Validity of the Department of Defense Standard Inpatient Data Record for quality management and health services research.

ABSTRACT

OBJECTIVE: To assess the validity of the Department of Defense's Standard Inpatient Data Record (SIDR) for health services research and quality measurement. SUBJECTS: Patients whose inpatient charts were abstracted through the Civilian External Peer Review Program's studies of acute myocardial infarction (N = 1,432) and 1993 review of the birth product line (N = 9,705). METHODS: Separate databases of professionally abstracted (the clinical data set) and hospital-reported (the administrative SIDR data set) diagnoses and procedures were compared for each patient, and the sensitivity and specificity of the SIDR for elements in the Civilian External Peer Review Program's clinical "gold standard" data set were calculated. Agreement beyond chance was examined with kappa statistics. RESULTS: The clinical data set's principal procedure was found as a SIDR principal or secondary procedure in 92.5% of cases. Sensitivities of the SIDR data for common diagnoses in the clinical data ranged from 64% (viral infection) to 97% (diabetes), with kappa statistics ranging from 0.55 to 0.96. Procedural sensitivities ranged from 77% (echocardiography) to 99% (cesarean section), with kappa statistics ranging from 0.7 to 1.0. CONCLUSIONS: Our analyses indicate that the Department of Defense's SIDR is a reliable source of administrative data that compares favorably with traditional civilian quality management and health services research data sources, such as those from the Health Care Financing Administration and large insurers.