CHAPTER 2
OSM-5 as a Framework for Psychiatric Diagnosis
In medicine and public health, diagnostic classifications provide a framework for understanding and communicating about clinical conditions. They offer a systematic presentation of diagnostic categories, based on delineation of specific clinical conditions organized into meaningful groups. After examining patients, clinicians specify the diagnoses—that is, state which disorders they believe are present. This complex task demands skill in assessment as well as knowledge of the basis for defining and categorizing mental disorders.
Classification systems reflect the current state of scientific knowledge, so they need to be revised periodically. Besides research, classifications are also influenced by their intended uses, prior history, and underlying concepts of illness (Moriyama et al. 2011; Sadler 2005; Sartorius et al. 1990; Zachar and Kendler 2007).
In the United States, the current classification for psychiatric practice is the Diagnostic and Statistical Manual of Mental Disorders, 5th edition (DSM-5; American Psychiatric Association 2013).
The "official" classification for public health reporting is the International Classification of Diseases and Related Health Problems (ICD), developed and periodically revised by the World Health Organization (WHO). All member countries of WHO have agreed to use this system to report disease statistics to its surveillance system, although they may not adopt each updated version as soon as it is available. Each country does have the ability to make its own changes, for example, to include terms that are used locally or to add precision to the system. In the United States, the National Center for Health Statistics oversees this process, which produces a "clinical modification" of ICD. It collaborates with specialty organizations, such as the American Psychiatric Association (APA), to incorporate diagnostic categories used in the United States.
An ideal classification system might be the periodic table. Chemical elements fit into an array based on their atomic numbers; this classification provides a clear, unduplicated placement for each element. The periodic table is exclusive, because each element fits into only one category; it is exhaustive, because every element can be classified within it; and it is discrete, because the defining characteristic, the atomic number, matches one integer or another, rather than one of the infinite real numbers between two integers. Its columns group elements by similar properties, and its rows demonstrate their electron structures (Kendell 1975; Zachar 2012).
If each illness could be shown to have just one cause, illnesses could be categorized on the basis of etiology and the associated mechanism of disease. Infectious diseases and nutritional deficiencies provide a model for this approach. Once an etiological agent or cause becomes established, researchers can study methods to cure the illness or prevent it. This model has been the accepted standard for thinking about diagnoses in medicine (Loscalzo 2011).
Defining illnesses, however, is challenging. Even with a single cause, a clinical condition can present with a variable mix of symptoms, as in myocardial infarction. In many conditions, especially psychiatric disorders, multiple biological, psychological, and environmental factors interact in complex ways over time to produce an illness. Specification of the way these factors interact is difficult because so many processes in neuroscience and psychology are complex and poorly understood (Jones et al. 2012; Loscalzo and Barabasi 2011). As a result, it is not possible to classify psychiatric illnesses as diseases based on etiology and mechanism, using a reductionistic, one-level explanation (Kendler 2012; Schaffner 2008).
Clinical conditions are not freestanding entities like chemical elements, so the periodic table is an unrealistic model for classification in medicine. Even for medical conditions with a single known etiology, types of causes vary considerably, such as infectious agents, chromosomal abnormalities, or neoplasms. There is no single defining characteristic, like an atomic number, to govern their classification (Jones et al. 2012; Kendler 2008). Even a single defining characteristic of a disease may produce ambiguity in distinguishing between normality and disease; for example, an individual can be a carrier of an infectious agent, such as tubercle bacillus, without having an illness. The defining characteristics of some conditions are continuous measures, such as blood pressure or hormone levels, and require a threshold to be set for a category to be identified. Research on epigenomics, proteomics, and other aspects of systems biology will lead to even greater complexity of disease classifications (ENCODE Project Consortium 2012; Hyman 2010; Institute of Medicine 2012; Loscalzo 2011).
When a defining characteristic has not been established, clinical conditions are identified as syndromes. A syndrome is based on pattern recognition and comprises a characteristic set of signs and symptoms that usually occur together, have a typical prognosis, and may have a similar response to treatment. A syndrome, such as anemia, is usually anticipated to be heterogeneous in terms of causes; eventually specific diseases may be identified within the broad syndrome, such as thalassemia or hemorrhagic anemia. Clinical conditions can present with an incomplete symptom picture, and they can be confusing because the signs and symptoms are not exclusive to a single syndrome. Assessment of a patient can be challenging, because an individual may present with more than one condition, or with one that does not fit any recognized syndromal pattern. With this reliance on syndromal diagnosis, the goal of producing a psychiatric classification that is exclusive, exhaustive, and based on discrete entities is not attainable without new knowledge and some unforeseen paradigm shift (Kendler and First 2010).
In contemporary psychiatric classifications, many categories derive from long-recognized syndromes, which clinicians and researchers have refined over time based on experience, research, and shifts in their explanatory models of illness. In the past two centuries, syndromes have been introduced by leading figures in psychiatry ("the great professors," such as Emil Kraepelin or Eugen Bleuler) (Kendler 1990). Since 1952, classification systems have been developed by panels of experts, who have been charged with reviewing evidence and developing refinements based on a consensus of opinion. The judgments that produce any given classification must balance widely varying concepts of psychiatric illness, such as the assumptions made about causation, the threshold for determining abnormality, and standards for interpreting research evidence (Fulford et al. 2006; Jablensky 2009; Pepe 2005; Zachar and Kendler 2007).
Because the etiology and pathogenesis of so many psychiatric conditions are not yet established, contemporary classification systems describe the conditions being categorized as disorders rather than diseases, a term associated with conditions whose causes are known. Both DSM and the ICD's section on mental disorders have acknowledged that disorder is a convenient but inexact term.
This imprecision in describing diagnostic categories has led critics to question how clinicians can distinguish mental disorders from reactions to normal life, or problems in living. Over the last half-century, the task of identifying and classifying mental disorders has drawn attack as doing nothing more than labeling unwelcome social, political, or other behaviors as mental illnesses. These lingering criticisms have led to a more formal effort in recent editions of DSM to offer a general definition of mental disorders, even though clinicians typically only worry about specific disorders and the rest of medicine is untroubled by the lack of rigorous definitions of disease and illness (Kendell 1975; Zachar 2012).
The introduction to DSM-IV acknowledged the imprecision of the concept: "it must be admitted that no definition adequately specifies precise boundaries for the concept of 'mental disorder'" (American Psychiatric Association 1994, p. xxi). It listed the characteristics that the system used to conceptualize mental disorders:
When there is no way to identify the causal factors or mechanisms of an illness, and the clinical presentation does not produce a clear break between normality and clinically relevant illness, specifying when a syndrome crosses the boundary from normal to abnormal is one of the challenges for a classification system (Kendell and Jablensky 2003). In DSM-IV, this challenge was met by requiring the association of a syndrome with distress, impairment, or risk of other harmful outcomes. More than 70% of the specific disorders in DSM-IV included distress or impairment among the required criteria for the condition. Despite the relevance of the third associated characteristic, "increased risk" of a harmful consequence, this factor was not included in DSM-IV's standard criterion of "clinically significant distress or impairment" (Lehman et al. in Kupfer et al. 2002).
This pragmatic approach generated problems of its own. Terms such as clinically significant and impairment are not defined. Requiring impairment for diagnosis of a clinical disorder could preclude early intervention in preclinical phases of the disorder. It could also complicate efforts to study the evolving relationship between the syndrome and any subsequent disability. Trying to attribute distress or impairment to a particular disorder when more than one disorder is present is difficult at best (Lehman et al. 2002; Sartorius 2009).
In DSM-5, the definition of mental disorder is phrased in terms of dysfunctions and their underlying processes:
A mental disorder is a syndrome characterized by clinically significant disturbance in an individual's cognition, emotion regulation, or behavior that reflects a dysfunction in the psychological, biological, or developmental processes underlying mental functioning. Mental disorders are usually associated with significant distress or disability in social, occupational, or other important activities. (American Psychiatric Association 1994, p. 20)
This definition still does not provide operational guidance on specifying a threshold for diagnosis. The difficulty in determining when a dysfunction has produced a change from normal variation to pathology has meant that the coupling to "significant distress or impairment" could not be removed completely. DSM-5 criteria for some individual disorders have retained DSM-IV's specific reliance on distress or impairment to establish the diagnosis.
This definition in DSM-5 serves mainly as a guidepost to the future. It anticipates a time when the "dysfunction" can be measured directly and the underlying processes can be identified (Hyman 2010; Insel and Wang 2010).
