Differential Diagnosis In Psychiatry Research Paper

1. The Definition Of Diseases

In the seventeenth century all the diseases that were then recognized were defined by their clinical syndromes, that is by the symptoms the patient complained of, the clinical signs elicited by the physician on examination, and the evolution over time of these symptoms and signs. Over the last three or four hundred years, however, increasing understanding of the etiology of most of these diseases has resulted in their being redefined at some more fundamental level. Smallpox is now defined, not by its characteristic rash, high fever, and mortality, but by the presence of the variola virus. Cancer of the stomach is defined, not by the patient’s rapidly worsening indigestion and palpable abdominal mass, but by the histopathology (detailed microscopical structure) of the tumor. Thyrotoxicosis is defined, not by the patient’s goiter, weight loss, and bulging eyes, but by a raised blood level of the thyroid hormone thyroxin, and so on. As a result, the diagnosis of most diseases depends on the results of laboratory investigations. Although the physician usually has a shrewd idea which diagnoses are likely after examining the patient, the provisional clinical diagnosis needs to be confirmed by the laboratory. A diagnosis of smallpox is only provisional until the variola virus has been identified; a diagnosis of cancer of the stomach has to be confirmed by histological examination of the tumor; and a diagnosis of thyrotoxicosis has to be confirmed by demonstrating a raised blood level of thyroxin.

Because understanding of the etiology of most psychiatric disorders is still rather rudimentary they still have to be defined by their clinical syndromes. This means that diagnosis, and hence differential diagnosis, is determined by the patient’s symptoms (abnormal or distressing subjective experiences reported by the patient) and behavior (observed or reported by others), and the temporal evolution of these abnormalities. Apart from age and sex, information of other kinds, particularly from laboratory and radiological investigations, is rarely relevant, and even if it is relevant it is not decisive. The difference between psychiatry and the rest of medicine is not absolute, however. Several neurological disorders, such as migraine and torticollis, are still defined by their syndromes, and so too are some skin disorders, such as pityriasis rosea. Conversely, some psychiatric disorders, such as Alzheimer’s disease, vascular dementia, and Creutzfeldt-Jakob disease, are defined by their neuropathology. Their defining characteristics are comparable, in other words, with those of cancer of the stomach. But even here there is an important practical difference. The histopathology of a gastric tumor can be determined by taking a biopsy through a gastroscope, or at operation, whereas the histopathology of the brain can normally only be determined postmortem, and this means that the diagnosis on which treatment is founded still has to be based on the clinical syndrome.

At present there are only a few psychiatric disorders which are defined by the presence of a fundamental abnormality which can be identified or excluded during life. Down’s and Klinefelter’s syndromes are both defined by the presence of chromosomal abnormalities—the former by an additional chromosome 21 and the latter by an additional X chromosome. Huntington’s chorea and the Fragile X syndrome are both defined by abnormal genes at specific sites, and phenylketonuria (like Fragile X, an important cause of mental retardation) is defined by a specific biochemical abnormality. Eventually, of course, as understanding of their causation develops, most psychiatric disorders will either be redefined in terms of an underlying abnormality of some kind, or be replaced by new diagnostic concepts defined in this way. At that stage differential diagnosis in psychiatry will be no different from that in any other branch of medicine. Until then, however, problems of differential diagnosis have to be handled differently. In other branches of medicine diagnostic problems usually lead to more extensive laboratory and radiological investigations. In psychiatry this is rarely appropriate, and the best approach is usually to try to obtain a more detailed account of the development of the patient’s illness, often from other informants, and to monitor its course and response to treatment as closely as possible. Because most psychiatric disorders are defined by the patient’s symptoms and abnormal behavior, and changes in these over time, differential diagnosis has to be determined by information about these things, and uncertainties can only be resolved by obtaining more accurate or more detailed information, or by following the course of the illness.

