Abstract
Background
Accurate ethnicity reporting is a fundamental component of monitoring inequities in health outcomes for Indigenous Peoples and minoritised peoples. The Australia and New Zealand Dialysis and Transplantation registry (ANZDATA) is the national clinical quality registry for people treated with kidney replacement therapy that forms the evidentiary basis for monitoring healthcare delivery and outcomes of kidney failure in Aotearoa New Zealand.
Objective
We evaluated the accuracy of ethnicity recording in ANZDATA for Māori and Pacific Peoples.
Design, setting, participants
Population-based analysis of ethnicity recording using non-standardised methods in people commencing kidney replacement therapy between 2006 and 2019 in Aotearoa New Zealand. Registry data (ANZDATA) were compared with hospitalisation data (National Minimum Dataset) using standardised methods as the reference standard using deterministic data linkage.
Main outcome measures
Measures of agreement between datasets (sensitivity, specificity, positive and negative predictive values) overall and by demographic and clinical characteristics; trends in sensitivity over time.
Results
Of 7678 people commencing kidney replacement therapy, 2436 (31.7%) were identified as Māori using standardised methods in hospitalisation data, of which the registry recorded 2299 (94.6%). Overall sensitivity of the registry for Māori was 94.6% (95% confidence interval (CI) 93.6–95.4%), with high specificity (98.5%, CI 98.1–98.8%). Overall, 1718 (22.4%) people were identified as Pacific Peoples by hospitalisation data, with 1552 (90.4%) recorded in the registry. Overall sensitivity for Pacific Peoples was 90.4% (CI 89.0–91.8%), with high specificity (99.1%, CI 98.9–99.3%).
For Māori and Pacific Peoples, inter-rater reliability (Cohen’s kappa) exceeded 0.81 both overall and in subgroup analyses, suggesting high agreement between the datasets. Non-reporting of Māori identification in ANZDATA was associated with a shorter time to kidney transplantation (HR 2.20, CI 1.39–3.51), however there was no evidence of an association of ethnicity reporting concordance with all-cause death (HR 0.80, CI 0.60–1.06). Among Pacific Peoples, there was no evidence that agreement of ethnicity reporting between data sources was associated with kidney transplantation (HR 1.15, CI 0.72–1.83) or death from any cause (HR 1.29, CI 1.00–1.67). There was no evidence of variation in the agreement of ethnicity reporting between data sources over time.
Conclusion: Although ethnicity data collection in ANZDATA is currently not standardised, there was consistently high agreement between registry and hospitalisation data over the study period.