From its first publication in 1952, DSM's title has referred to dual purposes. The manual was intended to serve as both a diagnostic manual for clinicians and a guide to statistical reporting useful for hospital and public health officials. Trying to serve both purposes was another challenge for the experts developing the classification system.
For clinicians, DSM-I (American Psychiatric Association 1952) presented a list of diagnostic terms, to serve as a nomenclature. Although the terms could have been listed alphabetically, the list was presented according to the coding system in the then-current ICD-6 (World Health Organization 1949). DSM-I also provided a "definition of terms," with a brief description of each condition within a diagnostic category. The standardized terms for the diagnostic categories and the corresponding definitions were the essential elements of the manual for clinicians.
For institutional and public health officials, the manual explained how the classification could be used to report the number of cases seen in a clinical setting, usually a public mental hospital, over a given period of time. For this reporting process, coding clerks noted the diagnostic term used by a clinician and counted the patient in the appropriate category within a tabular aggregation of total cases. For the clerk's reporting effort, the definition of terms was irrelevant. Rather, the code corresponding to the clinician's diagnosis was the key information used from the manual.
These two different uses highlighted conflicts that initially affected the way psychiatric diagnoses were made. The manual offered a classification of disorders; the statistical report offered a classification of cases, with numbers of cases in each given class of disorder presented in a table. For statistical reports, each person should only be counted once, so the goal of limiting a patient's diagnosis to a single disorder influenced clinical practice. It became one of the major controversies addressed in the preparation of DSM-II (American Psychiatric Association 1968). A compromise allowed clinicians to document more than one disorder, with the understanding that one of these would be considered the "primary" diagnosis for statistical reporting (Gruenberg 1968).
A second problem arose from the tension between "lumping" and "splitting," a decision that continues to be confronted in each revision of a classification. Clinicians wanted the conditions to be described at an appropriate level of precision to allow meaningful distinctions to be made. Statisticians needed to have a manageable number of categories, so they envisioned a system whose individual diagnostic classes could be collapsed into broader but still meaningful categories. They favored a hierarchical system that would allow this aggregation and still preserve the logic of a classification (American Psychiatric Association 1952, pp. 88-89).
Over time, these two conflicts have eased. DSM no longer expects the clinician to restrict the number of diagnoses artificially; in fact, one recent concern has centered on how many "comorbid" conditions are diagnosed in a given patient. Also, the classification no longer tries to construct a formal hierarchy of conditions, but instead groups them by some perceived similarity. As these original conflicts among different types of uses have faded, additional problems have arisen more recently from several nonclinical applications of DSM. In the United States, commercial and government payers have selectively applied DSM classifications to determine which disorders would be covered by a beneficiary's insurance plan. Some disorders have also been further restricted by payers on the basis of Global Assessment of Functioning (GAF) scores that were introduced by DSM-III in 1980 (American Psychiatric Association 1980). Payers have predetermined whether a GAF score for some conditions justifies a level of service, such as hospitalization for a depressive episode. This approach uses the diagnostic manual as an indicator of "need for treatment" rather than presence of a disorder, thereby reducing the complicated problem of treatment planning to an arbitrary, nonclinical approach. DSM-5 has emphasized that a diagnosis has clinical utility for clinicians, such as aiding them in developing treatment plans, but does not by itself determine a need for treatment. Need for treatment depends on multiple factors beyond diagnosis (American Psychiatric Association 2013, p. 20), such as any of the following:
In other fields of medicine, a clinical condition often receives clinical attention even when the condition is not considered a disease, as with childbirth (Kendell 1975). Risk conditions, such as the metabolic syndrome or intestinal polyps, have also been legitimate objects of clinical intervention even without a diagnosis of a disease state. DSM-5's distinction between a diagnosis and a need for treatment is intended to support the same flexible application of clinical judgment in psychiatric practice.
The diagnostic system has also been broadly used in judicial proceedings, such as to bolster a defense in a criminal trial, or in social welfare systems to justify claims for disability payments. However, DSM revisions have noted that a diagnosis can only be made by well-trained clinicians. DSM-5 has reiterated that the presence of a mental disorder does not substitute for these other judgments:
This definition of mental disorder was developed for clinical, public health, and research purposes. Additional information is usually required beyond that contained in the DSM-5 diagnostic criteria in order to make legal judgments on such issues as criminal responsibility, eligibility for disability compensation, and competency. (American Psychiatric Association 2013, p. 20)
Although these cautions are intended to avoid misuse of DSM-5 diagnoses, other users besides practitioners rely on the manual for legitimate purposes. Trainees in clinical disciplines benefit from the extensive description of each diagnostic category. This breadth of content also makes the manual a useful resource for patients and their families, as well as for the public and policy makers.
In the nineteenth century, psychiatrists in Europe and the United States concentrated on identifying and describing individual disorders encountered in their hospitals or offices, but had less concern about formulating an overall classification system to organize the full range of disorders found in a population. Not until the mid-twentieth century did the profession recognize the need for a system that incorporated diagnostic terms for a broad array of patients who were seen outside a public hospital (American Psychiatric Association 1952; Grob 1994).
In the United States, public health concerns drove the initial efforts to classify mental disorders. Beginning in 1840 and continuing for the next century, components of the federal government enlisted leaders of the new specialty of psychiatry to help develop classifications of psychiatric illnesses for use in public health reporting (American Psychiatric Association 1952, 1968; Grob 1991).
From 1840 through 1890, the U.S. Census attempted to count people who had a mental disorder, characterized as "insanity" or "idiocy." In 1880, a special Census survey tried to apply a system with seven categories, but officials found it impossible to construct a list that would be endorsed by leaders of psychiatry and psychology (American Psychiatric Association 1952).
After 1890, the Census abandoned its effort to determine psychiatric illness in the population. In 1917, the Census began planning a survey of individuals who resided in mental institutions. It asked the precursor of the APA for assistance in developing a classification system for psychiatric illnesses. This system, revised periodically through consultation with the APA and the American Medical Association, was used to survey psychiatric patients residing in hospitals beginning in 1923 and continuing to World War II.
When the United States entered World War II, the need to assess volunteers and to treat soldiers and veterans exposed the shortcomings of the existing system. "Military psychiatrists, induction station psychiatrists, and Veterans Administration psychiatrists found themselves operating within the limits of a nomenclature specifically not designed for 90% of the cases handled" (American Psychiatric Association 1952, p. vi).
To provide a more useful nomenclature, the U.S. Army's Office of the Surgeon General prepared a technical bulletin that provided a comprehensive set of diagnostic terms, explained by definitions and grouped into categories, such as psychoneurotic disorders, character and behavior disorders, and psychotic disorders. Definitions of several categories, such as psychoneurotic disorders, reflected a psychodynamic orientation (Houts 2000).
After the war, the APA's Committee on Nomenclature and Statistics prepared a new classification based on the version created by the army. The logic of the classification used in DSM-I applied the major viewpoints in American psychiatry (Grob 1991):
Despite the hope that the new manual would improve the basis for uniform application of diagnostic terms, clinicians' use of psychiatric diagnoses continued to vary widely within the United States and around the world. An international survey conducted for WHO found little agreement about concepts or even names to be used for the conditions. Its recommendation for increasing agreement was for future classifications to rely on "operational" statements based on observation rather than prototypes developed locally (Stengel 1959). These results prompted WHO to expand the coverage of mental disorders in ICD as it began planning a new revision (ICD-8) in the mid-1960s.
In concert with this effort, the APA's Committee on Nomenclature and Statistics produced DSM-II in 1968. It mirrored the format of DSM-I and continued a psychodynamic approach to psychoneurotic disorders. Reflecting the same lack of consensus found in the WHO survey a decade earlier, the committee members disagreed on the descriptions, or even the names to use, for conditions such as schizophrenia. Growing research evidence also demonstrated that agreement on clinical diagnosis of mental disorders remained low (Spitzer and Fleiss 1974).