2. The Adoption Of Operational Definitions

Before the 1970s psychiatric diagnoses were alarmingly imprecise. There was broad agreement on the core symptoms of the major syndromes, and this was reflected internationally in textbook descriptions. It was generally agreed, for example, that the characteristic symptoms of schizophrenia were thought disorder, flattening of affect and delusions, and hallucinations of particular kinds, and that some patients also exhibited other typically schizophrenic symptoms such as apathy and stereotypies. There was no agreement, however, on which combinations of these symptoms were adequate to establish a diagnosis of schizophrenia. Indeed, that question was almost never posed, and different schools of psychiatry and individual psychiatrists made different assumptions, and in practice trainee psychiatrists learnt how to diagnose syndromes such as schizophrenia and mania largely by modeling themselves on their teachers. As a result, several studies of the reliability of psychiatric diagnoses in the 1960s found the reliability of the diagnoses even of experienced psychiatrists to be distressingly low, and international studies like the US UK Diagnostic Project and the World Health Organization’s International Pilot Study of Schizophrenia revealed major between-country differences in the usage of key diagnostic terms such as schizophrenia and manic-depressive illness, and even between different cities or teaching centers in the same country.

As soon as this chaotic situation became apparent it was agreed by the clinical research community that operational definitions were required, at least for all the major syndromes. Essentially, an operational, or semantic, definition is a statement or set of statements which spells out precisely which combinations of symptoms and behaviors are adequate to establish the diagnosis in question. The first set of fifteen operational definitions for use in clinical research was published by the Department of Psychiatry in St Louis in the USA (Feighner et al. 1972) and other rival sets quickly followed. A few years later the American Psychiatric Association provided operational definitions for most of the 200 syndromes described in the third (1980) edition of its Diagnostic and Statistical Manual of Mental Disorders (DSM-III) (American Psychiatric Association 1980). It also demonstrated in the field trials conducted before the glossary was published that these operational definitions made it possible to achieve much higher levels of reliability than had previously been attained. Subsequently the World Health Organization provided similar operational definitions for most of the mental disorders listed in the tenth revision of the International Classification (ICD-10) (World Health Organization [WHO] 1993).

As a result of this general recognition, at least by clinical research workers and epidemiologists, of the need for operational definitions, and their incorporation into the two most widely used glossaries, DSMIII and its successors and ICD-10, diagnosis and differential diagnosis have been transformed. Reliability is now much higher; and the unrecognized differences in diagnostic criteria between different countries and schools, which had caused so much confusion previously, have either been eliminated or, if different operational definitions are being used, the differences are now overt and explicit. This in turn has transformed differential diagnosis.

3. The Consequences Of Definition By Clinical Syndrome

There are, even so, still a number of fundamental problems arising from the fact that most mental disorders are defined by their clinical syndromes. The most important is that there is no scientific way of deciding which of two or more alternative operational definitions of a single disorder is correct, or even which is the best. The choice between alternative clinical definitions of Down’s syndrome would be straightforward. One would pick whichever definition most accurately identified individuals with Trisomy 21 (an additional chromosome 21 in all the cells of the body), because Trisomy 21 is now the defining characteristic of Down’s syndrome. In the case of most psychiatric disorders, however, there is no comparable external criterion to appeal to, because the clinical syndrome is itself the defining characteristic. This is why alternative operational definitions of disorders such as schizophrenia and agoraphobia remain in widespread use, side by side, and it is commonplace for individual patients to meet one of these definitions but not the other. A patient with typical schizophrenic symptoms for the past six weeks, for example, might well meet the formal criteria for schizophrenia in ICD-10, but not the corresponding criteria in DSM-IV, because the latter require evidence of continuous signs of illness for at least six months. In practice this means that discussions about diagnosis need to take place in a context in which it is agreed which glossary (i.e., which set of operational definitions) is being used, and also with a clear recognition of the fact that the choice between alternative glossaries or sets of definitions is arbitrary.

