Published in:
Open Access 01-12-2010 | Research article
Under-ascertainment of Aboriginality in records of cardiovascular disease in hospital morbidity and mortality data in Western Australia: a record linkage study
Published in: BMC Medical Research Methodology | Issue 1/2010
Login to get accessAbstract
Background
Measuring the real burden of cardiovascular disease in Australian Aboriginals is complicated by under-identification of Aboriginality in administrative health data collections. Accurate data is essential to measure Australia's progress in its efforts to intervene to improve health outcomes of Australian Aboriginals. We estimated the under-ascertainment of Aboriginal status in linked morbidity and mortality databases in patients hospitalised with cardiovascular disease.
Methods
Persons with public hospital admissions for cardiovascular disease in Western Australia during 2000-2005 (and their 20-year admission history) or who subsequently died were identified from linkage data. The Aboriginal status flag in all records for a given individual was variously used to determine their ethnicity (index positive, and in all records both majority positive or ever positive) and stratified by region, age and gender. The index admission was the baseline comparator.
Results
Index cases comprised 62,692 individuals who shared a total of 778,714 hospital admissions over 20 years, of which 19,809 subsequently died. There were 3,060 (4.9%) persons identified as Aboriginal on index admission. An additional 83 (2.7%) Aboriginal cases were identified through death records, increasing to 3.7% when cases with a positive Aboriginal identifier in the majority (≥50%) of previous hospital admissions over twenty years were added and by 20.8% when those with a positive flag in any record over 20 years were incorporated. These results equated to underestimating Aboriginal status in unlinked index admission by 2.6%, 3.5% and 17.2%, respectively. Deaths classified as Aboriginal in official records would underestimate total Aboriginal deaths by 26.8% (95% Confidence Interval 24.1 to 29.6%).
Conclusions
Combining Aboriginal determinations in morbidity and official death records increases ascertainment of unlinked cardiovascular morbidity in Western Australian Aboriginals. Under-identification of Aboriginal status is high in death records.