In an effort to understand the wide variations in diagnostic practice as shown in hospital statistical reports through the 1960s, a cross-national study examined patients hospitalized in New York and London. Investigators developed sets of explicit criteria for the disorders under study and then trained American and British psychiatrists to use a common set of data about each patient to assign a diagnosis based on these criteria. The patient data included results from systematic clinical interviews and historical information from case records. Using a consensus diagnosis, the study demonstrated that reported variations in the clinical diagnoses came from variations in practice, not from true variations in the patient mix (Cooper et al. 1972). With a similar perspective, psychiatrists in St. Louis, Missouri, who wanted to pursue research on mental disorders developed explicit criteria sets for 16 disorders. They published these criteria sets so other researchers who wanted to study similar groups of patients could use them (Feighner et al. 1972). Soon after, a multisite study of depression sponsored by the National Institute of Mental Health (NIMH) adopted this approach with modified criteria (Spitzer et al. 1978).
As this approach of specifying explicit criteria for individual disorders gained rapid acceptance among clinical researchers, WHO undertook another revision of ICD. In 1974, Robert Spitzer, one of the leaders in these efforts to improve reliability, became chair of the APA Task Force on Nomenclature and Statistics to prepare DSM-III in parallel with ICD-9. He recruited task force members largely from the growing clinical research community, and their commitment to enhance clinical and research diagnoses led to a new paradigm for classifying mental disorders (Klerman et al. 1984).
DSM-III provided an explicit set of criteria for each clinical disorder. These criteria specified characteristics that must be present in some combination ("inclusion criteria") and characteristics that would prevent the disorder from being diagnosed ("exclusion criteria"). The task force used inclusion criteria that were descriptive in terms of psychopathology, without any assumptions about etiology unless it had been clearly established. DSM-III dropped the concept of psychoneurotic disorders that had been introduced in the army's nomenclature and continued in DSM-I and DSM-II. DSM-III also removed the term neurosis from the classification system. Rather than giving a brief one- or two-sentence definition of a disorder, DSM-III provided extensive information beyond the criteria sets in an effort to help clinicians understand the disorder; typically, each disorder also contained a description of associated features, course of illness, and differential diagnosis (Spitzer et al. 1980).
Besides developing criteria for specific disorders, the task force also provided clinicians with a framework to record nondiagnostic data about a patient's condition that would be relevant in planning treatment and gauging prognosis. DSM-III recommended that each diagnostic formulation should include five types or "axes" of information. Axis IV, the severity of psychosocial stressors, and Axis V, the highest level of adaptive functioning in the past year, constituted novel additions to the traditional diagnostic assessment, to support a comprehensive evaluation (Spitzer et al. 1980; Spitzer and Williams 1994).
Because so many of the disorders in DSM-III had not been well studied before 1980, and because some aspects of the classification such as its length and multiaxial structure were new, APA undertook a review process beginning in 1983 to examine points that needed refinement to improve ease of use, or to reflect new research that had been stimulated by publication of DSM-III. The revised version of DSM-III, referred to as DSM-III-R, was published in 1987 (American Psychiatric Association 1987). One of the principal changes was to remove many of the diagnostic hierarchies that had been introduced in DSM-III through its exclusion criteria; for example, DSM-III did not permit a diagnosis of panic disorder if the panic attacks occurred only in the course of major depression. Studies of clinical and epidemiological populations demonstrated that the use of hierarchies had precluded the clinician from assigning more than one diagnosis when multiple syndromes occurred in an episode. The particular pattern of exclusion criteria in DSM-III did not explain the co-occurrence of syndromes, so it was simplified (American Psychiatric Association 1987, p. xxiv; Boyd et al. 1984).
In concert with preparation of ICD-10 by WHO in the late 1980s, APA established the DSM-IV Task Force to produce the next version of the classification, published in 1994 (American Psychiatric Association 1994). This group had the advantage of considerable research stimulated by DSM-III and DSM-III-R, enhanced by diagnostic interviews designed to provide systematic assessment of a respondent by DSM-III and DSM-III-R criteria.
After a thorough review of each disorder, consisting of literature reviews, reanalysis of existing data sets, and field trials of selected criteria, 13 work groups simplified, clarified, and sometimes modified the criteria and thresholds for diagnosis. Major changes in the system included the decision to drop "organic mental disorders," because the term implied that other mental disorders did not have an "organic" component. Similarly, "physical disorders," which had been used to refer to other medical conditions, was replaced by "general medical conditions" to remove the suggestion of a mind-body split for illness. The threshold for adding new disorders to the classification was raised from DSM-III; clinical relevance was generally not sufficient as a rationale, and some research on the disorder was required. For conditions that might need further study, such as binge-eating disorder, an appendix was created to present criteria sets for further study. Although both Axes IV and V were refined in concept, the focus on aiding a comprehensive evaluation of the patient was maintained.
To avoid any disruption in continuity for clinical researchers and to reduce the burden on clinicians of learning new criteria sets, APA decided not to publish a major revision of DSM-IV until at least 2010. However, to keep the text descriptions current with research findings, and to correct any ambiguities or errors, APA did publish a revised text of DSM-IV, referred to as DSM-IV-TR, in 2000 (American Psychiatric Association 2000). None of the criteria sets were modified, and no structural changes were undertaken.
In 2000, APA sponsored a workshop to identify high-priority areas for study in preparation for DSM-5. A series of workshops over the next several years refined the issues to address and led to formation of the DSM-5 Task Force. David Kupfer, M.D., was appointed chair, and Darrel A. Regier, M.D., M.P.H., became vice chair. Members of the task force who would lead work groups on categories of specific disorders were appointed in 2008, and several cross-disorder study groups were also established.
As for DSM-IV, guidelines established at the initial stages of planning for DSM-5 included an emphasis on clinical relevance. This emphasis was coupled with a requirement that significant changes from DSM-IV should be based on empirical research. Although no changes from DSM-IV were precluded, APA's leadership also wanted to preserve continuity with DSM-IV whenever possible to minimize unnecessary disruptions for clinicians or researchers. (Key issues that arose during the preparation of DSM-5 are summarized in Table 2-1.)
One concern arose from the growing worry that commercial and other interests might hope to influence the selection of disorders or the formulation of their criteria. To forestall possible conflicts of interest among the many experts who would be asked to participate in the deliberations about DSM-5, the APA Board of Trustees developed eligibility criteria to ensure the independence of participants in any level of the revision work. Some critics objected that the standards restricting financial support should have been even more stringent, but APA leadership reiterated their commitment to the principle of independence, their confidence in the guidelines, and their acceptance of the fact that many leading experts in psychiatry and related disciplines had been excluded from any role in DSM-5 by virtue of their participation in industry-sponsored research or consultations.
The APA Board of Trustees also established three types of peer review groups to support the task force. One group monitored the extensive process of managing the effort to ensure that work proceeded efficiently and to guard against inadvertent conflicts of interest. A scientific review committee reviewed proposed changes to DSM-IV in terms of specific research findings on improvements in validity, when such data existed (Kendler et al. 2009). For some conditions, a clinical and public health committee also reviewed proposals that were based on clinical or public health considerations or other scientific evidence beyond the correlation with validators. Both of these substantive review committees prepared critiques for consideration by members of the task force, whose final recommendations were then reviewed by leaders of the APA and ultimately its board of trustees.
|
Table 2-1. Challenges for DSM-5 |
||
|
|
Issue |
Responses |
|
Process |
Potential conflict of interest |
Annual disclosures Restrictions on commercial activities, income Peer review of task force process |
|
Communication with field, public, advocates |
Public Web site with current drafts Annual comment periods open to public |
|
|
Harmonization with ICD |
Periodic meetings, sharing drafts Adopt WHODAS II to replace Axis V |
|
|
Mobilization of expertise |
Sustained effort by experts from multiple disciplines, from U.S. and abroad, participating as task force, work groups, study groups, consultants, advisers, peer reviewers |
|
|
DSM-5 system |
Standards for changes |
Peer review (science; clinical and public health) Specification of validators for science reviews |
|
Dimensional measures |
Stepwise development and testing for specific areas and specific disorders |
|
|
"Clinical significance" |
Restate definition of mental disorders with less reliance on "distress or impairment" |
|
|
Organizing principle for grouping disorders |
Clinical similarity, with lifespan approach in each section Omit "childhood" disorder section |
|
|
Multiaxial structure |
Change to uniaxial structure |
|
|
Comorbidity |
Introduce dimensional measures Collapse disorder categories Modify criteria to clarify boundaries |
|
|
Culture, gender, neuro-science |
Expand text descriptions of each disorder |
|
|
Specific disorders |
Clarity and complexity of criteria |
Revise wording Separate compound disorders |
|
Excessive NOS usage |
Identify need for new disorder categories |
|
|
Poor reliability |
Modify criteria Collapse closely related disorders |
|
Note. ICD=International Classification of Diseases; NOS = not otherwise specified; WHODAS II=World Health Organization Disability Assessment Schedule 2.0.