4. Single Versus Multiple Diagnoses

In medicine as a whole an attempt is generally made to explain all the patient’s symptoms with a single diagnosis, and to try to avoid multiple diagnoses. The rationale for this is that several important diseases, including infections such as tuberculosis and syphilis, and collagen diseases such as polyarteritis nodosa, can affect almost every organ in the body, and the correct diagnosis is more likely to be made if the physician consciously strives to account for all the patient’s apparently unrelated problems with a single unifying explanation. In fact, of course, people often have several different things wrong with them simultaneously, particularly as they age. Old people are rather like old cars in this respect.

The same attempt to avoid multiple diagnoses is apparent in psychiatric practice, and the underlying rationale is similar. Patients are commonly given two diagnoses simultaneously if one is a lifelong state, such as mental retardation or a personality disorder, and the other an illness such as schizophrenia or a mood disorder with an identifiable onset and a distinctive course. A diagnosis of some form of substance misuse is also commonly combined with a wide range of other diagnoses. There is a deep-seated reluctance, however, to diagnose a neurotic or stress-related disorder in the presence of psychotic symptoms such as delusions and hallucinations, despite the fact that psychotic and neurotic symptoms commonly coexist, and there is a similar reluctance to diagnose any kind of nonorganic psychosis if there is evidence of brain disease. There is an even greater reluctance to diagnose two psychotic disorders simultaneously, or a mood disorder and a neurotic disorder simultaneously, despite the fact that depression and anxiety commonly coexist.

5. The Hierarchy Of Diagnoses

This reluctance to make multiple diagnoses achieves formal expression as a diagnostic hierarchy. Organic disorders come at the top of this hierarchy. This means that if the patient has clear evidence of a brain disease—a tumor, for example—it is legitimate for them to have a wide range of other psychiatric symptoms, psychotic or neurotic, without any other diagnosis being required. Next in the hierarchy come the so-called functional or nonorganic psychoses. If the criteria for one of these disorders—schizophrenia, for example—are met, the patient may also be depressed or anxious, or have phobic or obsessional symptoms, without any additional diagnosis being required. They must not, however, have evidence of overt brain disease. Neurotic and stress-related disorders come at the bottom of the hierarchy, which means that diagnoses at this level cannot normally be made if psychotic symptoms or evidence of brain disease are also present. It is no accident, either, that this is the sequence in which disorders are listed in contemporary glossaries, and also the sequence of the decisions required by the flowcharts of computer programs for generating diagnoses from clinical ratings.

The requirements of this hierarchy are only relevant if the symptoms in question all appear at much the same time. If two groups of symptoms have a quite different time course from one another, it may be both legitimate and appropriate to make two independent diagnoses. In the past the hierarchy used to be implied rather than explicit. Now it is built into the operational definitions of individual disorders. At the same time it has become more complicated, particularly in situations where schizophrenic and affective symptoms coexist, or follow one another in quick succession.

It is also still uncertain whether the hierarchy is simply an artifact, a device for reducing multiple diagnoses to a minimum, or whether it is a reflection of a fundamental characteristic of psychiatric symptomatology. It is certainly true that a high proportion of people with psychiatric symptoms, but not all, move up and down the hierarchy as they relapse and remit. Phobic and obsessional symptoms almost invariably become more widespread and severe if depressive symptoms develop, for example, and the development of hallucinations and delusions is usually accompanied by affective and neurotic symptoms which do not wane until after the hallucinations and delusions have disappeared.

