Investigating the accuracy of ethnicity data in New Zealand hospital records: Still room for improvement

Background: The accuracy of ethnicity information in the New Zealand hospital data was reported on in 1994. Data collected in the Barriers: to Diabetes Care in the Waikato Study enables further evaluation of the accuracy of ethnicity information in hospital records. Aims: One aim of public health policy is addressing health disparities between ethnic groups. Monitoring disparities depends on accurate outcome data, such as that from hospitals. It would be expected that this data would improve over time. This paper reports on the contemporary accuracy of ethnicity data in hospital records in the Waikato district. Methods: Self-identified ethnicity data were gathered as part of the Barriers to Diabetes Care in the Waikato mail survey. Hospital record data were collected for those participants who had consented for access to their hospital records. Results: Complete data was available for 3500 people with diabetes. Ethnicity in the hospital record was correct for one of the sometimes multiple, self-identified ethnicities for 97.7 (95CI96.8-98.3)% of respondents. Ethnicity data were concordant for 71 (67-75)% of Māori and 99 (99-100)% of non-Māori. The non-Māori ethnic group was disaggregated into component groups: the hospital record agreed with self identified ethnicity for 89 (87-91)% of Europeans, 67 (55-78)% of Pacific groups, 70 (57-81)% of South Asian groups, 64 (48-47)% of Asian groups, and 41 (27-57)% of 'Other" ethnic groups. Conclusions: Hospital records continue to mis-record ethnicity when compared to a self-identified ethnicity. Mis-recording occurs for all ethnic groups, and is more pronounced at more specific levels of ethnic group. Researchers, clinicians, and policy makers must be cognisant of these continuing discrepancies when using hospital record data to describe ethnic variations in health status, service utilisation, or for policy planning activities.