Work group members undertook extensive literature reviews and analyses of preexisting data sets as they reviewed DSM-IV criteria. Several problems were identified in their reviews:
The structure of the classification, in the way it grouped disorders into chapters, and in its overall use of a multiaxial system, was also examined (Kendler et al. 2009; Kupfer and Regier 2011).
Underlying these varied issues, one fundamental problem for almost all of the conditions was the proper determination of diagnostic boundaries—that is, the separating of a specific disorder from other disorders and from normal, nonclinical variations. The difficulty of setting boundaries in prior DSM versions had led to high rates of comorbidity and to uncertainty about setting a threshold for diagnoses without using "distress or impairment" as indicators (Kendler et al. 2009; Regier et al. 2009). Although it seemed unlikely that a sufficient basis existed for DSM-5 to introduce a new paradigm for defining and classifying psychiatric illness, leaders of the APA and NIMH felt that a systematic revision effort might start a process that would lead to improvements in diagnostic categorization over time (Kendler et al. 2009). As part of this exploration of different approaches, the task force considered introducing dimensional ratings into the diagnostic process to refine the use of binary categories in DSM-5 (Kupfer and Regier 2011).
In DSM-5, a multiaxial framework is no longer used. Although this framework had the beneficial effect of encouraging clinicians to conduct a full-range evaluation, including personality disorders, intellectual development, other medical conditions, stressors, and level of functioning, the multiaxial system was not applied uniformly in practice. Other important factors, such as consideration of gender or cultural characteristics, also need to be included in a comprehensive evaluation but did not have a separate axis. The development of a more robust and informative rating scale, the WHO Disability Assessment Schedule 2.0 (WHODAS II; Ustün et al. 2010), resulted in an instrument that could replace the Global Assessment of Functioning that formed Axis V.
Grouping disorders into chapters, organized in a particular sequence, is intended to convey information that a simple alphabetical listing would not. For clinical use, the organizing principles should make the classification easy to use and logical in expressing some similarities among disorders. The organization may be helpful in guiding differential diagnosis and in stimulating research on closely related conditions. Early in the revision process, DSM-5's Diagnostic Spectra Study Group offered a suggestion for grouping disorders into five major "clusters" to create a simpler system that might reflect both common "causal risk factors" and clinical utility, as well as lessen the impact of high comorbidity rates (Andrews et al. 2009). However, this system was thought to present new problems (Wittchen et al. 2009). Until a more substantial research base can be interpreted confidently as supporting the creation of an overall organizing principle for a psychiatric classification, and until key dilemmas can be resolved with some degree of agreement about the trade-offs (e.g., lumping vs. splitting) to be reflected in a new structure (Kendler 2009b), the task force has adopted an incremental approach to restructuring DSM-5 (Table 2-2).
In DSM-5, disorders are grouped by development over the lifespan, starting with the chapter "Neurodevelopmental Disorders," which includes intellectual disabilities, communication disorders, autism spectrum disorder, attention-deficit/hyperactivity disorder, specific learning disorder, and motor disorders such as Tourette's disorder. Within each chapter, disorders likely to occur in childhood are listed first. This approach made it possible to drop the earlier section on disorders usually arising in childhood, and to integrate a lifespan perspective into all the mental disorder chapters. Each chapter was placed in close proximity to disorders thought to be closely related, at least clinically; for example, "Bipolar and Related Disorders" follows "Schizophrenia Spectrum and Other Psychotic Disorders" and precedes "Depressive Disorders."
Placement of disorders into chapters has also been reconsidered for DSM-5. Gambling disorder, for example, has been moved from the chapter "Impulse-Control Disorders Not Elsewhere Classified" in DSM-IV to "Substance-Related and Addictive Disorders" in DSM-5, reflecting recent evidence about behavioral addictions. Conditions such as posttraumatic stress disorder, previously in the chapter "Anxiety Disorders," and adjustment disorders, previously in their own chapter, have been aggregated into the new chapter "Trauma- and Stressor-Related Disorders."
The judgments involved in refining individual disorders can be seen most clearly by examining the types of changes made by the DSM-5 Task Force.
Some diagnoses described separately in DSM-IV and prior versions have been shown to have poor reliability and are poorly distinguished from each other. In DSM-5, five disorders previously distinguished as separate categories are combined into a single autism spectrum disorder. The task force found that this broader disorder can be distinguished from normal development and other "nonspectrum" disorders in a reliable way, but that distinguishing among the individual disorders showed variation over time and across settings, and was influenced by other factors such as language level. This change prompted expressions of concern from advocacy groups, who feared that it would increase stigma for patients who had previously been diagnosed with Asperger's disorder, for example. But the task force felt that stigma should be directly addressed as a problem on behalf of all of the patients with a condition in this spectrum. Some clinical researchers also suggested that the revision would lead to different prevalence rates, but experts in the working group disagreed with this report (Swedo et al. 2012).
|
Table 2-2. Changes in chapter organization from DSM-IV to DSM-5 |
|
|
DSM-IV |
DSM-5 |
|
Axis I |
(Removed) |
|
Disorders Usually First Diagnosed in Infancy, Childhood, or Adolescence |
Neurodevelopmental Disorders (Also moved to Trauma- and Stressor-Related Disorders; Disruptive, Impulse-Control, and Conduct Disorders; Feeding and Eating Disorders; Elimination Disorders) |
|
Delirium, Dementia, and Amnestic and Other Cognitive Disorders |
(Moved to Neurocognitive Disorders) |
|
Mental Disorders Due to a General Medical Condition |
(Moved to specific chapters) |
|
Substance-Related Disorders |
(Located below) |
|
Schizophrenia and Other Psychotic Disorders |
Schizophrenia Spectrum and Other Psychotic Disorders |
|
Mood Disorders |
Bipolar and Related Disorders Depressive Disorders |
|
Anxiety Disorders |
Anxiety Disorders Obsessive-Compulsive and Related Disorders Trauma- and Stressor-Related Disorders Dissociative Disorders |
|
Somatoform Disorders Factitious Disorders |
Somatic Symptom and Related Disorders |
|
Dissociative Disorders |
(Located above) |
|
Sexual and Gender Identity Disorders |
(Moved to Sexual Dysfunctions; Gender Dysphoria; Paraphilic Disorders) |
|
Eating Disorders |
Feeding and Eating Disorders Elimination Disorders |
|
Sleep Disorders |
Sleep-Wake Disorders Sexual Dysfunctions Gender Dysphoria |
|
Impulse-Control Disorders Not Elsewhere Classified |
Disruptive, Impulse-Control, and Conduct Disorders Substance-Related and Addictive Disorders Neurocognitive Disorders Personality Disorders Paraphilic Disorders Other Mental Disorders |
|
Adjustment Disorders |
(Moved to Trauma- and Stressor-Related Disorders) Medication-Induced Movement Disorders and Other Adverse Effects of Medication |
|
Axis II |
(Removed) |
|
Personality Disorders (Axis II) |
(Located above) |
|
Other Conditions that May Be a Focus of Clinical Attention |
Other Conditions that May Be a Focus of Clinical Attention |
For substance use disorders, the task force agreed to collapse the categories of abuse and dependence. In DSM-5, a dimensional approach is used to grade the severity of dependence.
A work group review showed that many somatoform disorders in DSM-IV relied on a determination of "medically unexplained symptoms." The task force found that this assessment could not be performed reliably and that distinctions among the individual disorders were unclear. Several of the DSM-IV somatoform disorders are now grouped into a single category, somatic symptom disorder, based on common features of somatic symptoms and cognitive distortions.
In DSM-IV, agoraphobia and panic disorder had been defined in relation to each other (panic disorder with agoraphobia, panic disorder without agoraphobia, and agoraphobia without panic disorder). The task force agreed to simplify this system by uncoupling the disorders and presenting two separate disorders defined without regard to the presence or absence of the other.
One change to criteria that generated public concern was the decision to eliminate the so-called bereavement exclusion in the criteria for major depressive disorder. The goal of this change was to make it possible for clinicians to determine when treatment might be needed for a major depressive episode occurring after a loss, when grief can resemble a depressive episode. After concerns were expressed about appearing to "medicalize" normal life reactions, the task force provided language to clarify that grief reactions are not automatically considered evidence of a major depressive disorder (Zisook et al. 2012).
