* Ian R. McWhinney, M.D., FRCGP, FCFP, FRCP
^t Ronald M. Epstein, M.D.
⁺ Tom R. Freeman, M.D., CCFP
January, 1997

*Centre for Studies in Family Medicine, The University of Western Ontario, London, Ontario, Canada N6A 5C1
t Highland Hospital Primary Care Institute, and Departments of Family Medicine and Psychiatry, University of Rochester, Rochester, N.Y. U.S.A. 14620
Dr. Epstein is a Robert Wood Johnson Foundation Physician Faculty Scholar.
+Department of Family Medicine, The University of Western Ontario, London, Ontario, Canada N6A 5C1

Reprint requests to:

Dr. I.R. McWhinney
Centre for Studies in Family Medicine
The University of Western Ontario
London, Ontario Canada N6A 5C1

Patients with the so-called "somatoform" disorders are very common, especially in primary care (Kirmayer & Robbins 1991a; Bridges & Goldberg 1985). They are also a source of frustration and difficulty for physicians, and frequently strain relationships between doctor and patient. The purpose of this article is to suggest that one source of the difficulty is the unsatisfactory status of the concept of somatization, and of the assumptions on which it is based.

Lipkowski (1988) defines somatization as "a tendency to experience and communicate somatic distress and symptoms unaccounted for by pathological findings, to attribute them to physical illness, and to seek medical help for them," adding that it is usually assumed that the tendency becomes manifest in response to psychosocial stress. Lipkowski's formulation has four components: experiential (patients' experience of distress); cognitive (patients' attribution of the distress to physical illness); observational (the physician's negative findings); and behavioral (the patient's decision to seek care). McDaniel ,Campbell & Seaburn (1990) observe that persistent somatization (somatic fixation) can occur either with or without organic disease. Somatic fixation is defined as "a process whereby a physician and/or a patient or family focuses exclusively and inappropriately on the somatic aspects of a complex problem." McDaniel's formulation thus recognizes the role of the physician and patient's family in the process. Originally, somatization was related to the psychoanalytical concept conversion: the transduction of a psychological conflict into bodily symptoms. Conversion was viewed as a defense mechanism through which the patient avoided having to deal with the conflict and gained some relief from threatening circumstances (secondary gain). Classical psychosomatic theory distinguished between the direct physiological effects of emotion and conversion effects in which psychological conflict could be translated into bodily symptoms by symbolic processes. The symptoms of conversion were therefore forms of communication rather than the experience of physiological disturbances(5) (Schur 1955). Lipkowski's definition is independent of psychoanalytical concepts.

Somatization has now replaced the older terms "hysteria" and "neurasthenia." In the DSM-III (Diagnostic and Statistical Manual of Mental Disorders, 3^rd edition)(1980),"the classic concept of hysteria was split to form a number of `somato form' disorders" (Kirmayer & Robbins 1991b). Conversion disorder (hysterical disorder, conversion type) was retained as a physical expression of psychological conflict or need not under the voluntary control of the patient. Evidence of the psyc hological origins of the symptom includes secondary gain and a temporal relationship between an environmental stimulus and the origin of the symptoms. Conversion disorder was unique in this section of the DSM-III in providing an explanatory theory to acco unt for the symptoms. Somatization Disorder was defined by the age of onset (before 30), chronicity, number of symptoms, and absence of organic pathology. Psychogenic pain disorder was defined as pain without, or out of proportion to organic pathology. Ev idence of secondary gain or an environmental stimulus should be present. Hypochondriasis was retained in its classical form of "an unrealistic fear or belief of having a disease... despite medical reassurance..." Atypical somatoform disorder was a residual category, for use when physical symptoms are not explainable by organic findings or known pathophysiological mechanisms and are "apparently linked to psychological factors".

In the DSM-IV (1994), atypical somatoform disorder is replaced by undifferentiated somatoform disorder, and defined as: one or more physical complaints of more than 6 months duration and causing significant distress; the symptoms either cannot be e xplained by a known general medical condition or drug, or are in excess of what would be expected from physical findings; the disorder is not feigned and is not accounted for by another mental disorder. Although evidence of psychological problems is not a mong the diagnostic criteria, the description of the condition states "medically unexplained symptoms and worry about physical illness may constitute culturally shaped 'idioms of distress' that are employed to express concerns about a broad range of perso nal and social problems, without necessarily indicating psychopathology". Neurasthenia, characterized by fatigue and weakness, is included in this category. Somatoform disorder not otherwise specified is a somatoform disorder which has been present for le ss than 6 months.

From the looseness of these definitions, it is a short step to diagnosing any patient with unexplained symptoms as having a "somatoform disorder." Moreover, to say that symptoms are "employed to express concerns" implies an intention on the part of the patient. After considering all the difficulties with the categorization of these disorders, Kirmayer and Robbins (1991b) suggest that "with the possible exception of somatization disorder per se, the somatoform disorders appear to be best thought of as symptoms or patterns of reaction rather than discrete disorders with a discrete natural history."

