|
PERSPECTIVEMental Disorder Diagnostic Theory And Practical Reality: An Evolutionary Perspective
In the current legislative debate about mandating parity of insurance coverage for mental disorders, many question the use of DSM-IV diagnostic criteria to indicate benefit eligibility. Some have indicated that resistance to adopting parity legislation has been driven by inadequacies in the theory underlying psychiatric diagnosis. This paper takes issue with that perspective and reviews the scientific basis for the current classification and the advances in research and clinical practice made possible by reliable diagnostic criteria. As hypotheses that are subject to empirical testing, the DSM-IV criteria have set the stage for further advances—independent of the economic and political forces that are now playing out in the parity debate.
In the preceding paper David Mechanic addresses a recent policy debate on the scope of parity in insurance coverage for mental disorders by focusing on the nature of diagnostic conceptions and criteria contained in the fourth edition of the American Psychiatric Association’s Diagnostic and Statistical Manual (DSM-IV).1 The issue is framed by identifying a concern expressed by opponents of insurance parity legislation that covering all disorders defined in the DSM would vastly increase demand and cost—both because of the overinclusiveness of the disorders and because some of the available treatment modalities may be sought for self-improvement and self-actualization. Although the latter issue of treatment specificity is an important subject in its own right, the crux of the issue addressed in this Perspective is the relationship between DSM-IV definitions of mental disorders and a credible concept of service need. Possible reasons for opposition to parity legislation raised include (1) the current unconfirmed diagnostic theories of mental disorders that have impeded an ability to predict treatment response; (2) the impact of psychoanalytic theory on society, which has led many to believe that serious disorders are not discrete conditions but are on a continuum with mild dysfunctional emotions and behaviors that can only be corrected by assessing unconscious conflicts and motivations to increase understanding and control; (3) the high rates of mental disorder identified in recent epidemiological studies that have led some policy analysts to fear a bottomless pit of treatment need; and (4) fear that some people with minimal symptoms or even those without currently defined disorders will inappropriately use mental health services. Since some of these perceptions could play a role in the public’s and policymakers’ judgments about the value of mental health treatment, I have been asked to address those related to the development and current status of DSM-IV diagnostic criteria. I do so from the standpoint of one involved in the development of diagnostic criteria and assessment instruments and the use of these for epidemiological and health services research studies over the past twenty-five years.
The initial thesis of Mechanic’s paper is that linking a diagnosis to treatment need requires a "valid disease theory." In this framework, such a theory is necessary to point to an "underlying disorder," which then provides information on expected course or natural history, causes, and needed specific treatment. The clear implication is that without a valid disease theory, psychiatry has no scientifically credible way of saying something intelligible about the natural history of disorders, their clinical course, or useful treatment strategies. The characterization of mental disorders is that existing diagnoses have few objective measures and are "more conveniences to assist communication and further inquiry than they are representative of confirmed diagnostic theories" (page 10). Clearly, every physician would prefer to have explicit etiological agents, well-described organ pathophysiology, anatomical localization, and specific treatment interventions to eliminate every disease, disorder, or syndrome. However, to hold psychiatry or any other branch of medicine to this standard is not possible or even necessary to advance our understanding and treatment of diseases.2 Although all medical researchers continually test and seek to improve the knowledge base for existing diagnostic concepts, even a cursory reading of medical history reveals that syndrome-based diagnoses, with no known causal mechanisms or "disease theory," enabled the control of cholera, pellagra, and dropsy (congestive heart failure) long before the cholera vibrio, niacin deficiency, or action of digitalis were understood.3 To imply that causal mechanisms are needed to develop research and treatment strategies for mental disorders is simply belied by the evidence. Certainly, pharmacologic treatments for psychotic, depressive, and anxiety disorders have emerged in the absence of clear causal mechanisms and often serendipitously as the result of attempts to develop anesthetics or treat other diseases such as tuberculosis.
Instead of being a requirement for defining treatment need, one can argue that valid diagnostic theories, as defined by Mechanic, follow major treatment advances more often than they precede them. There is also little question that an inaccurate diagnostic theory can have a strong impact on society and the progress of scientific research. It is hard to dismiss the impact on parents and families that psychoanalytic disease theories have had in assigning blame to "schizophrenogenic" mothers, when the only causal mechanism considered for all mental disorders, on a continuum from mild anxiety conditions to schizophrenia, was a "reaction" to early childhood development and child-rearing experiences.