Binge-eating disorder was listed in Appendix B of DSM-IV as a disorder that needed more study before being included in the list of Axis I disorders. In DSM-5, the disorder has been formally moved into the new chapter "Feeding and Eating Disorders." This move drew criticism as an unwarranted expansion into diagnosis of normal behavior. The task force disagreed with this criticism for several reasons. As noted in reviews of eating disorders, the average proportion of patients given an NOS or "other disorder" designation using DSM-IV and ICD systems ranged from 40% to 60% of patients seen in different clinical settings. The use of more adequate criteria for the narrowly defined categories, such as anorexia and bulimia, and the addition of binge-eating disorder ("bulimia without purging") have been broadly recommended for both DSM-5 and ICD-11 (Uher and Rutter 2012).
Mild neurocognitive disorder has also been added to DSM-5, because clinical and epidemiological data have demonstrated the presence of a clinical condition that does not reach the severity of major neurocognitive disorder but nevertheless requires assessment, care, and follow-up. A varied set of underlying illnesses can produce this disorder, including HIV-related cognitive changes and traumatic brain injury. The description of this condition reflects a new tension in the task of writing diagnostic criteria. In addition to reports of observable decline in intellectual functioning, formal neuropsychological testing is specified in the required criteria for a diagnosis. The task force recognized that some clinical settings may not have access to this service and referred to an "equivalent clinical evaluation" as an acknowledgment of the problems in requiring a specific testing procedure in a manual intended for widespread clinical use.
One area that posed a great challenge for the DSM-5 Task Force was personality disorders. Studies using DSM-IV had demonstrated that most individuals with a diagnosis of personality disorder qualified for more than one type and were usually given an NOS designation. For those people given a specific diagnosis, each type showed great heterogeneity among those with the same diagnosis. Within a given diagnosis, there was little temporal stability, so that diagnoses changed over time. Some researchers also felt that the 10 personality types included in DSM-IV did not provide adequate coverage of the range of disorders.
The work group considered revising the classification of personality disorders by adopting a hybrid system, which combines the traditional categorical model and a new model using quantitative ratings of personality domains and traits. In this approach, four of the prior personality disorder types would be eliminated from the classification. To assess the patient, a clinician would determine whether the individual has an impairment in either 1) self-identity/self-direction or 2) interpersonal functioning, as in intimacy and empathy. If so, and if one of the six remaining types is present, a diagnosis for that specific disorder would be given; if not, a series of rating scales would be used to characterize the features of the disorder, designated as personality disorder—trait specified. This category would replace the NOS category and would provide specific information about the individual's characteristics. Because it does not affect the judgment of whether a personality disorder is present, this category serves as a way to characterize patients, not to determine a diagnostic threshold (Krueger et al. 2012). This approach intrigued the task force, but the peer review committees felt it was not well enough established to be incorporated into the manual for routine clinical use. Section III of DSM-5 contains this alternative model for personality disorders, which will need more study before being formally incorporated into the classification system at some future date.
Along with making several revisions of the criteria for schizophrenia, such as removing the special significance given to bizarre delusions and Schneiderian first-rank hallucinations in DSM-IV, the task force deleted the five subtypes of schizophrenia. Evidence suggests that subtypes do not explain the heterogeneity of the disorder, and they do not correlate with longitudinal course or treatment outcome. Perhaps for these reasons, most of the subtypes have been used only infrequently. Replacing the subtypes is a new specification of psychotic dimensions, such as rating negative symptoms. This approach is expected to provide a better description of heterogeneity and of change over time than the subtypes provided.
An early focus of the effort to introduce dimensional measures for use with DSM-5 was on the assessment of important aspects of psychopathology that should be evaluated in an ongoing way with almost all patients (Helzer et al. 2008; Hyman 2010). These measures cut across diagnostic boundaries; for example, anxiety, somatic concerns, and substance use would be assessed in patients with any diagnosis. These measures were expected to help in identifying changes in a patient during treatment, as well as indicating areas for more clinical assessment outside the criteria sets of the specific diagnostic category. These cross-cutting measures performed well in the field trials for reliability and for clinical usefulness as reported by patients and clinicians (Narrow et al. 2013). Before they can be expected to gain widespread use, however, more tests of their performance in tracking clinical change will be needed. The cross-cutting dimensional measures are included in Section III of DSM-5.
In a classification system, the term categorical applies to a response that can be only one of two responses: yes or no. In the context of diagnosis, either the person is positive or negative for a particular diagnosis. The term dimensional means that the individual differences between persons in expression of a disorder are reflected in one or more ordinal scales (e.g., level of relevant symptoms). Therefore, a dimensional approach can identify individual clinically important differences (e.g., symptom intensity, frequency, duration) among those persons who have a positive categorical diagnosis. Even more important, a dimensional approach is meant to detect such differences among persons who have a negative categorical diagnosis (such as subsyndromal or prodromal manifestations)—that is, those patients given the NOS labels in the earlier DSM system.
Use of categorical diagnosis sacrifices power in testing and precision in estimation compared to dimensional diagnosis, resulting in slow progress in studies of the cause, course, or cure of psychiatric disorders. This problem arises because clinical research based on a categorical diagnosis assumes that everyone with a positive diagnosis is homogeneous, and that everyone with a negative diagnosis is also homogeneous, with regard to disorder status. Ignored individual differences within those two groups are regarded in statistical analysis as "random error," thus misleading the results of such research (Kraemer et al. 2004).
A major issue in defining a categorical diagnosis is the choice of "cut point," which is used to identify individuals of interest. Setting a cut point with a dimensional scale offers more precision in making the diagnosis: the higher the dimensional diagnosis score, the more likely the patient is to have the categorical diagnosis and thus the disorder. The extremes of such a dimensional diagnostic scale would be 1) absolutely no probability of a positive diagnosis and 2) absolutely no probability of a negative diagnosis. Deciding where between those extremes the optimal cut point should be placed depends on the use to which the cut point would be put. To screen for presence of a disorder, a lower cut-point would reduce false-negative rates. To identify cases that might benefit from a specific treatment, a higher cut point would reduce false-positive rates. Other purposes would produce other placements of the threshold; for example, different cut points may be needed to distinguish those individuals for whom a particular treatment may be most effective or those who are likely to have onset at some future time. Indeed, it may be that different cut points would be determined depending on which particular treatment is being considered (a specific psychotherapy or a specific drug). An advantage of having a dimensional diagnosis is the possibility of setting different cut points for different clinical applications, with each cut point optimal for one clinical application (Kraemer 2007).
A dimensional diagnosis is one special case of a dimensional measure. In DSM-5, the cross-cutting measures are dimensional but are not specific to any disorder, as a dimensional diagnosis would be. Cross-cutting measures are meant to be used at periodic clinic visits, whereas dimensional diagnoses are used when a specific diagnosis needs to be established or checked. Cross-cutting measures are meant for longitudinal follow-up of patients, particularly for management purposes, whereas dimensional diagnoses determine clinical status at one time point to guide decision making in the next time span. Cross-cutting measures may sometimes remind clinicians to check certain diagnoses, but are generally not screening measures themselves.
Some DSM-5 dimensional diagnoses did not perform well in the DSM-5 field trials, and the problems that dimensional diagnoses were meant to address were no better addressed in this instance than in earlier DSM versions (Regier et al. 2013). It was perhaps overly optimistic to try to move to dimensional diagnoses for psychiatric disorders en masse. Establishing a successful dimensional diagnosis may be easier one diagnosis at a time, with a greater focus on individual differences among those with and those without a specific disorder.
Just after the earliest U.S. classification of psychiatric illness was used in the 1840 Census, members of the newly formed American Statistical Association submitted to Congress a detailed critique of the system and its use. Even though the "system" used only two diagnostic terms, they had not been defined, the census enumerators had not been trained to use them, and no one tested whether the assessments were conducted reliably. The 1840 Census results had remarkable internal inconsistencies and, when comparisons to external measures were possible, could be shown to have tremendous inaccuracies (Gorwitz 1974).
That experience indicates the need to demonstrate how well a classification serves its purpose. The groups responsible for any new revision must set priorities for which standards to emphasize, such as valuing reliability higher than acceptability to clinicians or ease of learning.