6. The Boundary Problem

There is an old axiom that the art of classification consists in ‘carving nature at the joints.’ This is an elegant way of saying that a good, or a valid, nosology places the boundaries between adjacent disorders, and between these disorders and health, where there are genuine discontinuities, either in symptomatology or in etiology. If the defining characteristic of the disorder is a fundamental biological abnormality of some kind it may be self-evident that a discontinuity is involved. An additional chromosome or an abnormal gene, for example, clearly involve qualitative deviations from normality. If, however, the defining characteristic of the disorder is still its syndrome, or a quantitative deviation of some kind, such as a raised blood pressure or a raised blood level of thyroxin, it is not self-evident that there is a valid boundary between the syndrome in question and normality, or between this syndrome and others. Indeed, the fact that there are several rival operational definitions for many mental disorders, each of them encompassing slightly different populations of patients, strongly suggests that it is not obvious where the boundaries of those syndromes should be drawn.

Ideally, a point of rarity needs to be demonstrated between each syndrome and its neighbors (Kendell 1975). For example, it is widely assumed that schizophrenic and affective psychoses are distinct conditions. Indeed, the distinction Kraepelin drew at the end of the nineteenth century between what he called dementia praecox and manic-depressive insanity is still regarded as a fundamental feature of our contemporary classification of mental disorders. The fact that patients with a mixture of schizophrenic and affective symptoms—so called schizo-affective patients—are quite common does not in itself prove that there is no genuine boundary between the two syndromes, provided that these mixed forms are less common than the pure forms. The ‘greys’ must be less common than the ‘blacks’ and the ‘whites’; or to express the issue graphically, the distribution of symptomatology along an axis extending from typical schizophrenia through schizo-affective states to typical affective psychoses must be bimodal rather than unimodal (see Fig. 1). A number of attempts have been made to address this issue using discriminant function analysis and the results are summarized below.

An alternative to demonstrating discontinuities in the distribution of symptomatology is to identify a discontinuity in the relationship between symptomatology and some other important variable, such as prognosis or response to treatment. For example, it is well established that, on average, the outcome of affective psychoses is much better that that of schizophrenic illnesses, both in the short and the long term. The key issue, however, is not whether the two disorders have a significantly different outcome, but whether, when the variation in symptomatology is portrayed as a linear dimension (as described above) there is a discontinuity at some point in the relationship between symptomatology and outcome, or whether there is simply a linear improvement in outcome as the symptomatology becomes increasingly more affective (see Fig. 2).

To date, most attempts to demonstrate valid boundaries by these means have been concerned with the boundaries between what used to be called endogenous and reactive depressions, between schizophrenic and affective psychoses, and between major depressive disorder and normality. The attempts to demonstrate a valid boundary between endogenous and reactive depressions were unsuccessful; when representative populations of patients were studied the distribution of individual patients along the discriminant function was consistently unimodal rather than bimodal. Partly for this reason, contemporary glossaries such as ICD-10 (WHO 1993) and DSM-IV (American Psychiatric Association 1994) no longer distinguish between these two types of depressive illness. Similar attempts to demonstrate a valid boundary between schizophrenic and affective psychoses, either in symptomatology or in the relationship between symptomatology and outcome, were also unsuccessful, but a bimodal distribution of scores was subsequently demonstrated on a linear function discriminating between schizophrenia and all other psychiatric disorders combined. More recently, an attempt to identify a valid boundary between major depression and normality by demonstrating, in a population of 2,000 twins, a discontinuity in the relationship between depressive symptomatology and either outcome or concordance between twin pairs ended in failure (Kendler and Gardner 1998).

7. The Consequences Of Failure To Validate Boundaries

None of these studies proves that there are no valid boundaries between the syndromes in question. For one thing, it is always possible that more extensive data, or a different choice of criterion, might have provided a different result. Cumulatively, however, the failure of most of these attempts to demonstrate valid boundaries between syndromes emphasizes that it is unsafe to assume that there are valid boundaries there to be found. We must take seriously the possibility that we are not yet ‘carving nature at the joints,’ or even that there are no ‘joints’ there, and that we are trying to impose an artificial set of discrete categories on what is in reality continuous variation.