The verb "to somatize" and the noun "somatizer" are unique in the vocabulary of medicine, implying an action performed by a patient on his own body. For most diseases, there is no word which signifies the unique patient. We only say that a pati ent "has" pneumonia or cancer. For some, the ending "-ic" (diabetic, schizophrenic) implies that the disease is inseparable from the person, although there is no suggestion that the patient is responsible for the disease. Only with the stigmatizing term s omatizer is it implied that the patient is the author of his own bodily suffering.

The term somatization is a product of Western medicine's dualistic ontology. The assumption is that emotions, instead of being expressed symbolically in words, are "transduced" to bodily events. A further assumption is that our emotions are not emb odied in the first place. Our ethnocentricity hides from us how aberrant this belief is. In many societies the concept of somatization has no meaning, since distinctions between mental and physical illness are not prevalent (Fabrega 1991). In Ayurvedic an d traditional Chinese medicine, illness (the experience of symptoms) and disease (biological processes) are not separate categories. Illnesses are conceived in terms of imbalance and are rooted in the report of symptoms. There is, therefore, no logical pl ace for the idea of somatization. Even in the European tradition, the experience of illness in all forms was accorded substantial validity until the emergence of modern diagnostic technology in the nineteenth century focused attention almost exclusively o n the bodily processes.

Bodily forms of communication are accepted in all cultures, though in the West, especially among the educated, the verbal expression of feeling is usually considered more desirable. Even the most articulate patients, however, may find it difficult to put their feelings into words. William Styron (1990) described his severe depression as "so mysteriously painful and elusive in the way it becomes known to the self, to the mediating intellect, as to verge close to being beyond description." If this is so of the highly literate, how much more difficult must it be for the inarticulate. Words may also fail because the emotion is too overwhelming, or because the trauma giving rise to it was preverbal (Seaburn 1995).

The notion of the disembodiment of the emotions is quite recent even in Western medical thought. Classical and neoclassical medical theory maintained that a definite relationship existed between emotions such as anger, fear and love, and physic al states (Rather 1965). Contrary to modern assumptions, Descartes did not deny mind-body interactions, but maintained that most aspects of affective states are primarily somatic (Brown 1989). Until the nineteenth century a unitary view of illness prevail ed and diagnosis often meant diagnosis of a patient rather than of a disease. The replacement of this unitary view by the notion of diseases having a bodily location led eventually to the conceptual separation of mind from body. Commenting on the manuals of clinical method which appeared at the turn of the twentieth century, Crookshank (1926) noted that they: "give excellent schemes for the physical examination of the patient whilst strangely ignoring, almost completely, the psychical.". An enquiry into t he emotions was by that time no longer deemed necessary for the clinician's understanding of illness.

This change in the medical worldview was reflected in a transformation in the popular view of the human body. For the eighteenth century patient there was no separation between the emotions and the body. Nor was there a distinct boundary between th e doctor's diagnostic vocabulary and the feelings of the patient (Duden 1991). To an eighteenth century patient, the idea of the emotions as being "in the head" would probably not have occurred.

Even as these concepts were becoming entrenched in modern medical thought, some dissenting voices were raised. William James(1950) held that bodily sensations were prior to, and the origin of, feelings of grief, joy, anger, etc. Although James' the ory could not account for the neurophysiologic complexities of emotional response, his central insight into the body's involvement is still valid. With the somatic marker hypothesis, Damasio (1994) postulates a neural mechanism throu gh which bodily feelings influence human response, either consciously or beneath the level of awareness: a process which Ciompi(1991) calls "affect logic."

Symptom attribution, a key element in the concept of somatization - is an interpretive process, strongly influenced by cultural factors (Kirmayer, Young & Robbins 1994). Cultures vary widely in their openness in expressing emotion and in assump tions about mind-body relationships. In the West, the assumption that bodily expressions of emotions are "in the mind" and therefore not "real," and the blame attached to mental illness, naturally encourage the interpretation of bodily expression of emoti on as physical disease. For some psychiatrists, blame-avoidance is the key feature of somatization (Bridges & Goldberg 1985).

Although not explicit in the concept, the language of somatization suggests that the expression of emotion in bodily symptoms is abnormal and that cure requires that the mental "causes" of the symptoms to be acknowledged, verbalized, and resolved, and that symptomatic treatment is not appropriate, even though it has some empirical justification (Tyrer 1973). The requirement that the physician convince the patient of the mental origins of his symptoms may create the ground for an irreconcilable conf lict, for why should a patient acknowledge something he does not feel? A diagnosis of somatization, especially when associated with the idea of primary and secondary gain, carries with it the implication of moral failure with all the consequences in stigm atization and the breakdown of relationships (Kirmayer 1988).

