McWhinney and colleagues [1] make a spirited argument for the incorporation of the biopsychosocial model into the management of somatic symptoms. It is hard to dispute the verity of this model, but we fail to see the relevance of abandoning the categorization of chronic unexplained symptoms as somatoform disorders. The authors contend that the label somatizer points to patients as the "authors of their own bodily suffering." However, this attribution of moral culpability is not unique to somatization but applies to mental disorder diagnoses in general. Only recently have the public and the health care profession begun to classify depression, alcohol disorders, and schizophrenia as legitimate illnesses rather than personal failings.

Although limited empirical data support some somatoform disorders (other than somatization disorder, a disorder relatively rarely seen in primary care) as valid constructs, ample evidence now suggests that as a discrete disorder, chronic unexplained symptoms are linearly associated with substantial functional impairment (that is, the more unexplained symptoms present, the greater the functional impairment) [2]. Dysfunction-as measured by self-reported disability days and health-related quality of life-is comparable to depressive and anxiety disorders when the somatizing patient has three or more unexplained symptoms lasting for more than 2 years; this entity has been newly defined and validated as multisomatoform disorder [2]. When clinically significant somatization is identified and treated as a discrete disorder (by establishing a relationship with the patient, providing frequent follow-up, and avoiding excessive testing), health outcomes improve and costs decrease [3]. Identifying valid and meaningful patterns of illness for which focused and effective therapies can be implemented is what matters to busy clinicians. We need more of this kind of "human science" to verify the usefulness of the biopsychosocial model. Establishing a valid categorization of chronic unexplained symptoms makes it easier for the busy clinician to cease excessive, low-yield diagnostic testing and to begin to focus on the supportive, psychological needs of these patients. This is consistent with Engel's biopsychosocial concept of disease [4].

McWhinney and colleagues' argument seems impractical and extreme and might even exacerbate the holist-reductionist dichotomy. Somewhere in between lies the truth. We agree that the patient's experience of illness is of primary concern and that diagnostic abstractions are secondary. But if such abstractions are useful in alleviating illness, as current research in somatoform disorders shows, then abandoning that construct would be counterproductive. Would you do the same for anxiety and depressive disorders?