Reliability of Psychiatric Diagnoses: 2. A Study of Consistency of Clinical Judgments and Ratings

An investigation of the reliability of psychiatric diagnoses was designed to minimize factors that would artificially lower or inflate the rate of concordance. A series of 153 patients were examined independently by paired psychiatrists and diagnoses were made independently according to the standard nomenclature. The degree of agreement (54%) on specific diagnoses was statistically significant (p<.001) and was higher than that obtained in other comparable studies. 2. In cases where both diagnosticians indicated they were certain of the diagnosis, the agreement rate (81%) was found to be significantly higher than in the remaining cases. 3. When the diagnosticians gave both a preferred diagnosis and an alternative diagnosis, it was found that the rate of agreement between either diagnosis offered by one diagnostician and either diagnosis of the other was 82%. This suggested that the diagnosticians may have been closer in their appraisals than indicated by the scoring of only the preferred diagnoses. 4. An additional method of classification consisted of rating the patients on a 4-point scale along a single dimension, viz., the depth of depression. It was found that, when they used this method, the diagnosticians agreed within one scale unit in 99% of the cases. 5. It was suggested that the present system of diagnosis could be improved by focusing greater attention on appropriate training in diagnostic skills, the identification of the defects in current interviewing techniques, and the development of more uniform clinical procedures.