Test-retest reliability of the PRIME-MD: limitations in diagnosing mental disorders in primary care

Abstract

The primary care evaluation of mental disorders (PRIME-MD) can be seen as characteristic for successive refinements of criteria and structured interview techniques for diagnosing psychiatric disorders in primary care. It is one of the most widely used instruments, but there is no evidence to support its test-retest reliability. With 1-week intervals between interviews, a test-retest study of the PRIME-MD was conducted in a general practice population of 100 distressed patients (20- to 60-years old) who were on sick leave. Almost everyone (89%) received one or more diagnoses at both measurements, and there was fair total agreement (kappa = 0.27). The best agreement was found for more severe threshold disorders [major depressive disorder (kappa = 0.58), dysthymia (kappa = 0.57), and generalized anxiety disorder (kappa = 0.59)], while we found indefinite results for the sub-threshold disorders [anxiety disorder not otherwise specified (NOS) (kappa =.30), minor depressive disorder (kappa = -0.03), and somatoform disorder NOS (kappa = 0.11)]. The PRIME-MD is one of the few instruments in primary care that actually diagnoses specific mental disorders according to the DSM criteria. However, there was a failure to adequately classify sub-threshold disorders. Mental disorders, as seen in primary care, encompass important specific symptoms and clinical syndromes that vary in duration and severity over time, but they also encompass an admixture of somatic and psychological symptoms that do not match current diagnostic systems. This most likely resulted in methodological uncertainty about the level of agreement. Diagnostic criteria in psychiatry need to be operationalized for use in primary care and require further evaluation.