Identifying young Aboriginal and Torres Strait Islander children in linked administrative data: A comparison of methods

BACKGROUND: In the ongoing debate on optimum methods for identification of Indigenous people within linked administrative data, few studies have examined the impacts of method on population counts and outcomes in family-based linkage studies of Aboriginal children. OBJECTIVE: To quantify differences between three algorithms in ascertaining Aboriginal and Torres Strait Islander children in linked administrative data. METHODS: Linked administrative health data for children born in Western Australia (WA) from 2000-2013, were used to examine the cohorts identified by three methods: A) the Indigenous Status Flag (ISF, derived by the WA Data Linkage Branch using a multistage-median approach) for the children alone; B) the ISF of the children, their parents and grandparents; and C) Indigenous status of the child, mother or father on either of the child’s perinatal records (Midwives or birth registration), to determine differing characteristics of each cohort. RESULTS: Method B established a larger cohort (33,489) than Method C (33,306) and Method A (27,279), with all methods identifying a core group of 26,790 children (80-98%). Compared with children identified by Method A, additional children identified by Methods B or C, were from less-disadvantaged and more urban areas, and had better perinatal outcomes (e.g. lower proportions of small-for-gestational age, 10% vs 16%). Differences in demographics and health outcomes between Methods C and B were minimal. CONCLUSIONS: Demographic and perinatal health characteristics differ by Aboriginal identification method. Using perinatal records or the ISF of parents and grandparents (in addition to the ISF of the child) appear to be more inclusive methods for identifying young Indigenous children in administrative datasets. KEYWORDS: Aboriginal health, identification, data linkage, Indigenous, child, methodology