|The Overlap of DSM IV Syndromes: Potential Implications for the Practice of Polypsychopharmacology, Psychiatric Drug Development, and the Human Genome Project
SHELDON H. PRESKORN, MD
BRYAN BAKER, RN, MSM, CCRC
Journal of Psychiatric Practice, May 2002, 170-177
Since its inception, the focus of this column has been to discuss the variables that influence the ability to use medications to treat patients safely and effectively. Topics have included the pharmacology of existing medications, the drug development process leading to the production of new medications, and genetic and acquired factors underlying interindividual differences in response to the same dose of the same medication. Major series have been devoted to polypharmacy and to the human genome project and its implications for drug development. Polypharmacy can be viewed as a state or acquired difference among patients (i.e., state dependent changes in their internal environment) that can alter their response to a given dose of a given drug by interacting with it either pharmacodynamically or pharmacokinetically. The focus of the human genome project is in part to determine genetic or trait variables that lead to interindividual differences in drug response.
Age, genetic differences, and internal environment, including polypharmacy, are state dependent factors that fall under variable 3 in equation 1:
Effect = affinity for x drug x biological variance:
site of action concentration (age, disease, genetics,
Disease is another state dependent factor that can alter drug response. Recall that the purpose of all drugs (except anti infectives) is to alter human biology. For that reason, their presence in the body changes the body (i.e., they are a "state" variable) and hence can change the patient's response to other drugs (i.e., a drug drug interaction). Also recall that drugs are developed to treat disease and therefore the response to a drug is dependent on having a disease that is responsive to the drug's mechanism of action. That is the reason why drugs have labeled indications and why inclusion criteria in clinical trials specify the disease being treated.
This column will address some conceptual issues relevant to psychiatric diagnoses as they relate both to drug development and to clinical practice. It will also discuss how these issues may be relevant to the human genome project and its search for the genetic basis for specific psychiatric disorders and to the apparently growing tendency to use polypsychopharmacology in clinical psychiatric practice.1
Role of Diagnoses in Medicine
Diagnosis, the process of naming and classifying diseases, serves many purposes in medicine. It simplifies communication about patients with a common condition and permits prognostication, treatment planning, and research. Diagnosis is to the practice of medicine what a hypothesis is to research. A hypothesis is the researcher's prediction of what will happen under specific experimental conditions. Its value resides in its ability to predict the outcome of the study. The same is true of diagnosis in clinical medicine. A diagnosis is only as good as its ability to predict the natural course of the patient's condition and its response to specific interventions. A positive research study supports and strengthens the hypothesis. A negative study requires reconsideration of the hypothesis. The same is true for diagnosis. To the extent that the diagnosis correctly predicts what will occur either naturally or in response to treatment, it is useful. When the outcome is not consistent with the diagnosis, the clinician should reconsider the diagnosis.
The goal of diagnosis is to group together patients suffering from the same disease. Of course, any diagnostic system is only as good as the knowledge of the disease at the time the system is developed. Hence, diagnostic systems are constantly evolving as new knowledge becomes available. The development of diagnoses is an iterative process. As diagnoses become refined and more narrowly define the disease, patients with that disease become more homogeneous with regard to their pathophysiology and pathoetiology. That homogeneity improves the signal to noise ratio in both treatment studies and more basic studies aimed at further elucidating the mechanisms underlying the disease. The knowledge resulting from those studies in turn provides more knowledge, which in turn should lead to further refinement of the diagnostic criteria. The process continues in this fashion until the pathophysiology and pathoetiology of the disease have been fully elicited.
|Figure 1 - Diagnostic criteria pyramid |
All diagnoses can be grouped into four hierarchical levels of understanding as illustrated in Figure 1. The least sophisticated level is symptomatic diagnoses (e.g., headache or psychosis). Patients generally present with symptomatic diagnoses (e.g., "I have a headache").