In addition to the criteria for specific disorders, a classification system can be evaluated on several overall characteristics. One question is about feasibility, asking whether the classification is clear enough to be used by clinicians in their practice; for an international system like ICD, feasibility includes the ease of translating its terms into local languages. Another characteristic to assess is the suitability of the system for the patients seen in a clinical practice, through measures such as the rate of using residual diagnoses (e.g., "NOS" categories in DSM-IV). Surveys of users may be helpful in assessing these questions, as may examining results from field trial users or large-scale data sets of routine clinical encounters (First et al. 2004; Reed et al. 2011).
Several questions are important in considering the reliability and validity of diagnostic criteria:
A disorder is defined by the set of criteria specified in the classification. In a test of these disorders, the study relies on a diagnosis, which is an expert opinion that a certain disorder is found in a patient. For any diagnosis, categorical or dimensional, the total variance (individual differences among subjects) in a particular population comprises three nonoverlapping parts:
The reliability of a measure is the percentage of total variance free of noise, and validity is the percentage specifically from the signal, free of both noise and interference. Reliability is always at least as great as validity. Thus, a measure that is completely reliable may have zero validity, but a measure that has very poor reliability must also have very poor validity. For that reason, DSM-III and subsequent classifications have taken reliability as an essential first goal for disorder specifications.
To measure reliability, investigators take a sample of subjects from the population of interest and have each subject "blindly" evaluated using the diagnostic criteria over a period of time that is long enough to avoid carrying over noise from the first to the second diagnosis, but short enough that the disorder is unlikely to disappear in those who have it or to appear in those who do not. Agreement is measured as the percentage of variance without the noise component. The preferred reliability coefficients are the intraclass kappa for a categorical diagnosis and the intraclass correlation coefficient for a dimensional diagnosis.
Measuring validity is a greater challenge because of the difficulty of distinguishing between the signal and interference components of the total variance. Testing the validity of the diagnosis and the validity of the underlying concept of disease are difficult to separate. One approach to testing the accuracy of a diagnosis would be to study a sample of subjects and correlate the diagnosis with a "gold standard" indication of the disorder. However, psychiatry has no "gold standard." Some alternatives have relied on "best estimate" diagnoses from a panel of experts, or on routine use of a standard interview to guide the diagnostic assessment.
A more robust approach is to challenge validity in various ways. A diagnosis might be correlated with a diagnosis using another system of classification that has proved useful or with future outcomes known to be associated with the disorder. If the correlation is high, the results show convergent or predictive validity. The diagnosis might be correlated with the diagnosis of a different disorder, and if the correlation is low, the results show discriminative validity. The more such challenges a diagnosis can pass, the more likely it is to be valid. However, the process of documenting validity is long and tedious. In assessing the empirical evidence to make revisions for DSM-5, work group members considered both reliability and correlations with a set of validators derived from classic papers and prioritized by importance (Kendler 1980; Robins and Guze 1970; Table 2-3).
Although the DSM-5 Task Force and its work groups carefully reviewed each disorder and considered whether changes would be warranted in the descriptions or criteria sets, the reviews did not result in extensive changes to DSM-IV. This conservative outcome could be interpreted in two different ways:
|
Table 2-3. Validators used for DSM-5 |
||
[Note: While categories A and B would most typically be assessed after illness onset, they also could be assessed prior to illness onset as premorbid characteristics.] |
||
* Indicates highest priority.
Source. Kendler et al. 2009.
The most extensive changes were proposed for personality disorders, as explained in Section III of DSM-5. Through the removal of a multiaxial structure, these conditions were placed on the same level as other mental disorders. Along with mild neurocognitive disorder and some sleep disorders, personality disorders were redefined using a combination of categorical and dimensional approaches. This change represented an effort to continue with a clinically useful structure in retaining six categorical diagnoses, even though personality researchers had proposed revising it more radically. Because this proposal represents the most ambitious use of dimensional ratings considered for DSM-5, and was questioned by the peer review committees, it will need more testing before clinicians can be asked to use it in routine practice settings.
Of the various changes envisioned for DSM-5 and its near-term revisions, the use of quantitative ratings may require the largest adjustment by clinicians. As with personality disorders, this effort demonstrates the tension that can arise between the desire to introduce changes based on research and the desire not to disrupt current clinical practice. This conflict will arise in the future as more research-based and quantitative measures are proposed for incorporation into the classification. Disagreements about the timing of change and about the necessary strength of research findings to support notable changes will likely occur. In this respect, the development of a peer review mechanism to assess the scientific evidence and the clinical or public health importance of a change has the potential to become one of the most important innovations of the DSM-5 process. However, even the most thoughtful peer review can neither compensate for a lack of research nor overcome strong disagreements about the proper type of research to consider in reviewing proposed changes. In cases such as the proposal for personality disorders, task force members and the peer review groups disagreed about the nature and strength of evidence that should be applied in considering changes from DSM-IV criteria.
From the beginning of the DSM-5 process, the extensive rate of co-occurrence of mental disorders had been identified as a concern. Research has not yet clarified the possible explanations of this apparent comorbidity: Does it reflect varying presentations of a single, underlying condition that has been fragmented into separate disorders; or a shared pathway of different conditions; or the production of one disorder by a preexisting one? One approach to the issue was to consider whether different groupings of disorders—that is, lumping rather than splitting—would offer a new, more useful way to account for these presentations.
The proposal for a new "metastructure" was not adopted, because the empirical basis is underdeveloped for such a rethinking. The DSM-IV categories that were collapsed into a broader-spectrum disorder were closely related and thought to represent unnecessary distinctions among a closely related set of conditions. The creation of these new, broader categories (autism spectrum disorder and somatic symptom disorder) was not aimed at addressing comorbidity so much as poor reliability and conceptual overlap. Development of a new metastructure to use in classifying disorders awaits more definitive research that accounts for the comorbidity.
The most immediate application of a dimensional approach is the use of crosscutting measures to follow elements of psychopathology that are not tied to specific diagnostic criteria. The next step would be to develop dimensional measures to help set the threshold for diagnosis for an individual disorder. Initially, such a dimensional diagnosis may be tied to the criteria used for defining the category; with quantitative ratings, different cut points can be studied to determine where to place the threshold of diagnosis, as has been accomplished in hypertension or hyperlipidemia, for example (Hyman 2010; Kraemer 2007; Kupfer and Regier 2011).
Clinicians are likely to face increasing demands to use quantitative ratings as payers and public rating sites begin to focus on reported outcomes of treatment across medicine. These multiple demands may lead to redundant, contradictory, or cumbersome sets of instruments that are forced into clinical practice. If quantitative measures tied to the diagnostic system can be shown to be useful, it may be possible to simplify the expectations on practitioners even as the pressure to document outcomes increases. Ultimately, the use of quantitative ratings of underlying biological, psychological, and developmental processes may help in the clinical application of research findings on etiology and causal mechanisms, as forecast in the DSM-5 definition of mental disorders.
With the use of Internet connectivity and electronic publishing, the sponsors of DSM-5 have the potential to maintain it as a "living document" by revising it according to developments in scientific research and clinical practice, rather than being tied to long-term planning calendars.
In the past, revisions of DSM and ICD have involved cycles of relatively short periods of intense, large-scale activity to develop the classification, followed by one to two decades of seeming hibernation once the manual was produced. New technology has made it possible to introduce a new "process paradigm" for future revisions.
By the end of 2012, the DSM-5 revision process had demanded considerable effort by hundreds of participating experts from the United States and abroad, as well as consultants, advisers, field trial investigators, and peer reviewers. During this process, the increasing acceptance of electronic publishing and use of the Internet for communication and dissemination of information made it apparent that regular meetings of large work groups and associated committees may not be needed in the future. Rather than having major revisions scheduled according to the calendars and resources of the sponsoring organizations such as APA and WHO, smaller updates could be planned to follow incremental advances in scientific knowledge or changes in clinical practice.
For that reason, the new version of DSM was christened with an Arabic numeral, and updates will be numbered as with updated versions of computer software (DSM-5.1, DSM-5.2, etc.). This change in publishing and ease in communicating suggest that revisions can be undertaken in a series of smaller and timelier updates by a "permanent revision infrastructure," and only when needed for specific disorders. This change may also spare future major revisions from needing such a large mobilization of experts to rewrite the system two decades after DSM-5 is first published (Regier et al. 2009). Initial revisions of DSM-5 are likely to result from the introduction of new technology, such as clinical or biological measures similar to those recommended for assessing cognitive impairment and sleep disorders.