This state of affairs holds several important implications for the differential diagnosis of psychiatric disorders. The first is the importance of recognizing that most of the categories of disorder recognized in contemporary nosologies, even those such as schizophrenia with pedigrees dating back to the nineteenth century, have not yet been adequately validated, and that their formal definitions are simply agreed conventions. The second is a consequence of the first. All decisions about diagnosis, and all discussions about differential diagnosis, should take place in a setting in which it is agreed which set of diagnostic categories and corresponding set of operational definitions (in effect, which glossary) are being used, and with a clear understanding that both the categories and the definitions are arbitrary and inadequately validated, and therefore likely to be modified as understanding of the underlying etiology accumulates.

The third implication is a consequence of the likelihood that many existing boundaries, either be-tween syndromes or between psychiatric disorder and health, are not situated at points of rarity in the distribution of symptomatology. To the extent that this is the case, relatively minor changes in the operational definition of a syndrome, which produce corresponding changes in the position of its boundaries, may result in quite large changes in the numbers of people, or patients, who fall within those boundaries. This is almost certainly why small changes in the definition of common disorders such as major depression or generalized anxiety—particularly changes in the number of typical symptoms required, or in their duration—have been associated with major changes in the recorded prevalence of these disorders in community surveys.

Finally, if it is indeed the case that there are no genuine boundaries within large tracts of the territory of psychiatric disorders, and we are imposing an artificial set of discrete categories on what is in reality continuous variation, psychiatrists will have to consider seriously the possibility of replacing their traditional categorical classification of disorders with a dimensional classification. This would never be done lightly, if only because the rest of medicine has always been committed to and content with categorical classifications of its disorders. If a dimensional format ever were to be adopted for mental disorders, however, the consequences for differential diagnosis would be profound. Disputes about which disorder or combination of disorders an individual was suffering from would be replaced by discussions about the precise locus in a multidimensional space at which that individual should be located. These discussions would probably also be calmer than the previous disputes, because in most cases it would be clear that the alternative loci under discussion were all quite close to one another, and not separated by dividing lines implying major differences in treatment or prognosis.

8. Practical Issues In Differential Diagnosis

As in other branches of medicine, some differential diagnoses are more critical than others. The most important functions of a diagnosis are to determine treatment and to predict outcome, and the most important differential diagnoses are those involving a disorder requiring urgent and specific treatment. All psychiatrists are keenly aware of the importance of diagnosing, or at least suspecting, cerebral tumors, and other brain diseases such as viral encephalitis and subdural hematoma, which may also present with psychiatric symptoms, as quickly as possible. Because these conditions may all present with depressive symptoms, with apathy or other behavioral changes, or even with psychotic symptoms, and because they all require urgent treatment, they enter into the differential diagnosis of a wide range of psychiatric disorders. Many reputations for diagnostic acumen are, rightly, based on successful detection of relatively uncommon brain diseases of this kind, or of metabolic disorders such as thyrotoxicosis and Cushing’s syndrome, which underlie some anxiety states and depressions. It is also important to detect brain diseases such as Creutzfeldt–Jakob disease, which sometimes present with psychiatric symptoms even though there is not yet any treatment capable of altering the course of the disease, because of the implications for prognosis. If someone has an illness which is inevitably going to result in permanent disability or death, the sooner that is established the better, both for the patient and the family.

There are other circumstances in which the critical diagnosis in the differential diagnosis between brain disease and psychiatric disorder is the latter rather than the former. Dementias are common in old age and most are associated with progressive decline to dependency and death. Sometimes, however, depressive illnesses present with such widespread evidence of intellectual impairment that they closely resemble a dementia. Because depressions are eminently treatable it is vital that these ‘depressive pseudodementias’ are quickly recognized and treated, and depression is therefore an important part of the differential diagnosis of dementia. Similar considerations apply to rare curable dementias, such as those associated with a deficiency of thyroid hormone or vitamin B12.

