The accuracy of ethnicity data in primary care and its impact on cardiovascular risk assessment and management. PREDICT CVD-8

Abstract

Background Accurate ethnicity data are a prerequisite for evidence-based cardiovascular risk assessment and management according to national guidelines. Aims (i) To investigate the accuracy of ethnicity data in primary care medical records by comparing them with self-identified ethnicity. (ii) To determine the clinical impact of ethnicity misclassification on cardiovascular risk assessment and management. Methods A random sample of 870 patients from 18 general practices (who had ethnicity collected from their medical record as part of cardiovascular risk assessment using PREDICT, a web-based decision support tool) were sent a postal questionnaire asking their self-identified ethnicity using the 2001 Census ethnicity question. Results Data were available for 665 people (77% response rate) who completed the postal questionnaire. Ethnicity in the primary care record and self-identified ethnicity from the questionnaire were identical for 68% of respondents at Statistics New Zealand Level 2 coding. Data concordance varied from 9.8% for the non-New Zealand European ethnic group to 90.9% for New Zealand European. The primary care record agreed with self-identified ethnicity for 64.9% of Māori respondents. Fortunately, when the same ethnicity data were categorised using the Statistics New Zealand ethnic group prioritisation rules and applied within PREDICT, which adds a risk weighting for Māori, Pacific, and Indian subcontinent peoples, the impact of misclassification was small. The main reason was that about half of misclassifications occurred between ethnic groups classified in the same high cardiovascular risk category. For about 6% of Māori, Pacific, and Indian subcontinent people in our study this misclassification could potentially have delayed risk assessment and resulted in under-treatment. In contrast, about 1.5% of those with other ethnicities may have undergone a premature risk assessment and been over-treated. Conclusion The clinical impact of ethnicity misclassification on cardiovascular risk assessment and management in primary care is modest because much of the misclassification does not alter cardiovascular risk classification. Nevertheless, efforts to improve the accuracy of ethnicity classification in primary care need to continue in order to support the sector’s ability to monitor health service utilisation, outcomes, and performance related indicators.