Abstract
It is an inherent aspect of human nature to believe that one lives at a momentous time in history, at a cross-road. Often this is an illusion born of short horizons. Most history is evolutionary, not revolutionary. Nonetheless, as far as psychiatric diagnosis is concerned, we may indeed be approaching such a cross-road. Before very long, our complex, yet limited descriptive nosology of psychiatric disorders may help us to discover much simpler, more incisive explanatory models of psychiatric illness.
Virtually every civilisation, as well as nearly every tribe, has developed a model for classifying the range of human behaviour and the deviance from norms, which we now clumsily call mental disorders. The reason for this universal interest seems clear: unpredictable behaviour can be terrifying and disruptive both for the individual and society. Labels provide an usually false yet comforting sense of being able to control the uncontrollable. Some classification models have been descriptive, while others have attempted to be explanatory. The consistency of descriptions across cultures and times suggests that people and psychiatric illnesses have not changed significantly in the last 5000 years. The explanatory models such as the four humours, demonic possession, or even our current alchemy of neurotransmitters, may be useful bridges to future explanations but tend to age quickly because they are based on limited knowledge, which is easily superceded by new discoveries. The current descriptions of causal attributions in psychiatry reflect the fact that our medical theories have been diverse and are evolving.
The Diagnostic and Statistical Manual of Mental Disorders, fourth edition (DSM-IV) [1], is an important current system of diagnostic classification that was developed as a descriptive nosology. This evidence-based classification system has shaped psychiatric research and clinical practice by providing agreed-upon definitions of psychiatric disorders based on the current state of empirical data. We now know a great deal about the epidemiology, courses and treatment responses of psychiatric disorders. We also know a great deal about the neurobiology of the brain and the genetics of transmission and expression of psychopathology. We continue to learn at a remarkable rate. What we do not yet know is how the neurobiology of the brain becomes disordered to cause the major psychiatric illnesses. Disorder by disorder, this knowledge will probably become available within the next one to three decades. Soon, we are likely to have a clearer understanding of some of the causes of obsessive–compulsive disorder, schizophrenia and bipolar disorder. As time passes, more of these disorders will be diagnosed based on causes rather than descriptions, just as pneumonia went from a descriptive to an aetiological diagnosis.
In this way, the explosion of knowledge from neurobiology and human genetics is enabling us to understand the causes, not only the presentation of psychiatric disorders. Just as the periodic table made atomic theory possible and as Linnaeus' ordering of the species facilitated Darwin's unifying explanatory evolutionary theory, so the most current versions of a descriptive psychiatry will serve as the necessary stepping stone for the development of a deeper level of aetiological synthesis. We need the descriptive model to guide the science toward the explanatory model. A descriptive diagnostic nosology is a necessary tool, a method not an end product.
Despite its limitations, DSM has served a useful purpose by improving psychiatric research, facilitating clinical communication, and establishing the epidemiology of psychiatric disorders. In this paper, we will describe the development of the DSM-IV, its purposes and limitations, and how we hope it will be superceded by future advances in our field.
A short history of DSM-IV
Most systems of psychiatric classification and diagnosis are the result of a wise professor model in which an especially influential man or woman describes his or her patients and classifies these observations to create a list of names. The most important modern example derives from the work of Kraepelin. His classification system, which informs DSM-IV and ICD-10, began as a careful textbook with a detailed table of contents that eventually became the basis for a widely accepted system of modern psychiatric nomenclature.
The early versions of the DSM and the ICD were based on the extension of the professor model. Instead of relying on the observations of a single individual, these diagnostic systems have been developed by ever multiplying committees of experts. The essential problem with an expert model, whether it involves one individual or a committee of individuals, is that it risks being arbitrary. Who is designated an expert and how are his or her observations and systems of classification validated? The innovation of DSM-IV was to reach beyond the expert model to attempt to develop an evidence-based psychiatric nosology based on a systemic review of the available scientific data about psychiatric disorders. It was not meant to be paradigm shift but rather a good workman-like improvement informed by a comprehensive review of available scientific knowledge.
The primary goal of DSM-IV was to ground the diagnostic categories within an empirical framework, by moving beyond expert consensus and using documentation derived from comprehensive literature reviews, re-analysis of existing data sets, and rigorous and extensive field trials. For example, specific criteria sets, diagnostic thresholds, frequencies and durations of specific symptoms were based, as much as possible, on empirical data. The revision process succeeded in providing a systematic and comprehensive review of the current empirical data and documentation of reasons for changes.