Syndromic diagnosis is the next level. Such diagnoses are based on the observation that a group of patients are presenting with the same cluster of symptoms and signs. Such naturally occurring clusters suggest that a common disease process may explain the clustering. A relatively recent example was acquired immunodeficiency syndrome (AIDS). In this case, an astute clinician noticed that a group of patients were presenting with a cluster of what had previously been relatively rare signs, symptoms, and diseases, including opportunistic infections and malignancies such as Kaposi's sarcoma.2,3 The definition of this syndrome facilitated research, which first led to an improved understanding of the pathophysiology underlying the syndrome (i.e., a progressive loss of specific types of lymphocytes) and then to an understanding of the pathoetiology (i.e., infection with the human immunodeficiency virus, HIV). Pathophysiology and pathoetiology are the final two levels of diagnostic sophistication.
The goal of the clinician and the researcher is to achieve the highest level of diagnostic sophistication possible, since higher levels of understanding permit better prognostication and improve the ability to alter the course of the disease. For example, the identification of HIV as the causative agent in AIDS permitted the development of practices that reduce the risk of acquiring the disease and drugs that prevent the replication of the virus once acquired, thus arresting the progression of the disease. Now work is underway to develop vaccines to further prevent the spread of the disease.
Diagnosis in Psychiatry
Diagnosis serves the same functions and follows the same pattern in psychiatry as in the rest of medicine. Most psychiatric diagnoses are at the level of syndromes and are codified in the United States in the fourth edition of the Diagnostic and Statistical Manual of Mental Disorders (DSM-IV)4 of the American Psychiatric Association. These syndromes are important because they are the target of drug development and thus have implications for what prescribers use and why. These syndromes are also the basis by which patients are grouped into "disease" clusters and then examined for potential genetic traits underlying those "diseases."
Psychiatric diagnoses for the most part have not progressed beyond the syndromic level largely because of the difficulty of studying and relating disturbance in brain function to behavior. The brain is by far the most complex and challenging organ to study. Due to the
complexity of brain circuits and the fact that brain function is regionally based, multiple different pathoetiologies may present in a clinically similar manner.
To illustrate this point, consider hemiplegia as a clinical syndrome. Phenomenologically, it might look the same whether it was due to an ischemic infarct, an intracerebral hemorrhage, a multiple sclerotic plaque, a benign tumor, a malignant tumor, a gumma, or an abscess. While the signs and symptoms might look the same, the underlying diseases, their course, prognosis, and appropriate treatment can obviously be quite different. The same could well be true for patients grouped together with the same psychiatric syndrome. The signs and symptoms may well reflect a disturbance in a given region of brain function, as is the case with hemiplegia, but the etiologies of the dysfunction in this region could be quite different. This issue may partly account for the relatively poor signal to noise ratio found in trials of antidepressants, as discussed in earlier columns,5,6 as well as in trials of medications for other psychiatric syndromes.
Researchers need to keep this issue in mind when considering drug development and when attempting to do studies to test for genetic predisposition to specific psychiatric disorders. There may be significant problems with phenocopies, which can confound detection of an association. In addition to these problems, the overlap in symptoms and signs in psychiatric syndromes as currently defined in DSM-IV also poses problems. Defining the proper boundaries of syndromes has generally been problematic in medicine. While psychiatric diagnoses can be traced back to Hippocrates, a concerted effort to codify these syndromes is relatively recent, with most of the effort occurring in the last 30 40 years. It may be helpful to briefly review this recent history.
Psychoanalysis dominated American psychiatry for much of the last century and even today has some residual influence on the field. Diagnosis was not a principal concern of psychoanalysts. In fact, they viewed it as somewhat irrelevant because they believed that they knew what caused psychiatric illness and had the treatment. According to their theory, psychiatric illness resulted from a breakdown of psychic defense mechanisms and from unresolved psychosexual developmental issues. In their view, the nature and the severity of psychopathology was a function of how regressed the patient was in terms of the breakdown of psychic defense mechanisms. Given their confidence in their theories and their treatment, it is easy to see why the patient's psychiatric diagnosis was not a concern for psychoanalysts.
For this reason, the earliest versions of the DSM published by the American Psychiatric Association (DSM-I7 published in 1952 and DSM-II8 published in 1968) were notably smaller than either DSM-III9 or IV. They gave case descriptions rather than criteria for psychiatric diagnoses. Clinicians would read a thumbnail vignette of the "typical" patient with a specific disorder and decide whether that description came closest to fitting their patient out of all the descriptions contained in the manual. Due to the absence of criteria for making these decisions, the same patient could receive a different diagnosis from two different experienced clinicians based on what facets of the description and the patient's mental status and history resonated with one clinician compared with another. There were no rules to help guide decisions but instead only the interpretation of each vignette and how well it matched with the history and presentation of the clinician's patient.