Traditionally, much time and intellectual effort have been devoted by psychiatrists to distinguishing accurately between schizophrenic and affective psychoses. To some extent this is appropriate. The two disorders require different treatments and there are major differences in long-term prognosis between them. Some disagreements are due simply to undisclosed differences in diagnostic criteria which would be resolved by the adoption of an agreed operational definition. However, a more fundamental conceptual issue may also be involved. Although there are many instances, including the differential diagnosis between schizophrenic and affective psychoses, where no valid boundary has yet been demonstrated between the two syndromes, psychiatrists often still behave as though they were discussing the distinction between two ‘disease entities.’ In reality, if the patient’s symptoms are ‘schizoaffective,’ it may not matter much which side of an arbitrary boundary they are deemed to lie, because their therapeutic needs and their prognosis may both be genuinely intermediate between those of the two classical syndromes.

To diagnose schizophrenia is always a grave decision. Although some people develop schizophrenic illnesses which resolve completely, even without treatment, and never recur, for many more a schizophrenic episode is the prelude to a life of invalidism and the death knell of hopes and ambitions. It is understandable, therefore, that many psychiatrists are reluctant to diagnose schizophrenia, particularly in an adolescent, unless the evidence is incontrovertible. If some less doom-laden diagnosis, such as depression, is plausible they will opt for that, and hope that time will justify their decision. Scandinavian psychiatrists sometimes make a diagnosis of psychogenic psychosis with similar motives. Recently, however, psychiatrists have become increasingly convinced by the evidence that early treatment with neuroleptics improves the long-term outcome of schizophrenic illnesses and reduces the risk of a ‘defect state.’ Although this evidence is not yet buttressed by random allocation trials it is sufficiently strong to swing the balance of advantage in patients with ambiguous symptomatology in favor of early diagnosis. A similar mix of social and therapeutic considerations has an influence on differential diagnosis in other areas too.

Other important issues of differential diagnosis arise when patients with psychiatric disorders present to other medical disciplines, as they commonly do. Depressive illnesses are extremely common at all ages and depressed patients may present with a wide variety of complaints, particularly if they are reluctant to accept that they may have a psychiatric disorder. They may complain of ‘feeling off color’, of loss of appetite or weight, of difficulty sleeping, or of ill-defined pain almost anywhere in the body. As a result, the correct diagnosis is often missed unless questions are asked about characteristic depressive symptoms such as difficulty concentrating or loss of the ability to enjoy things. Phobic anxiety is another common disorder which often presents with somatic symptoms such as palpitations, or recurring bouts of upper abdominal discomfort or difficulty breathing. Unless specific questions are asked about the environmental circumstances in which these symptoms develop, their phobic nature may easily be missed and the patient may be subjected to a long series of irrelevant investigations. In both cases the fundamental problem is that the doctor concerned does not realize, or remember, that a depressive illness or phobic anxiety should be important elements in the differential diagnosis of the patient’s presenting complaints, even though these may be entirely somatic.

Bibliography:

1. American Psychiatric Association 1980 Diagnostic and Statistical Manual of Mental Disorders, 3rd edn. (DSM-III), American Psychiatric Association, Washington, DC
2. American Psychiatric Association 1994 Diagnostic and Statistical Manual of Mental Disorders, 4th edn. (DSM-IV), American Psychiatric Association, Washington, DC
3. Feighner J P, Robins E, Guze S B, Woodruff R A, Winokur G, Munoz R 1972 Diagnostic criteria for use in psychiatric research. Archives of General Psychiatry 26: 57–63
4. Kendell R E 1975 The Role of Diagnosis in Psychiatry. Blackwell Scientific, Oxford, UK
5. Kendler K S, Gardner C O 1998 Boundaries of major depression: An evaluation of DSM-IV criteria. American Journal of Psychiatry 155: 172–7
6. World Health Organization 1993 The ICD-10 Classification of Mental and Behavioural Disorders: Diagnostic Criteria for Research. World Health Organization, Geneva, Switzerland