Despite the emphasis on empirical data, it must be recognised that many important questions remain unanswered in the currently available databases. The expert consensus model remained an important part of the development of the DSM-IV criteria. To decrease the arbitrary nature of expert opinion, the American Psychiatric Association attempted to create a comprehensive, inclusive and open process for gathering and using expert opinion. A task force of 27 members with Allen Frances as the Chairman was convened. Work groups of five to 10 members were established to focus on individual diagnoses. These work groups were supported by advisory groups of 50–100 members. Members of all groups were chosen to reflect psychiatric expertise, to represent differing views and unresolved controversies in the field, and to have an ability to adopt a consensus view.
As decisions were made about criteria for inclusion and definitions of psychopathology, comments from the entire field were widely sought and incorporated into the final decision-making process. Before a draft was finalised, an options book [2] was published and distributed to solicit opinion about possible alternative criteria sets. Finally, DSM-IV source books [3], companion publications to the DSM-IV itself, were published providing the field with the opportunity to read the literature reviews, the data re-analyses and the field trial results and the details of the task force's decision-making process.
A three-stage review of the current state of psychiatric knowledge and expert opinion was developed. First, systematic reviews of the published literature on current or proposed diagnoses were done. Second, previously collected but unpublished data sets were reviewed and re-analysed. Third, new data were collected and analysed under the auspices of field trials.
The inclusion of focused field trials provided an opportunity to test the validity and reliability of diagnostic categories in real-life clinical settings. Twelve diagnostic categories were chosen for the field trials: antisocial personality disorder; disruptive behaviour disorders; mixed anxiety depressive disorders; mood disorders; obsessive–compulsive disorder; panic disorder; pervasive developmental disorders; posttraumatic stress disorders; schizophrenia; sleep disorders; somatisation disorder; and substance-related disorders. Each field trial compared DSM-III, DSM-III-R and ICD-10 criteria with DSM-IV proposed criteria in five to 10 sites with five to 100 subjects per site. Diverse sites were selected with patients drawn from a range of sociocultural and ethnic backgrounds to ensure the results could be generalised. Each field trial collected information on the reliability and performance characteristics of each criteria set as a whole, as well as the specific items within each criteria set.
The final DSM-IV Work Group recommendations were informed by the data obtained through the three-stage process of empirical documentation, the extensive written comments from the advisors, and the correspondence in response to the DSM-IV options book. The final DSM-IV was published in 1994.
Why the current descriptive system is necessary but not sufficient
DSM-IV represents our current, best attempt to provide a workable, reproducible psychiatric classification system. The DSM reliance on a descriptive nosology based on observable behaviours, signs and symptoms, rather than theories of aetiology, has facilitated reliability of diagnosis, communication about disorders, and research about psychiatric illness. There is no doubt that the current state of scientific knowledge of psychiatric disorders and treatment was made possible by a valid, reproducible, standardised psychiatric nosology. We cannot agree on causes or treatments if we do not have a shared descriptive language.
Yet the current nosology also has significant limitations for research and clinical practice. Our understanding of the nature and definition of diseases changes as knowledge, methodologies and theoretical approaches advance. The understanding that DSM-IVcategories are shaped by the current state of knowledge and measurement tools is critical for them to be used effectively, both in research and clinical practice. The researcher must use the DSM criteria as a means to move beyond the confines of the current classifications, while the clinician must combine clinical judgement and comprehensive, multidimensional assessment and treatment with the DSM categories.
Both researchers and clinicians must ask, as Peter Jensen does in a recent article about the limitations of DSM-IV, ‘What kinds of explanations are favoured by the diagnostic system? What clinical and societal needs are met by the classification system? What types of discoveries might be precluded by the classification system because certain classes of data are difficult to assess within the taxonomic system?’ [4, p.235]. Without these questions, the categories themselves limit, rather than expand, our understanding of psychiatric illness and our patient's experiences.
In the area of research, the DSM emphasis on observable symptoms may facilitate reliability without confirming the validity and/or clinical usefulness of these descriptive categories. The relative strength of empirical data in biological, pharmacological and behavioural psychiatry necessarily results in their increased influence on the descriptive system, with relative neglect of developmental, psychodynamic or interpersonal issues. Clearly, development of more sensitive tools, in all areas of psychiatry will help to refine, and even unify, the picture of psychopathology.
An effective clinician must know the symptoms, the diagnostic categories and the predictive power of each diagnosis in DSM-IV, but also must be aware of the limitations of a categorical approach. He or she must not worship at the altar of diagnosis, because in real life patients do not fit neatly into diagnostic categories. Clinicians must continue to appreciate the importance of clinical judgement and an individual patient's personal history. Unfortunately, many young psychiatrists training in the DSM era are learning diagnostic classification as the primary way of thinking about patients, without understanding the patient and his or her presentation within a biopsy-chosocial context. The result is that patients become their symptoms and the descriptively defined syndromes become narrow containers limiting the clinician's understanding of the patient's illness and treatment needs. Some managed care review criteria are a good example of how a literal and narrow reading of psychiatric illness using DSM criteria can lead to refusal to provide psychiatric care.