A tradition of active interest in psychiatric diagnosis dates back to the mid 1800s in both England and Europe. Psychiatrists in these countries were active in trying to study the possible biological basis for psychiatric conditions. For example, Kraepelin and his contemporaries conducted both pathological and physiological studies to search for possible biological correlates to various psychiatric syndromes.10,11
Interest in psychiatric diagnosis in the United States began to assert itself in earnest in the second half of the 20th century. That was in part the result of the chance discovery of the first effective medications for psychiatric syndromes such as lithium for bipolar disorder, chlorpromazine for schizophrenia, imipramine and iproniazid for major depression, and diazepam for anxiety disorders. These discoveries stimulated interest in finding new and better medications, which led to the need for clinical trials, which in turn led to development of inclusion and exclusion criteria defining the conditions (i.e., psychiatric syndromes) to be studied. At the same time, the National Institute of Mental Health began funding studies to examine the epidemiology and course of psychiatric illnesses.
It became apparent that these lines of research could not be effectively conducted if researchers at different institutions did not have a common way of determining whether a given group of patients had the same psychiatric disorder. Without an agreed upon mechanism, one could not be certain whether a group of patients with a given DSM-I or II diagnosis in one research study was the same as a group of patients in another study.
One of the early and most successful attempts to develop standardized criteria for codifying psychiatric syndromes was undertaken by psychiatric researchers at Washington University in the mid 1960s. Their criteria were published in a landmark paper in 1972.12 Following their publication, these criteria, generally referred to as either the Washington University criteria or the Feighner criteria based on the last name of the first author of the publication, became quite popular. The diagnoses were based on a set of objective criteria that the patient had to meet to receive the diagnosis.
The Washington University criteria were developed by a small group of researchers who wanted to define discrete conditions. Under the Washington University system, 25% or more of patients who presented for treatment did not fit into one of the discrete psychiatric syndromes defined by the Washington University criteria principally because they had too many rather than too few symptoms. These patients were termed undiagnosable and were excluded from study as being too heterogeneous. After all, the goal was to define diagnostic groups that were as homogeneous as possible so that they could be studied in terms of the natural history and familial nature of their condition, the potential biology of the illness, and its response to specific treatment interventions.
The next iteration of this criteria driven approach to psychiatric diagnosis was the publication in 1974 of the Research Diagnostic Criteria (RDC).13,14,15 There were approximately twice as many diagnoses in the RDC as in the Washington University criteria. The RDC were shortly followed by the DSM-III, published in 1980, which expanded the number of diagnoses to more than 200. The revised version of DSM-III (DSM-III R)16 was published in 1987, followed by the DSM-IV in 1994. DSM-V is in the works.
Overlap in Psychiatric Syndromes
There were multiple reasons for the explosion in the number of psychiatric syndromes that occurred with the publication of DSM-III. In part, it stemmed from new information generated by the studies funded by NIMH and other government agencies as well as by clinical trials that were being conducted by pharmaceutical companies searching for new medications. DSM-III was also developed by a number of committees that included a much more heterogeneous group of psychiatrists than those who developed the Washington University and the RDC criteria. DSM-III also had purposes that went beyond research and were important for practitioners. While researchers could choose to call some patients undiagnosable and leave them out of studies because they did not fit into more narrowly defined syndromes, that was not a palatable solution to the practitioner who did not have the luxury of turning such patients away.
Nevertheless, the explosion in the number of psychiatric syndromes may have produced some blurring of the boundaries that need to be considered when conducting research either in an attempt to develop new drugs or to understand the possible genetic basis of at least some psychiatric syndromes. It may also be relevant to the issue of polypsychopharmacology To explore this issue, we examined the diagnostic and associated features listed in DSM-IV for bipolar disorder type I and then compared those features with those for other psychiatric syndromes, beginning with the next closest diagnoses, bipolar II disorder and cyclothymic disorder. We then looked at major depressive disorder and dysthymic disorder, followed by schizoaffective disorder, two specific anxiety disorders, generalized anxiety disorder and posttraumatic stress disorder, and finally two personality disorders, borderline personality disorder and antisocial personality disorder. We focused on those diagnostic and associated features that were shared by several of these psychiatric diagnoses. The results are shown in Table 1.