The analogy of the classification to a living organism also suggests that unnecessary or potentially harmful interventions should be minimized. Relying on a mammoth outpouring of voluntary effort by a system of work groups, study groups, and consultants may need rethinking, so that specific expert groups are convened only when it seems likely that changes are needed; otherwise, the process of organizing a panel for every condition may by its own momentum result in a feeling that some changes must be introduced, even prematurely, because another chance might not arise for years or decades.
For clinicians, who have established patterns of interpreting and applying the DSM-IV criteria, learning a new way of understanding and communicating about mental disorders will likely be unsettling. For researchers, who have invested time and resources into developing assessment protocols for particular disorders, especially in longitudinal designs, even minor changes to criteria can disrupt a program of orderly investigation. For trainees, who must shift from a system they have been studying to a new one that even their teachers do not yet know, the change may be intimidating because they are less likely to have the clinical experience or confidence to navigate the transition as easily as their faculty (Modi et al. 2012).
In the future, initial changes will likely concentrate on text descriptions of associated features of the disorders, with the increased integration of biological and other factors into the description of each disorder. Other changes are likely to include the introduction of additional tools, such as quantitative rating scales for general use (Hyman 2010). Presumably, smaller changes introduced when they are judged ready rather than en masse on a system-wide basis will be more accessible to clinicians, and will offer a timely, effective translation of research findings into clinical practice.
Since 1980, each revision of DSM has had the goal of basing changes on the best evidence available. However, few studies have conducted formal comparisons of different sets of criteria. The prevailing research strategy for understanding mental disorders has been to examine associated characteristics, such as risk factors, familial patterns, treatment response, or biological markers. This "top-down" approach uses the pre-specified disorder in a classification like DSM or ICD as the starting point for an investigation, but it can lead to a reification of the existing categories (Hyman 2010).
A more open-ended approach would be to examine basic cognitive, psychological, social, or biological processes and then determine how any dysfunctions in them are expressed clinically. This "bottom-up" strategy might lead to new paradigms for specifying and classifying clinical disorders. The Research Domain Criteria (RDoC) initiative of NIMH intends to use this approach to examine problems from the molecular level through the developmental history of an individual (Insel and Wang 2010). Such an ambitious undertaking fits well with the definition of mental disorder in DSM-5 but will require much time and funding; it will span research from genomics to social interactions. The complexity may make the findings hard to interpret and integrate across such different levels of explanation (Institute of Medicine 2012; Kendler 2012).
In preparing to develop ICD-11, WHO staff surveyed psychiatrists about their views on classification. More than two-thirds of the 4,887 respondents endorsed a desire to have a simpler, more flexible description of disorders without the explicit inclusion and exclusion criteria of the DSM approach (Reed et al. 2011). One approach to suit this preference would be providing descriptions of prototypes for each disorder. For personality disorders, for example, clinicians would determine how closely the patient matched the template description of a particular type. This approach has shown promise but remains controversial (Eaton et al. 2010; Westen 2012). Because it has not been as well studied beyond personality disorders, it is not yet clear how it would be used for other disorders; how general or less experienced clinicians would be able to match the patient to a prototypic description; or whether it would reduce the problems of comorbidity and distinguishing normal and abnormal (Maj 2011).
As future efforts to study mental disorders address the complex, multilevel interactions across multiple levels of analysis, mathematical modeling (factor analysis, cluster analysis, latent structural modeling, etc.) will be essential in searching for meaningful patterns in data. Finding the most useful models will be the challenge; textbooks present the classic caution that all models are wrong, and the task is to determine how wrong they must be before they stop being useful (Box and Draper 1987, p. 74). Besides examining the assumptions underlying each technique, researchers must follow a multistage process of hypothesis discovery, field validation, and dissemination with clinical validation (Institute of Medicine 2012).
DSM-5 has maintained continuity with prior versions by retaining a basic reliance on categorical diagnosis. At the same time, it has identified the opportunity for a new approach in the future, through use of quantitative measures such as the following:
The desire to establish psychiatric illnesses as "diseases" based on etiology and mechanism has motivated many new research efforts since the publication of DSM-III. Increased understanding of the complex interactions among different types and levels of contributing factors, and of the complicated mechanisms that contribute to the causal chain for a disorder, has led to a greater appreciation that single, easily comprehensible models of etiology and mechanism are unlikely to be found for psychiatric illnesses; the same recognition has developed in regard to many chronic conditions in other fields of medicine (ENCODE Project Consortium 2012; Hyman 2010; Kendler 2012; Loscalzo et al. 2007).
For the near future, progress in understanding and classifying mental disorders is likely to be incremental, with improvements occurring through a process of successive approximations ("epistemic iteration"; Kendler 2009a). Unless an unforeseen breakthrough produces a new paradigm for mental disorders, the process of improvement is likely to follow a process of accumulating small bits of knowledge, integrated in a piecemeal way to shape the evolution of the DSM classification system (Kendler 2005). Such deliberate, stepwise progress should make future revisions manageable for clinicians and satisfying for the field.
Key Clinical Points
American Psychiatric Association: Diagnostic and Statistical Manual: Mental Disorders. Washington, DC, American Psychiatric Association, 1952
American Psychiatric Association: Diagnostic and Statistical Manual of Mental Disorders, 2nd Edition. Washington, DC, American Psychiatric Association, 1968
American Psychiatric Association: Diagnostic and Statistical Manual of Mental Disorders, 3rd Edition. Washington, DC, American Psychiatric Association, 1980
American Psychiatric Association: Diagnostic and Statistical Manual of Mental Disorders, 3rd Edition, Revised. Washington, DC, American Psychiatric Association, 1987
American Psychiatric Association: Diagnostic and Statistical Manual of Mental Disorders, 4th Edition. Washington, DC, American Psychiatric Association, 1994
American Psychiatric Association: Diagnostic and Statistical Manual of Mental Disorders, 4th Edition, Text Revision. Washington, DC, American Psychiatric Association, 2000
American Psychiatric Association: Diagnostic and Statistical Manual of Mental Disorders, 5th Edition. Arlington, VA, American Psychiatric Association, 2013
Andrews G, Goldberg DP, Krueger RF, et al: Exploring the feasibility of a meta-structure for DSM-5 and ICD-11: could it improve utility and validity? Psychol Med 39:1993-2000, 2009
Box GEP, Draper NR: Empirical Model-Building and Response Surfaces. New York, Wiley, 1987
Boyd JH, Burke JD, Gruenberg E, et al: Exclusion criteria of DSM-III: a study of cooccurrence of hierarchy-free syndromes. Arch Gen Psychiatry 41:983-989, 1984
Cooper JE, Kendell RE, Gurland BJ, et al: Psychiatric Diagnosis in New York and London: A Comparative Study of Mental Hospital Admissions. New York, Oxford University Press, 1972
Eaton NR, Krueger RF, South SC, et al: Contrasting prototypes and dimensions in the classification of personality pathology: evidence that dimensions, but not prototypes, are robust. Psychol Med 41:1151-1163, 2010
ENCODE Project Consortium: An integrated encyclopedia of DNA elements in the human genome. Nature 489:57-74, 2012
Feighner JP, Robins E, Guze SB, et al: Diagnostic criteria for use in psychiatric research. Arch Gen Psychiatry 26:57-63, 1972
First MB, Pincus HA, Levine JB, et al: Clinical utility as a criterion for revising psychiatric diagnoses. Am J Psychiatry 161:946-954, 2004
Fulford KWM, Thornton T, Graham G: Oxford Textbook of Philosophy and Psychiatry. New York, Oxford University Press, 2006
Gorwitz K: Census enumeration of the mentally ill and the mentally retarded in the nineteenth century. Health Serv Rep 89:180-187, 1974
Grob GN: Origins of DSM-I: a study in appearance and reality. Am J Psychiatry 148:421-31, 1991
Grob GN: The Mad Among Us: A History of the Care of America's Mentally 111. New York, Free Press, 1994
Gruenberg EM: Epidemiology and medical care statistics, in The Role and Methodology of Classification in Psychiatry and Psychopathology. Edited by Katz M, Cole JO, Barton WE. Washington, DC, U.S. Government Printing Office, 1968, pp 76-99
Helzer JE, Wittchen H-U, Krueger RF, et al: Dimensional options for DSM-5: the way forward, in Dimensional Approaches in Diagnostic Classification: Refining the Research Agenda for DSM-5. Edited by Helzer JE, Kraemer HC, Krueger RF, et al. Washington, DC, American Psychiatric Association, 2008, pp 115-127
Houts AC: Fifty years of psychiatric nomenclature: reflections on the 1943 War Department Technical Bulletin, Medical 203. J Clin Psychol 56:935-967, 2000
Hyman SE: The diagnosis of mental disorders: the problem of reification. Annu Rev Clin Psychol 6:155-179, 2010
Insel TR, Wang PS: Rethinking mental illness. JAMA 303:1970-1971, 2010
Institute of Medicine: Evolution of Translational Omics: Lessons Learned and the Path Forward. Washington, DC, National Academies Press, 2012
Jablensky A: Towards ICD-11 and DSM-5: issues beyond "harmonisation." Br J Psychiatry 195:379-381, 2009
Jones DS, Podolsky SH, Greene JA: The burden of disease and the changing task of medicine. N Engl J Med 366:2333-2338, 2012
Kendell R: The concept of disease and its implications for psychiatry. Br J Psychiatry 127:305-315, 1975
Kendell R, Jablensky A: Distinguishing between the validity and utility of psychiatric diagnoses. Am J Psychiatry 160:4-12, 2003
Kendler KS: The nosologic validity of paranoia (simple delusional disorder): a review. Arch Gen Psychiatry 37:699-706, 1980
Kendler KS: Toward a scientific psychiatric nosology: strengths and limitations. Arch Gen Psychiatry 47:969-973, 1990
Kendler KS: Toward a philosophical structure for psychiatry. Am J Psychiatry 162:433-440, 2005
Kendler KS: Explanatory models for psychiatric illness. Am J Psychiatry 165:695-702, 2008
Kendler KS: An historical framework for psychiatric nosology. Psychol Med 39:1935-1941, 2009a
Kendler KS: Introduction to "A proposal for a meta-structure for DSM-5 and ICD-11." Psychol Med 39(12):1991, 2009b Kendler KS: Levels of explanation in psychiatric and substance use disorders: implications for the development of an etiologically based nosology. Mol Psychiatry 17:11-21, 2012
Kendler KS, First MB: Alternative futures for the DSM revision process: iteration v. paradigm shift. Br J Psychiatry 197:263-265, 2010
Kendler KS, Kupfer DJ, Narrow W, et al: Guidelines for Making Changes to DSM-5. Revised October 21, 2009. Available at: www.dsm5.org/ProgressReports/Documents/Guidelines-for-Making-Changes-to-DSM_l.pdf. Accessed August 13, 2012.
