considerations?”; (5) “Which validators should be given priority (treatment outcome, genetic markers, course of illness)?”; (6) “What happens if different validators give different answers?”; (7) “What guidelines should be established for the inclusion and performance of laboratory values and etiological criteria?”; (8) “How should the priority of even these questions be resolved?”
(2) One approach to developing them is to count how many words a language has to describe their various degrees or nuances. A “lexical” model means the dimensions originally were developed by studying the incidence of words describing traits within the semantics of an existing language. (Widiger & Lowe, 2007).
Fink, M. & Taylor, M. (2008). “Issues for DSM-V: The Medical Diagnostic Model.” Am. J. Psychiatry, 165(7), 799.
First, M., Pincus, H., Levine, J., Williams, J., Ustun, B. & Peele, R. (2004). “Clinical utility as a criterion for revising psychiatric diagnoses.” Am. J. Psychiatry, 161, 946 – 954.
Flanagan, E., Davidson, L. & Strauss, J. (2007). “Issues for DSM-V: Incorporating Patients’ Subjective Experiences.” Am. J. Psychiatry, 164(3), 391 – 392.
Huprich, S. & Bornstein, R. (2007). “An Overview of Issues Related to Categorical and Dimensional Models of Personality Disorder Assessment.” J. Personality Assessment, 89(1), 3 – 15.
Kendler, K., Appelbaum, P., Bell, C., Fulford, K., Ghaemi, S., Schaffner, K., Waterman, G., First, M. & Zadler, J. (2008). “Issues for DSM-V: DSM-V Should Include a Conceptual Issues Work Group.” Am. J. Psychiatry, 165(2), 174 – 175.
Krueger, R., Skodol, A., Livesley, W., Shrout, P. & Huang, Y. (2007). “Synthesizing dimensional and categorical approaches to personality disorders: refining the research agenda for DSM-V Axis II.” Int’l J. of Methods in Psychiatric Research, 16(S1), S65 – S73.
Kupfer, D., First, M., Reigier, D. (2002). “Introduction” in Kupfer, D., First, M. & Reigier, D. (eds.). A Research Agenda for DSM-V, pp. xv – xxiii. Washington, DC: American Psychiatric Ass’n.
McGorry, P. (2007). “Issues for DSM-V: Clinical Staging: A Heuristic Pathway to Valid Nosology and Safer, More Effective Treatment in Psychiatry.” Am. J. Psychiatry 164(6), 859 – 860.
Rottman, B., Ahn, W., Sanislow, C., Kim, N. (2009). “Can clinicians recognize DSM-IV personality disorders from Five-Factor Model descriptions of patient cases?” Am. J. Psychiatry 166, 427 – 433.
Rounsaville, B., Alarcón, R., Andrews, G., Jackson, J., Kendell, R. & Kendler, K. (2002). “Basic nomenclature issues for DSM-V” in Kupfer, D., First, M., Regier, D. (eds.) A Research Agenda for DSM-V, pp. 1 – 29. Washington, DC: American Psychiatric Ass’n.
Trull, T.; Tragesser, S.; Solhan, M.; & Schwartz-Mette, R. (2007). “Dimensional models of personality disorder: Diagnostic and Statistical Manual of Mental Disorders Fifth Edition and beyond.” Current Opinion in Psychiatry, 20(1), 52 – 56.
Trull, T. & Widiger, T. (2008). “Geology 102: More Thoughts on a Shift to a Dimensional Model of Personality Disorders.” Social and Personality Psychology Compass, 2(2), 949 – 967.
Widiger, T. & Lowe, J. (2007). “Five-Factor Model Assessment of Personality Disorder.” J. Personality Assessment, 89(1), 16 – 29.
Widiger, T. & Samuel, D. (2005). “Diagnostic Categories or Dimensions? A Question for the Diagnostic and Statistical Manual of Mental Disorders – Fifth Edition.” J. Abnormal Psychology, 114(4), 494 – 504.
Zachar, P., Kendler, K. (2007). “Psychiatric Disorders: A Conceptual Taxonomy.” Am. J. Psychiatry, 164, 557 – 565.