|Table 1. Shared diagnostic and associated features listed in the DSM-IV for selected psychiatric syndromes |
Table 1 lists 10 diagnostic features and 1 associated feature found in the DSM-IV section on schizoaffective disorder. A mixed episode of bipolar I disorder shares the same 10 diagnostic features and 1 associated feature. In addition, there is considerable overlap in the diagnostic and associated features of the various mood disorders, including bipolar II disorder, cyclothymic disorder, major depressive disorder, and dysthymic disorder. There is also considerable overlap with nonaffective disorders such as generalized anxiety disorder and posttraumatic stress disorder.
Perhaps even more surprising is the substantial overlap in diagnostic and associated features among borderline personality disorder, antisocial personality disorder, schizoaffective disorder, and a mixed episode of bipolar disorder. Borderline personality disorder has 8 of the same diagnostic features as schizoaffective disorder. In addition, DSM-IV lists substance abuse as a diagnostic feature for borderline personality disorder and an associated feature of schizoaffective disorder and of a mixed episode of bipolar I disorder. Antisocial personality disorder shares 5 diagnostic features with schizoaffective disorder as well as the associated feature of substance abuse. In addition, 2 associated features are listed in DSM-IV for antisocial personality disorder that are also listed as diagnostic features for schizoaffective disorder.
Given this considerable overlap in signs and symptoms, it should be no surprise that many patients have features and may meet criteria for more than one psychiatric syndrome. Depending on the orientation of the clinician or researcher, they may be given multiple "comorbid" diagnoses, even though the same signs and symptoms are the basis for these different diagnoses. Alternatively, patients with the same condition may receive different diagnoses based on which diagnosis is favored by a specific clinician. Thus, the same patient might be diagnosed as having schizoaffective disorder by one clinician and borderline personality disorder by another because of the overlap of symptoms; or the same patient might be diagnosed as having both schizoaffective disorder and borderline personality disorder. This situation ironically recapitulates the problem the Washington University criteria were developed to correct.
Implications for the Human Genome Project and Psychiatric Diagnoses
in the past and for the foreseeable future, a discrete phenotype is important in enhancing the ability to establish the genetic basis for disease. An example of a highly discrete phenotype would be Huntington's chorea, which is quite distinct from virtually any other diagnosis. Of course, the fact that the condition follows autosomal dominant Mendelian genetics is also helpful. However, discrete syndromes that are clearly distinct from other conditions have also been helpful in less genetically straightforward conditions such as hypertension.
Clearly, the situation with Huntington's chorea is quite different from the complexities involved with most psychiatric syndromes. There are quite real and unavoidable difficulties in attempting to draw boundaries around psychiatric syndromes without the benefit of a discrete phenotype, as is found in Huntington's disease. However, that acknowledgement does not obviate the potential problems created by the overlap in different psychiatric syndromes reflected in Table 1.
The substantial overlap between the diagnostic and associated features of otherwise disparate conditions such as schizoaffective and borderline personality disorders sets the stage for patients with these conditions being included in studies whose goal is the elucidation of the possible genetic mechanisms underlying the pathoetiology of specific psychiatric syndromes. Of course, such an approach would not be detrimental if the traits these patients share have the same genetic basis, even though the syndromes themselves do not have the same pathogenesis. However, it seems more likely that studies will need to define more narrow subsets with strong family histories and avoid "comorbid" patients to enhance the signal-to-noise ratio.
Implications for Clinical Trials
The overlap in diagnostic and associated features also poses problems for the clinical trial researcher. First, it can thwart at least one of the goals of having inclusion and exclusion criteria, which is to make the study population more homogeneous with regard to pathophysiology and thus enhance the signal-to-noise ratio and increase the chance of a positive or at least definitive study. Second, it can pose problems for the researcher who is attempting to qualify a patient for a study because the patient may have received multiple different diagnoses at different times as a result of this diagnostic overlap. Some of these earlier diagnoses may be exclusionary criteria for the study at hand. For example, the researcher may need to obtain copies of previous inpatient hospitalization or outpatient records to see why a patient received a diagnosis of bipolar disorder when the patient currently appears to have a borderline personality disorder. Perhaps the previous clinician favored the diagnosis of bipolar disorder for any patient who presented with affective instability. Alternatively, a review of the record may reveal a well-defined and unequivocal manic episode. Unfortunately, the records are frequently not clear, which again may reflect the diagnostic uncertainty created by the multiple diagnostic options contained in DSM-IV and their overlap.
