Background
Accurate recording of Aboriginality on health data collections is essential to measuring the effectiveness of government policies and programs aimed at reducing the health gap. Using record linkage, the Enhanced Reporting of Aboriginality (ERA) algorithm was developed in 2012 to assess the weight of evidence as to whether a person would be reported as Aboriginal or Torres Strait Islander for statistical purposes. Record linkage with the NSW Patient Survey Program (PSP) provides an opportunity to validate and improve the ERA algorithm.
Aim
To evaluate and assess possible improvements to the ERA algorithm using self-reported Aboriginality derived from the PSP.
Methods
Data source: Linked records from Hospital, Emergency Department (ED), Birth and death Registrations, and Perinatal Data from 2011 to 2017 for participants in the BHI patient survey for admitted patient and ED from 2013 to 2015
Data Analysis: Self-reported Aboriginality in the PSP was compared with 1) the ERA value calculated from the linked dataset, 2) Aboriginality as reported on the recorded associated with the survey, 3) alternative algorithms.
Results
Of the 133,510 people in the linked dataset, 5102 reported themselves to be Aboriginal or Torres Strait Islander on the PSP, compared to 4732 on the linked administrative data. Sensitivity, specificity, positive and negative predictive values of the original ERA algorithm were high. Accuracy measures varied by age, and sub-group analysis informed improvements to the ERA algorithm.
Conclusion
Record linkage with the PSP has allowed validation of the ERA algorithm and informed improvements.