Source validation of pancreatitis-related hospital discharge diagnoses notified to a national registry in the Netherlands

Background: National and regional disease registries are often used for epidemiological studies and validation analyses are scarce. We analyzed the reliability of the National Information System on Hospital Care (NISHC) registry used to classify admissions for acute and chronic pancreatitis in the Netherlands. Methods: All pancreas-related discharge diagnoses notified by the Academic Medical Center to the NISHC were retrieved for a 2-year time period. A review of multidisciplinary patient reports was conducted to verify these diagnoses. Results: 284 patients were notified to the NISHC, relating to 483 admissions with 523 pancreas-related discharge diagnoses. Of these diagnoses, 112 were coded as acute pancreatitis, 250 as chronic pancreatitis and 161 were classified as pseudocysts and other pancreatic diseases. The positive predictive value for acute pancreatitis diagnosis codes was 77.7% and 46 additional acute pancreatitis diagnoses were identified, ultimately leading to an underestimation of the total number of acute pancreatitis diagnoses of 15.8%. The positive predictive value for discharge diagnoses regarding chronic pancreatitis was 83.6% and 57 additional chronic pancreatitis diagnoses were identified, finally resulting in an underestimation of the total number of chronic pancreatitis diagnoses of 6%. Conclusion: There is a substantial miscoding on a person level of discharge diagnoses of acute and chronic pancreatitis. On a group level, when miscoding between categories is leveled out, actual numbers of admission are only slightly underestimated. Copyright (C) 2008 S. Karger AG, Basel and IAP