It has been claimed that 50% of the patients enrolled in recent trials of mood stabilizers met criteria for both bipolar disorder and borderline personality disorder. The overlap in symptoms makes that plausible. The question is whether they have the same pathophysiology or simply share some superficial similarities in terms of symptoms and/or signs.
Implications for Polypsychopharmcology
This overlap may also in part contribute to the apparent increase in the practice of polypsychopharmacology.1 Given this overlap, clinicians have not surprisingly begun to use medications that have a labeled indication for one disorder to treat patients who ostensibly have another diagnosis but share diagnostic and associated features with the indicated disorder (e.g., the use of mood stabilizers to treat borderline and antisocial personality disorders).
Clearly, there are a number of examples of situations in which psychotropic drugs work in quite different conditions. For example, anticonvulsants have been found to be effective for bipolar disorder. Antipsychotics such as haloperidol work in schizophrenia, psychotic mania, psychotic major depression, and even Huntington's chorea. In such instances, the drugs are most likely working on overlaps in the pathophysiology of these disorders, which nevertheless may have quite different pathoetiologies (e.g., Huntington's chorea versus psychotic major depression).
However, there is a potential danger in the overly exuberant and uncritical adoption of this approach. A prescriber may begin adding medications on the basis of symptoms rather than on the basis of having a specific syndrome. For example, a patient with borderline personality disorder may be treated with an anticonvulsant or lithium for mood lability, an antidepressant for depressive symptoms, an anxiolytic for anxiety symptoms, and an antipsychotic for brief psychotic symptoms. These medications may be added all at once or one shortly after the other before there has been an opportunity to determine what effect a single medication will produce. Several examples of such polypsychopharmacology have been described in earlier columns." 17,18,19
To minimize excessive polypsychopharmacology, a disciplined approach can be taken in which a medication directed at the underlying disorder or a broader range of signs and symptoms can be given an adequate trial before resorting to the addition of more medications. An example would be to start with a mood stabilizer and resist the temptation to immediately add a concomitant antidepressant and anxiolytic. Another approach is to stop medications that have not been successful rather than simply continuing to add medications. 17,20
Considerations for Future Nosological Revisions
DSM-IV diagnoses are based principally on signs and symptoms but there is reason to suspect that natural history and course of the illness may be as important or even more important. Reflect back on the earlier example of the syndrome of hemiplegia. The nature of the prodrome and the context in which that syndrome develops is more germane to the underlying disease process than are the signs and symptoms. A hemiplegia due to a hemorrhagic stroke has a much different onset and course than does a hemiplegia due to a benign tumor or an abscess.
In a similar manner, the original cardinal features of bipolar I disorder included the marked severity of the symptoms, the apparent disconnect between the mood symptoms and the situation of the patient (grandiosity and gregariousness when reality testing would dictate another emotional state), the classic good premorbid state of the patient prior to the onset of the first affective episode, and the return to that premorbid state after an episode had passed, at least in the early stages of the illness.10 Once a manic or depressive episode begins, the symptoms are reasonably stable for a sustained period, although the severity may fluctuate somewhat. In keeping with the concept of an episode, there is a clustering of symptoms involving different brain systems, including mood, sleep, sex drive, concentration/ attention, and reality testing.
In contrast, a personality disorder by definition begins early in life and is chronic. In the case of borderline personality disorder, the mood symptoms typically have a rather close temporal relationship to an environmental stimulus and are usually quite transient. They may also be self serving. There is not the clustering of symptoms involving different brain systems such as sleep disturbance or concentration/attention.
Of course, arguments could be mounted against the usefulness of each of the distinctions offered above. Certainly, the concept of rapid cycling can be invoked to explain the short-lived nature of the affective lability in borderline personality disorder. The concept of lower grade severity could be used to explain the absence of grandiose delusions. There are certainly examples of illnesses that display a continuum from severe pathology to virtually normal. Nevertheless, it has generally been most fruitful from a research standpoint to study the most discrete cases rather than mild cases. In fact, the elucidation of the pathophysiology or pathoetiology of the disease in discrete cases has generally made it possible to define the more mild forms of the illness which, in their earliest state, may in fact be difficult to separate from the apparently non-ill population.