Another pitfall is the opportunity for misdiagnosis of organic disease. The definition of somatization is broad enough for the clinician to diagnose as somatoform any illness with physical symptoms but no pathological findings with existing investi gative technologies. Chronic fatigue syndrome and irritable bowel syndrome have both been categorized as somatoform disorders, though the evidence indicates that these disorders cannot be so readily explained (Goodnick & Limas 1993, Drossman et al. 19 87). Even patients with early multiple sclerosis have been diagnosed as somatizers.

Although the problematic nature of somatization has been widely recognized, the logical step of abandoning the concept has not been taken. This requires us to change deep-seated assumptions. First we need to accept that emotions are normally ex perienced in the body. We should reject the idea of somatoform disorders as diagnostic entities and learn that a symptom may be an embodied emotion, indeed that all symptoms, whatever their origin, have some affective "coloring." Whether or not they are e xperienced as emotions may or may not be abnormal, depending on the patient's cultural background. If they are not so experienced, simply by telling the patient that his symptoms are emotional will not be helpful. The connection between emotions and bodil y state must be made at both the affective and cognitive levels by the patient. We would also have to abandon the belief that the only way to deal with embodied emotions is to "re-transduce" them to mental states. Physical therapies may also be effective in helping the patient to make the breakthrough to a new level of understanding without the requirement of verbalization. With the possible exception of somatization disorder, the only category with empiric validation, somatoform disorders should not appe ar in textbooks of medicine and psychiatry as disease categories which imply that they have the same epistemological status as discrete disorders with a verifiable natural history. A likely consequence of changing our assumptions would be a felt need to f ind a new name for these illnesses - one that integrates the physical and psychologic aspects.

Many of these problems would be avoided if physicians and patients became less concerned with cause and more with care (Goodnick & Limas 1993). Once remediable causes have been ruled out, many illnesses have to be managed without knowing their cause. Even when no specific remedy is available, wise physicians have always aimed to help the patient by attention to the particulars of the illness, to diet, rest, sleep, appropriate exercise, symptomatic treatment and emotional support. The support of the physician is crucial, and it is this which patients regarded as "somatizers" often lack. Building trust is especially difficult when a patient has experienced rejection from physicians.

In focusing on care, attention is switched from causes of the illness to causes of chronicity, exacerbation, and relapse. These may be different from one patient to another. There will be many occasions when the physician has good reason to think t hat the bodily symptoms are an expression of emotion which has not yet entered consciousness. Sometimes awareness of the link is so close to conscious recognition that exploration of the patient's feelings will enable him to make the connection. In other patients, reaching this self understanding will be a longer process. Focusing on care does not mean abandoning the idea of secondary gain, or forgetting that a patient can, by responding to illness maladaptively, inhibit its resolution. Secondary gain, ho wever, can accrue from any illness, and should not be tied particularly to the so-called somatoform disorders. Trying to modify maladaptive responses carries the risk of doctor-patient conflict, but this is lessened if there is genuine acceptance by the p hysician of the patient's suffering.

Patients vary widely in response to symptoms, but this need not imply that the symptoms are "somatized." Those who attend physicians are likely to be more anxious than those who cope with their symptoms in other ways (Drossman et al. 1987). This ex plains their illness behaviour, not the mechanism of their symptoms. At one extreme end of the scale of responsiveness to symptoms are those who are highly sensitive to bodily sensations, and especially liable to attribute them to bodily diseases. This cl assical description of hypochondria is well known and can be retained, without assuming that the symptoms arise from the transduction of emotions.

Changing the language we use will not make these illnesses go away. Nevertheless, language both expresses and influences how we think and act. Although changing the words will not necessarily change our practice, there must eventually be congruence between thought, language, and actions. What changes in thought and action are required? First, the recognition that the patient's experience of illness is primary. Our own system of diagnostic abstractions, though very powerful, is secondary. In the pat ient's experience, there may be no separation into mental and physical: illness is "a disturbance in a person's ability to relate to and function in the world"(Baron 1985), whether or not it is associated with identifiable organic pathology. The biopsycho social model (Engel 1980), and the patient-centred clinical method (Stewart et al. 1995) require that the clinician attend to the emotions as a routine part of the clinical inquiry. To attend to the emotions only in certain kinds of illness, or only after diagnostic testing is negative, perpetuates the prevailing dualistic distinction between mental and physical illness. All significant illness is a disturbance at multiple levels, from the molecular to the personal and social. This implies that some of th e skills that are at present considered "psychiatric" will need to be more general in all clinicians, especially those working in primary care, where so much general undifferentiated illness is seen.

We thank Dr. Theodore M. Brown for his helpful suggestions.