The Danish Registry on Regular Dialysis and Transplantation: completeness and validity of incident patient registration

The Danish National Registry on Regular Dialysis and Transplantation (NRDT) provides systematic information on the epidemiology and treatment of end-stage chronic kidney disease in Denmark. It is therefore of major importance that the registry is valid and complete. The aim of the present study was to evaluate the registration of incident patients on chronic renal replacement therapy (RRT).

Incident patients on chronic RRT in the period 2001-2004 were identified in NRDT and in the National Patient Registry, which contains information on hospital admissions and treatments. In the National Patient Registry, identification of patients was as follows: patients receiving the procedure of dialysis during a minimum of 90 days and for a minimum of 12 times or the procedure of renal transplantation. Only patients with at least 2 years of dialysis-free interval before and never being transplanted were included. The completeness of NRDT was calculated as the percentage of new patients on chronic RRT registered in the National Patient Registry also found in NRDT. Validity of data in NRDT was assessed by information from medical records and analysed using kappa statistics.

Completeness of NRDT: Of 3020 patients registered in the National Patient Registry as incident chronic RRT patients, 97.2% were found in NRDT but 22.5% with another year of entry. There were no differences in completeness between hospitals or regions. Validity of NRDT: Validity of common renal diagnoses and RRT modality was high: diabetic nephropathy (kappa = 0.98), adult polycystic kidney disease (kappa = 0.95), chronic glomerulonephritis (kappa = 0.78) and RRT modality (kappa = 0.94). The diagnosis CKD of unknown aetiology and type of diabetes were less valid (kappa = 0.62, 0.60 and 0.73, respectively). The date of RRT start had also high validity.

Completeness of incident patient registration in NRDT was highly acceptable. Validity of incident patient data was also good, except for type of diabetes.