The symptom overlap in psychiatric diagnoses demonstrated in Table 1 can produce diagnostic confusion, create high rates of "comorbid" psychiatric diagnoses, lead to an increased concomitant use of multiple psychiatric medications, decrease the efficiency of clinical trials in psychiatry, and impede the progress of studies into the potential genetic bases of psychiatric illnesses. In addition to the psychiatric researcher, the careful clinician must take these issues into consideration as he or she evaluates a patient and use both discipline and caution to remain rigorous diagnosticians and avoid the temptation to succumb to the noncritical use of multiple psychiatric diagnoses.
- Frye MA, Ketter TA, Leverich GS, et al, The increasing use of polypharmacotherapy for refractory mood disorders: 22 years of study. J Clin Psychiatry 2000;61:9-15
- Zakowski P, Fligiel S, Berlin GW, et al, Disseminated Mycobacterium avium intracellulare infection in homosexual men dying of acquired immunodeficiency. JAMA 1982;248: 2980-2
- Taff ML, Siegal FP, Geller SA, Outbreak of an acquired immunodeficiency syndrome associated with opportunistic infections and Kaposis sarcoma in male homosexuals: An epidemic with forensic implications. Am J Forensic Med Pathol 1982;3:259-64
- American Psychiatric Association, Diagnostic and statistical manual of mental disorders, 4th edition. Washington, DC: American Psychiatric Association; 1994
- Preskorn SH, A dangerous idea. J Pract Psychiatry Behav Health 1996;2:231-4
- Preskorn SH, Finding the signal through the noise: The use of surrogate markers. J Pract Psychiatry Behav Health 1999;5:104-109
- 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, tad 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
- Kraepelin E, Manic depressive insanity and paranoia. New York: Arno Press; 1976 (reprint of 1921 edition)
- Kraepelin E, Dementia praecox and paraphrenia. Translated by Barclay RM. Robertson GM, ed. Huntington. NY Krieger; 1971
- Feighner JP, Robins E, Guze SB, et al, Diagnostic criteria for use in psychiatric research. Arch Gen Psychiatry 1972;26: 573
- Spitzer RL, Endicott J, Robbing E, Research diagnostic criteria. New York: Biometrics Research, New York Department of Mental Hygiene; 1974
- Spitzer RL, Endicott J, Robbing E, Clinical criteria for psychiatric diagnosis and DSM-III. Am J Psychiatry 1975;132: 1187 92
- Spitzer RL, Endicott J, Robbins E, Research diagnostic criteria. Arch Gen Psychiatry 1978;35:773 82
- American Psychiatric Association, Diagnostic and statistical manual of mental disorders, 3rd edition, revised. Washington, DC: American Psychiatric Association; 1987
- Preskorn SH, Do you believe in magic? J Pract Psychiatry Behav Health 1997;3:99-103
- Preskorn SH, I dont see em. J Prac Psych Behav Hlth. 1997; 3:302-307
- Preskorn SH, A message from Titanic. J Prac Psych Behav Hlth. 1998;4:236-242
- Preskorn SH, The slippery slide. J Pract Psychiatry Behav Health 1999;5:50-5
- Goodwin DW, Preskorn SH. DSM-III and pharmacotherapy. In: Turner S, Hersen M. Adult psychopathology: A social work perspective. New York: Wiley; 1982:453 65.
- Preskorn SH. The future and psychopharmacology: Potentials and needs. Psychiatric Annals 1990;20:6253.
- Preskorn SH. Beyond DSM-IV. What is the cart and what is the horse? Psychiatric Annals 1995;25:53 62.
- Preskorn SH, Fast GA. Beyond signs and symptoms: The case against a mixed anxiety and depression category. J Clin Psychiatry 1993;54:24 32.
- Preskorn SH, Goodwin WG, Erickson HM. DSM-III R and pharmacotherapy. In: Hersen M, Turner SM, eds. Adult psychopathology and diagnosis, 2nd edition. New York: Wiley; 1991:482 99.