The multiaxial system was an attempt to facilitate a dimensional formulation of psychopathology to assess environmental factors, stressors, comorbidity, character pathology and functioning; yet in practice it is not carefully used. Clinicians have not embraced the multiaxial system because it is cumbersome. Axes II–V are often completed as a hasty afterthought and rarely used to formulate a comprehensive dimensional understanding of the patient's illness. A more user-friendly and meaningful system to assess the multiple dimensions of illness needs to be developed.
Future directions
The present diagnostic system based on a descriptive approach was a great advance over previous systems which had a very low reliability. This improvement has facilitated research and clinical communication. Yet, the descriptive system of the DSM-IV is only a temporary way station to be replaced, probably disorder by disorder, with more useful aetiologic models of classification as our research develops. This is an interactive process. It is impossible to do valid research without a good clinical system of description, but the findings of that research will eventually inform a much deeper and richer method of classification than can be provided by description alone. The path leading from the DSM-IV, from descriptive psychiatry to a psychiatry based on causality, will probably involve the following:
(1) The current explosion of knowledge in neuro-science and genetics of psychopathology will continue and eventually will reframe our understanding of the causes and treatments of mental illness. The current biological revolution will lead to an understanding of the neurobiology of clinical syndromes, including the effects of pharmacological and psychosocial treatments. Imaging techniques and human genetics will help us find markers of illness that will also help to define the categories of psychopathology. Within the next five to 25 years, we will develop an explanatory model for at least some of the major psychiatric disorders. The first diagnoses that will crack are those that present in the most homogeneous way, with strong family histories, and with relatively predictable treatment responses. Major depression with its heterogeneous presentation and treatment reponses will probably be much more difficult. We should not expect that a single gene will explain the transmission and expression of a complex psychiatric illness. Nonetheless, the information from genetics will, with other findings, effect the ways that we cluster symptoms into syndromes and how we choose treatments.
(2) We need to continue to validate our current descriptive categories while, at the same time, strive to define and operationalise all clinical phenomena, including functional interactions between the patient and his or her family and environment, as well as internal, intrapsychic phenomena. We need a ‘topography’ [4, p.243] of clinical course, a developmentally and longitudinally based model which aims to reflect the aetiologies, prognosis, and variable treatment responses of psychiatric illness as assessed within an individual's context of relationships, risk and protective factors, genetic profile, and current and past stressors. The need for a developmentally informed diagnostic system is most acute for the definition of psychopathology in children and adolescents because most of the current diagnoses are based solely on models of adult psychopathology. The absence of a developmentally sensitive, interactive or longitudinal perspective in the DSM system of classification limits the useful of the categories for both research and clinical assessment and treatment of children and adolescents.
Perhaps the best example of why a developmentally appropriate diagnostic system is important can be found in the role kindling probably plays in the pathogenesis of psychiatric disorders. Although the evidence is incomplete, it seems likely that the kindling model of psychiatric illness may extend far beyond bipolar disorder and schizophrenia to include most psychiatric disorders. Early identifications of psychiatric disorders and appropriate implementation of timely treatment are critical to reducing the risks of kindling and decreasing lifetime morbidity. Early identification is complicated by developmental differences in presentation. We must refine the descriptive nosology to include differences of presentation across the life cycle, while also seeking to identify reliable genetic tests or neuroanatomical changes that will aid in early identification and prevention of psychiatric disorders.
(3) Clinically, we need to continue to maintain a balance between the complexity and limitations of our existing diagnostic categories and the complexities and idiosyncrasies of our patients'lives, while keeping up to date with emerging scientific knowledge.
The increasingly awkward term ‘mental disorder’ reflects the limitations of the DSM to express the current and future state of our knowledge. The term suggests a mind–body dichotomy that is proving to be illusory, as we develop complicated neurochemical and anatomical models of psychopathology. Gaining further understanding of the pathophysiology of brain disorders, as well as the biological effects of psychiatric interventions, will enable us to move from a descriptive model to an integrative model that reflects the interplay of biological, psychological, environmental, and social variables affecting the expression and treatment of psychiatric disorders.
Science strives for simplicity of explanation. Descriptive models tend to be piecemeal and complicated. We are at the epicycle stage of psychiatry where astronomy was before Copernicus and biology before Darwin. Our inelegant and complex current descriptive system will undoubtedly be replaced by explanatory knowledge that ties together the loose ends. Disparate observations will crystallise into simpler, more elegant models that will enable us not only to understand psychiatric illness more fully but also to alleviate the suffering of our patients more effectively.