The publication of the DSM-III in 1980 established a scientifically credible approach to classifying mental disorders, based on characteristic symptom patterns reported by the patient and on observable signs such as psychomotor retardation and mental status changes, with specified duration, exclusion criteria, and some internal impairment criteria.4 These criteria sets are based on an extensive analysis of available research, have achieved wide international consensus by experts, and are intended to represent hypotheses that can be tested for their reliability across multiple observers and modified if more statistically significant criteria are shown to predict genetic aggregation, clinical course, and treatment response. The DSM-IV is now the accepted scientific standard in the field of mental health for determining indications for safe and effective treatment of medications in the U.S. Food and Drug Administration (FDA), for obtaining grants from the National Institutes of Health (NIH) for research on mental and addictive disorders, and for all mental disorder disability determinations in Medicare, the Department of Veterans Affairs (VA), and private disability insurance companies. It is used as one component of medical necessity determinations for all managed behavioral health care organizations (MBHOs), which manage more than 90 percent of private insurance mental health benefits, and it is used in all state Medicaid programs to assist in determining eligibility for mental health services. In brief, the DSM-IV criteria are incorporated into more than 650 state and federal laws as the existing standard for defining mental disorders. The evidence base for establishing diagnostic criteria has also been used in the development of treatment guidelines that are geared to the DSM-IV diagnoses. The American Psychiatric Association (APA) has now published eleven evidence-based guidelines that depend on clinical trials of treatment efficacy and effectiveness for both psychosocial and psychopharmacological treatments.5 These guidelines form the basis for most "medical necessity" determinations of MBHOs.
The DSM-III and DSM-III-R (revised) diagnostic criteria were used in large-scale epidemiological studies such as the Epidemiologic Catchment Area (ECA) study and the National Comorbidity Survey (NCS) in the early 1980s and 1990s. This usage revealed that the criteria were necessary but not sufficient to differentiate people with clinically significant levels of psychiatric symptoms from those with similar symptoms but no functional impairment.6 Determining prevalence. Considerable optimism accompanied the initial publication of the DSM-III and the launch of the ECA study, which incorporated DSM-III criteria in its case-identification instrument. The expectation was that the high estimates of mental disorder found in older epidemiological studies would now be much lower. For example, significant psychiatric symptoms were noted in more than 80 percent of the population in the classic midtown Manhattan study, and about 20–25 percent of this New York City population was identified as needing treatment.7 However, the ECA followed the president’s commission report, under President Jimmy Carter, which used the most conservative available epidemiological estimates and identified an annual need for treatment in 10–15 percent of the population.8 Hence, the 32 percent lifetime prevalence identified in the ECA first-wave survey and the 50 percent rate found in a single-wave survey of the NCS, based on DSM-III-R criteria, was unexpectedly high and suggested the value of using additional criteria for determining treatment need. In the absence of some "gold standard" for differentiating less severe forms of illness from extreme ends of normal variation, I and some others also suggested that a higher symptom threshold would be more useful for defining disorders.9 By 1994 the DSM-IV addressed this issue explicitly by requiring "clinically significant" impairment or distress, in addition to symptomatic criteria, to qualify for any mental disorder diagnosis. Afterward, as discussed by Mechanic, William Narrow and colleagues published a major reanalysis of both the ECA and NCS prevalence rates by applying previously unused but available data from questions embedded in the diagnostic instruments that had been inserted to assess clinical significance of syndromes.10 Important omission. Ironically, all of us connected with the development of these instruments and their scoring in the ECA were confident that an assessment of the clinical significance of component symptoms was sufficient to identify disorders. Hence, we simply failed to use the additional information that was obtained from the questions that were designed to determine if there was clinically significant distress or impairment when the aggregate group of symptoms occurred together for the duration specified in the DSM-III. Contrary to critiques by Mechanic and others, these additional clinical significance questions are virtually the same ones used to assess the intensity of individual symptoms and simply ask specifically if the entire syndrome interfered with a person’s life a lot or if it caused enough distress for the person to tell someone else about this experience. Mechanic would like to have disorder status assessed independently of help-seeking behavior for theoretical reasons. However, available methods for assessing distress associated with