Dynamics of Decision-Making: The Issue of Reliability in Diagnosis
Diagnosis and the DSM
Publications on diagnosis since the DSM-III often mention the heroic effort made by many in terms of improving its reliability in psychiatry. This is illustrated in the DSM-5 (p. 5), where it states that the “DSM has been the cornerstone of substantial progress in reliability.” Indeed, while many proponents of the DSM admit that its validity still poses an unresolved problem, most seem to believe that the manual facilitates a relatively reliable assessment of human suffering. Underlying this line of reasoning is the idea that in the pre-DSM epoch a certain Babylonian confusion of tongues posed problems for psychiatry: different theories existed next to one another, and the concepts used to refer to different disorders were not properly operationally defined, allowing diagnosticians’ personal interpretations to strongly guide their diagnostic assessments. With the use of descriptive criteria since DSM-III, all of this changed: finally a sufficiently unambiguous language was found to allow a more rigorous evaluation of observable behaviors. However, the commonly accepted idea of “substantial progress in reliability” is not a proven fact, but an assumption that deserves closer examination.
© The Author(s) 2017
S. Vanheule, Psychiatric Diagnosis Revisited,
The issue of reliability is complex and cannot be disconnected from the question of how one thinks of diagnosis as such. In the early days of psychiatric thinking, diagnosis started from prototypes: handbooks described different forms of psychopathology, which were documented with an array of clinical illustrations. Clinicians made use of these prototypical descriptions in assessing and describing the extent to which a given patient’s complaints resembled those depicted in the literature, usually ending with categorical assertions on the kind of pathology a patient was suffering from. In this chapter I argue that this prototype- based method of diagnosis lost momentum under the pressure of nonclinical concerns, culminating in a singular focus on classification. The DSM-I (1952) and II (1968) start from prototypes, but reduce prototypical description to a strict minimum, thus illustrating what remains left of diagnosis when classification is its main target. Below I argue that in the 1970s the issue of the reliability of diagnosis became a high priority. During those days, psychiatry was in crisis due to (at least) three important influences: empirical studies began to demonstrate the unreliability of psychiatric decision-making, critical scholars questioned the ethos of psychiatry, and societal changes facilitated a shift toward quantifying mental health policies. The DSM-III (1980) thrived on this crisis and promised a new era of scientific stringency. Prototype-based diagnosis was discredited, while criterion-based diagnosis along the principles of biomedical thinking was hyped. On this point, it is often assumed that since the 1980s psychiatric diagnosis has made great progress. I contend that the statistical reliability of the DSM is seriously overestimated. In 1997 H. Kutchins and S.A. Kirk made such a claim, but the field trials conducted for the DSM-5 in particular illustrate that the “good news parade” on the reliability of the DSM hid its weaknesses for decades. Nevertheless, let us start where it all began and first discuss the shift from prototype-based to criterion-based diagnosis.