Klerman GL, Vaillant GE, Spitzer RL, et al: A debate on DSM-III. Am J Psychiatry 141:539-553, 1984
Kraemer HC: DSM categories and dimensions in clinical and research contexts. Int J Methods Psychiatr Res 16 (suppl):S8-S18, 2007
Kraemer HC, Noda A, O'Hara R: Categorical versus dimensional approaches to diagnosis: methodological challenges. J Psychiatr Res 38:17-25, 2004
Krueger RF, Derringer J, Markon KE, et al: Initial construction of a maladaptive personality trait model and inventory for DSM-5. Psychol Med 42:1879-1890, 2012
Kupfer DJ, Regier DA: Neuroscience, clinical evidence, and the future of psychiatric classification in DSM-5. Am J Psychiatry 168:672-674, 2011
Lehman AF, Alexopoulos GS, Goldman H, et al: Mental disorders and disability: time to reevaluate the relationship? in A Research Agenda for DSM-5. Edited by Kupfer DJ, First MB, Regier DA. Washington, DC, American Psychiatric Association, 2002, pp 201-218
Loscalzo J: Systems biology and personalized medicine: a network approach to human disease. Proc Am Thor Soc 8:196-198, 2011
Loscalzo J, Barabasi A-L: Systems biology and the future of medicine. Wiley Interdiscip Rev Syst Biol Med 3:619-627, 2011 21928407
Loscalzo J, Kohane I, Barabasi A-L: Human disease classification in the postgenomic era: a complex systems approach to human pathobiology (abstract). Mol Syst Biol 3:124, 2007
Maj M: Psychiatric diagnosis: pros and cons of prototypes vs. operational criteria. World Psychiatry 10:81-82, 2011
Modi JS, Avari J, Ferrando S: A changing DSM: the perspective of a trainee (abstract). Acad Psychiatry 36:345, 2012
Moriyama IM, Loy RM, Robb-Smith AHT, et al: History of the Statistical Classification of Diseases and Causes of Death. Hyattsville, MD, U.S. Department of Health and Human Services, Centers for Disease Control and Prevention, National Center for Health Statistics, 2011
Narrow WE, Clarke DE, Kuramoto SJ, et al: DSM-5 field trials in the United States and Canada, Part III: development and reliability testing of a cross-cutting symptom assessment for DSM-5. Am J Psychiatry 170:71-82, 2013
Pepe MS: Evaluating technologies for classification and prediction in medicine. Stat Med 24:3687-3696, 2005
Reed GM, Mendonça Correia J, Esparza P, et al: The WPA-WHO Global Survey of Psychiatrists' Attitudes Towards Mental Disorders Classification. World Psychiatry 10:118-131, 2011
Regier DA, Narrow WE, Kuhl EA, et al: The conceptual development of DSM-5. Am J Psychiatry 166:645-650, 2009
Regier DA, Narrow WE, Clarke DE, et al: DSM-5 field trials in the United States and Canada, Part II: test-retest reliability of selected categorical diagnoses. Am J Psychiatry 170:59-70, 2013
Robins E, Guze SB: Establishment of diagnostic validity in psychiatric illness: its application to schizophrenia. Am J Psychiatry 126:983-987, 1970
Sadler JZ: Values and Psychiatric Diagnosis.
New York, Oxford University Press, 2005
Sartorius N: Disability and mental illness are different entities and should be assessed separately. World Psychiatry 8(2):86, 2009
Sartorius N, Jablensky A, Regier DA, et al (eds): Sources and Traditions of Classification in Psychiatry. Toronto, ON, Canada, Hogrefe & Huber, 1990
Schaffner KF: Etiological models in psychiatry: reductive and nonreductive approaches, in Philosophical Issues in Psychiatry: Explanation, Phenomenology, and Nosology. Edited by Kendler KS, Parnas J. Baltimore, MD, Johns Hopkins University Press, 2008, pp 48-90
Spitzer RL, Fleiss JL: A re-analysis of the reliability of psychiatric diagnosis. Br J Psychiatry 125:341-347, 1974
Spitzer RL, Williams JB: American psychiatry's transformation following the publication of DSM-III. Am J Psychiatry 151:459-60, 1994
Spitzer RL, Endicott J, Robins E: Research diagnostic criteria: rationale and reliability. Arch Gen Psychiatry 35:773-782, 1978
Spitzer RL, Williams JB, Skodol AE: DSM-III: the major achievements and an overview. Am J Psychiatry 137:151-164, 1980
Stengel E: Classification of mental disorders. Bull World Health Organ 21:601-663, 1959
Swedo SE, Baird G, Cook EH, et al: Commentary from the DSM-5 Workgroup on Neurodevelopmental Disorders. J Am Acad Child Adolesc Psychiatry 51:347-349, 2012
Uher R, Rutter M: Classification of feeding and eating disorders: review of evidence and proposals for ICD-11. World Psychiatry 11:80-92, 2012
Ustün TB, Chatterji S, Kostanjsek N, et al: Developing the World Health Organization Disability Assessment Schedule 2.0. Bull World Health Organ 88:815-823, 2010
Westen D: Prototype diagnosis of psychiatric syndromes. World Psychiatry 11:16-21, 2012
Wittchen HU, Beesdo K, Gloster AT: A new meta-structure of mental disorders: a helpful step into the future or a harmful step back to the past? Psychol Med 39:2083-2089, 2009
World Health Organization: Manual of the International Statistical Classification of Diseases, Injuries, and Causes of Death, 6th Revision. Geneva, World Health Organization, 1949
Zachar P: Progress and the calibration of scientific constructs, in Philosophical Issues in Psychiatry II: Nosology. Edited by Kendler KS, Parnas J. Oxford, UK, Oxford University Press, 2012, pp 21-34
Zachar P, Kendler KS: Psychiatric disorders: a conceptual taxonomy. Am J Psychiatry 164:557-565, 2007
Zisook S, Corruble E, Duan N, et al: The bereavement exclusion and DSM-5. Depress Anxiety 29:425-443, 2012