The sensitivity and specificity of computerized databases for the diagnosis of rheumatoid arthritis

Objective. To examine the accuracy of a computerized medical database for the diagnosis of rheumatoid arthritis (RA). Methods. The complete medical records of all prevalent cases of RA (according to the 1987 American College of Rheumatology diagnostic criteria) on January 1, 1987 were reviewed to determine the sensitivity, specificity, and predictive value of database diagnoses compared with those obtained by medical record review. Agreement between database and medical record diagnoses was calculated using the kappa statistic. Results. Computerized database diagnoses of RA had a sensitivity of 89%, a specificity of 74%, a positive predictive value of 57%, and a negative predictive value of 94% compared with diagnoses based on clinical information abstracted from the complete medical record. Agreement between database and medical record diagnoses was poor (K = 0.54). Conclusion. The sole reliance on such databases for the diagnosis of RA can result in substantial misdiagnosis.