Abstract
BACKGROUND AND OBJECTIVES: The United States Renal Data System (USRDS) is a commonly utilized database for epidemiologic research of ESRD patients. USRDS uses Medical Evidence Form 2728 to collect medical information about ESRD patients. The validity of the Form 2728 "primary cause of renal failure" field for glomerular diseases has not been evaluated, although inconsistencies between Form 2728 information and medical records have been documented previously with respect to comorbidities. DESIGN, SETTING, PARTICIPANTS, & MEASUREMENTS: Form 2728 information was linked with renal biopsy results from the Glomerular Disease Collaborative Network (GDCN) for 217 patients with biopsy-confirmed glomerular diseases who had reached ESRD. Biopsy results were compared with the Form 2728 "primary cause of renal failure" field. Diseases were considered individually, and also categorized into commonly used disease groups. Percentage of agreement and disease-specific measures of validity were calculated. RESULTS: Overall agreement between renal biopsy and Form 2728 was low (14.8% overall, 23.0% when categorized). Agreement was better after Form 2728 was revised in 1995 (10.0% before versus 23.2% after overall). The cause of ESRD field was left blank in 57% of the forms submitted for glomerular disease patients. Individual glomerular diseases had very low specificities, but tended to have high positive predictive values. CONCLUSIONS: Form 2728 does not accurately reflect the renal pathology diagnosis as captured by biopsy. The large degree of missing data and misclassification should be of concern to those performing epidemiologic research using Form 2728 information on glomerular diseases.