Abstract
OBJECTIVES: The purpose of this study was to assess agreement on cause of death reporting in end-stage renal disease patients by comparing death certificates and reports to an end-stage renal disease registry. METHODS: Death certificates and registry reporting forms were retrieved for a random sample of 335 treated end-stage renal disease patients who died between 1980 and 1986 in Maryland. On the registry form, patient death was ascribed to 1 of 22 precoded causes by the patient's nephrologist of record. Corresponding death certificates were coded, according to rules of the International Classification of Diseases, 9th edition, by a trained observer unaware of the registry report. Agreement was measured by the kappa statistic. RESULTS: Overall cause of death agreement was poor (31%), and varied by the following categories: renal disease (40% on death certificates vs 0% on registry reports), cardiovascular disease (26% vs 47%), infectious disease (16% vs 22%), cancer (7% vs 5%), and withdrawal from therapy (0% vs 3%). Agreement was higher for transplant recipients than for dialyzed patients. CONCLUSIONS: Death certificates and registry reports yield different descriptions of mortality in end-stage renal disease patients. These sources of information should not be used interchangeably. Improvements to International Classification of Diseases coding of renal diseases and the determination of the reliability and validity of the US Renal Data System reporting process are necessary steps in the development of renal disease epidemiology.