mental disorders require expression of such feeling states to someone, including health professionals, and make total isolation impossible. However, it is noteworthy that only 25 percent of people with clinically significant disorders had any contact with health care professionals in one year—with 75 percent meeting diagnostic criteria exclusively on the basis of disability or distress revealed to others not directly involved in providing mental health services. The impact of using the added syndrome-specific clinical significance scoring data in Narrow and colleagues’ reanalysis was a lowering of combined overall mental disorder and substance abuse prevalence rates from about 28 percent in one year to 18.5 percent—an estimate that still does not directly measure treatment need. Nondisease validators. Mechanic’s and Robert Spitzer and Jerome Wakefield’s critiques of Narrow and colleagues’ paper—that the resulting estimates are constructed on a misguided theoretical basis—once again misses the point that no disease theory is being invoked and that none is really needed.11 The only theoretical construct needed is an assumption that some disease process is at work, which we can still only begin to identify by validators such as family aggregation, separation of disorders by distinct clinical syndromes, course of illness, laboratory studies, and response to treatment.12 As a practical matter, the cutpoints of pathology are being made on a statistical rather than a theoretical basis. This involves assessments of the probability that syndromes defined by the DSM criteria will predict clinical course, treatment outcome, and genetic or other biological correlates in much the same way that hypertension is defined by the statistical association between arbitrary blood pressure thresholds and adverse cardiovascular, cerebrovascular, and renal events.13 Seemingly arbitrary cutoff points of five out of nine symptoms for major depression and various levels on symptom severity scales are closely correlated with clinical course and treatment response.
This may have been a serious question in 1980, when the DSM-III was first published; in 1984, when the ECA findings began to emerge; or even in 1994, when the NCS findings first came out. However, with the remarkable development of MBHOs in the late 1990s and into the twenty-first century, not many with knowledge of the current state of "mental disorder diagnosis" and the way in which "medical necessity" for care is determined will consider these two concepts to be isomorphic. Need for care should not be based solely on diagnosis, in mental health or the rest of medical care. The authoritative book Unmet Need in Psychiatry provides a rich analysis of how to conceptualize need and unmet need for mental health services. In addition to emphasizing the requirement for linking diagnosis, disability, and duration as a basis for these determinations, it also pays attention to the cost-effectiveness of available treatments for reducing disease burden on both the individual and society.14 Also, in the remarkable Global Burden of Disease report of the World Health Organization (WHO), there is already good evidence that statistical associations of disability with current DSM disease definitions have become operationally useful in defining need for care.15 Limits to necessity for care. Certainly, some advocates have tried to imply that all mental disorders require treatment and have tried to use the highest possible prevalence figures to argue for higher levels of service, training, and research funding. When state mental health commissioners, insurance companies, and members of Congress considered such high rates to lack credibility, they requested definitions of severe mental illness, biologically based illnesses, or serious mental illnesses to limit the "need" definitions to a manageable level.16 This was when demand-side benefit design controls were primarily being used to limit the flow of patients by the diagnoses that could be used to "demand" treatment services. The entire service delivery system changed when managed care companies began focusing on "medical necessity" criteria that included both the diagnostic criteria of the DSM-IV and estimates of disability, risk of harm to self or others, and the evidence base for predicting that available treatments could be cost-effective.17 In this new administrative environment, need is defined not by the patient or the clinician but by the managed care company. However, when abuses of this MBHO power have been egregious, the regulatory power of the state has been used to give greater weight to the need perceptions of both patients and clinicians.18 Whereas the previous "moral hazard" was that patients or clinicians would demand payment for unneeded care, the most important current hazard is that needed care will be denied to reduce health care costs and to boost MBHO profits. Avoiding the "bottomless pit." To address the critiques of Mechanic and others that the prevalence of mental disorders is still too high and that to treat all would result in a bottomless pit of costs, I offer several considerations. First, the more than 12,000 diagnoses contained in the official International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) for all of medicine, including about 250 in the mental disorders section, are used primarily for statistical purposes, ranging from documenting reasons that patients identify for a visit (as minor as male-pattern baldness) to vital statistics reports on causes of death. In general, it is not diagnoses that are of concern to insurance companies; it is entitlement to specific forms of treatment that are limited. Hence, a doctor who diagnosed normal baldness on an insurance claim form after a differential diagnosis of hair loss would not raise insurance reimbursement issues. However, if a cosmetic, surgical hair transplant were recommended, some insurance plans would disallow such charges. In the special case of mental disorders, both the WHO and the American Psychiatric Association (APA) have found that it is not sufficient to have simple names of diagnoses listed without having explicit diagnostic criteria to define certain major conditions. Major differences in diagnostic practices such as those identified in the historic U.S./U.K. study made the need for explicit definitions obvious—there were much higher rates of schizophrenia in the United States and much higher rates of manic-depressive illness in the United Kingdom until explicit criteria and assessment procedures were standardized.19 The second point is that our ability to improve descriptive precision has continued to advance. These improvements include the development of epidemiological studies that combine existing diagnostic criteria in the DSM-IV (which set the lowest threshold for a disorder category), with dimensional measures of severity (for example, the Hamilton Depression Scale) and specific measures of disability (for example, the SF-12 or SF-36). It will be interesting to see what the prevalence rates are when studies using this approach in follow-up NCS investigations are actually published. Ongoing changes in diagnoses. An additional approach will offer a more fundamental reformulation of diagnostic concepts in the process of developing the DSM-V revision. Over the past twenty-five years since the DSM-III revision, these highly reliable diagnoses have been used to define phenotypes for genetic studies and indications for pharmacologic and psychosocial treatments. As a result of a reliable diagnostic definition, a cumulative body of research has developed that now includes preclinical animal models, functional imaging, and other neurobiological studies of anatomical variations between patients and controls; clinical trials of many new treatments for specific disorders; and epidemiologic, cross-cultural, and health services research studies. The recent publication of A Research Agenda for DSM-V identifies a wide range of research studies that identify biological markers for specific mental disorders, pharmacogenomic studies that relate specific treatments to patients’ genetic subtypes, and the effect of fundamental reformulations of some diagnoses from categorical to dimensional entities such as hypertension.20 All of these evolutionary changes should enable us to get closer to an understanding of the pathophysiology and etiology that will guide future therapeutic and preventive interventions, with increasing benefit to patients. With less reliance on syndrome phenomenology for diagnostic criteria, it will be possible to uncouple the "clinical significance" concepts of disability and distress from disorder criteria and to evaluate treatment need on the feasibility of preventing the onset of disorders as well as curing established diseases.21 Allocation issues remain. Regardless of the advances that will come to improve diagnostic validity and treatment effectiveness, any future health care system will still need some mechanism for allocating scarce health care resources that will not simply depend on the validity of existing diagnostic theory. Indeed, the scientific basis of current diagnostic criteria for mental disorders has been sufficient to propel an enormous increase in the availability and effectiveness of psychiatric treatments. Likewise, our society’s ability to diagnose and treat other medical/surgical disorders has outstripped our ability to deliver and pay for services for all who could benefit. Systematic discrimination continues. Although major social policy and ethical decisions will confront the United States as we address these issues, one obvious area of systematic discrimination against a whole class of patients remains and can be easily remedied. This discrimination is demonstrated by a continuing difference in reimbursement rates for mental disorder treatment in Medicare and in many private insurance plans, even where MBHOs have drastically reduced costs. The reasons for such discrimination are to be found not in the DSM-IV diagnostic system, but in the historical patterns of assigning care for the mentally ill to the states and to the ongoing stigma associated with mental disorders—no matter how they are defined. Medicare reform to equalize copayments for all disorders and the enactment of the current mental health parity legislation for private insurance plans would do much to reduce the remaining inequities in mental health care.
Darrel Regier is executive director of the American Psychiatric Institute for Research and Evaluation of the American Psychiatric Association in Arlington, Virginia, and director of the association’s Division of Research. Health Affairs invited his response to the preceding paper by David Mechanic.
This article has been cited by other articles:
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Top
The Significance Of...
When A Diagnostic Theory...
