The Politics of Diagnosis

Over the years, psychiatrists have worked to reduce problems with diagnosis by re- fining the definitions of illnesses in the Diagnostic and Statistical Manual of Men- tal Disorders (DSM), first published by the APA in 1952. Virtually all psychiatrists use this manual for assigning diagnoses, as do most other clinicians, because insurers usually require a DSM diagnosis before they will reimburse clinicians for care.

The DSM and the subsequent DSM-II, published in 1968, instructed clini- cians to reach diagnoses based on the clinicians’ inferences about such intrapsychic processes as defenses, repression, and transference. Because clinicians can’t measure these processes, the same behavior often elicited quite different diagnoses from different clinicians.

Partly because of these problems, in 1974, the APA announced its decision to revise the DSM-II. Ironically, although the resulting DSM-III, published in 1980, was designed to quiet questions about the ambiguities of psychiatric diagnosis, it instead illuminated those ambiguities because its writing became an overtly political battle involving active lobbying by both professional and lay groups. This battle revealed wide differences among clinicians regarding what behaviors signified mental ill- ness, what caused those behaviors, who should treat them, and how they should be treated.

To encourage support for the DSM-III and avoid open political battles among psychiatrists, its authors decided to stress symptomatology and avoid discussing either causation or treatment. In addition, to increase the odds that clinicians would use the DSM-III, the authors described the various diagnoses based not on available research but on the consensus among practicing psychia- trists. These two strategies, they hoped, would produce a widely used and highly reliable document. Reliability refers to the likelihood that different people who use the same measure will reach the same conclusions—in this case, that differ- ent clinicians, seeing the same patient, would reach the same diagnosis. Yet even this modest goal was not achieved because studies continue to find high rates of disagreement over diagnosis. Moreover, reliability in the absence of validity is not particularly useful. Validity refers to the likelihood that a given measure accurately reflects what those who use the measure believe it reflects—in this case, that persons identified by the DSM-III as having a certain illness actually have that illness. As Phil Brown notes, “Anyone can achieve . . . reliability by teaching all people the ‘wrong’ material, and getting them to all agree on it. . . . The witch trials [of earlier centuries] showed a much higher degree of interrater reliability than any DSM category, yet we would not impute any validity to those social diagnoses.”