Auditing Cause of Death Reporting to Australia and New Zealand's Dialysis and Transplant Registry: A Retrospective Cohort Study.

Aim: National data registries provide a valuable source of data for epidemiological research but may be subject to inaccuracies. Whilst studies have compared agreement between cause of death (COD) data from Australia and New Zealand's dialysis and transplant registry (ANZDATA) to other databases, no studies have manually compared agreement with electronic medical records (EMR). This study aimed to assess the agreement between COD for dialysis patients in the Western Sydney renal service according to ANZDATA and EMR.

Methods: We conducted a retrospective cohort study on dialysis patients from the Western Renal Service who died between January 1, 2016 and July 31, 2022, inclusive. We matched ANZDATA patient records to our service's EMR, extracted data on COD from EMR and coded COD according to ANZDATA's categorisation system. Agreement between COD from EMR and ANZDATA was assessed using kappa statistics.

Results: There were 709 deaths in the study period. 449 (63.3%) were male and 260 (36.7%) were female. Of the 539 patients with available records, the most common COD categories were cardiovascular (201, 37%) and infection (154, 29%). There was poor agreement between individual COD from EMR and ANZDATA (kappa of 0.39). However, this improved when comparing broader COD categories (kappa of 0.53). Subgroup analysis showed no difference in agreement between males and females or patients who died between 2016-2018 and 2019-2022.

Conclusion: There is poor agreement between ANZDATA and EMR for individual COD. However, agreement improved to moderate using broader categories of COD, suggesting higher accuracy when utilising registry COD data for epidemiological analysis.