In a cross-sectional study, we used linked administrative health and ABS data to estimate RHD-related deaths (Figure 1). We estimated observed mortality rates by age at death and Indigenous status by using data from End RHD in Australia: Study of Epidemiology (ERASE) (9). We used the generated excess deaths rates to calculate expected RHD-associated deaths and compared them with ABS RHD-coded death counts.

The ERASE cohort has been described (2). In brief, prevalent and new RHD cases were identified from the RHD register, surgical registry, and hospitalization records (1,9,11,12). ERASE included 5 jurisdictions in Australia: Northern Territory, Queensland, South Australia, Western Australia, and New South Wales (Appendix Figure 1). We obtained probabilistically linked data from jurisdiction-specific linkage units; ERASE investigators harmonized variables between jurisdictions and data sources and determined vital status.

To create the RHD study cohort, we selected ERASE cohort members who had RHD, were <65 years of age, and were alive on January 1, 2013 (Figure 1). We used broad age groups (0–24, 25–44, and 45–64 years), which corresponded to those used in previous RHD mortality studies (2,13). We used multiple ERASE data sources to assign Indigenous status, minimizing known underidentification (9). We searched all hospitalization record diagnosis fields for comorbidities (Appendix Table 1).

We calculated observed and background mortality rates (both crude and age-specific per 100,000 person-years). We calculated age-standardized mortality rates by using the direct method, standardized to World Health Organization World Standard Population 5-year age groupings for 2000–2025. For observed mortality rates (Figure 1), RHD diagnoses from January 1, 2013, through December 31, 2017, contributed person-time from whichever time was latest (denominators): first diagnosis date or January 1, 2013. Deaths during 2013–2017 contributed to observed mortality rate numerators. For background mortality rates (Figure 1), we used age group–specific deaths of Indigenous and non-Indigenous Australians (numerators) and residential population denominators from the ABS (13).

We calculated excess mortality rates as the difference between the observed and background mortality rates (within matched age/population stratum; Figure 1). We derived 95% CIs by using nonparametric bootstrap methods, assuming a Poisson distribution (Appendix). We calculated expected RHD-related deaths by applying excess mortality rates to person-years within the RHD study cohort age/population stratum (Appendix Table 2). We calculated observed and excess mortality rate ratios (MRRs) with 95% CIs by comparing Indigenous with non-Indigenous populations with RHD.

Epidemiologic, demographic, and clinical characteristics of this cohort are described (Appendix Table 3). Among the 9,342 persons in the RHD study cohort (65.6% female, 24.6% <25 years of age, 55.6% Indigenous), comorbidities included atrial fibrillation (30.5%), heart failure (26.0%), hypertension (23.7%), diabetes (19.4%), chronic kidney disease (17.4%), and chronic obstructive pulmonary disease (10.6%) (Appendix Table 3). The 726 observed cohort deaths occurred most frequently among persons 45–64 years of age (72.3%) and among those who were female (58.7%) (Appendix Table 3). Among the 325 non-Indigenous persons who died, 36.0% were immigrants from low/middle income countries. Metropolitan residents accounted for 14.0% (n = 56) of deaths among Indigenous and 71.4% (n = 232) among non-Indigenous persons. Detailed causes of death within the study cohort were attributed to mostly noncardiovascular causes; most frequent were cancer, diabetes mellitus, and respiratory diseases (2).

In 2013–2017 in Australia, the background mortality rate was 193.6 deaths/100,000 Indigenous person-years and 72.3 deaths/100,000 non-Indigenous person-years (Table). Background age-specific mortality rates increased with advancing age in both populations but were always 2- to 3-fold higher for the Indigenous than non-Indigenous population (Table, Figure 2)

In the RHD study cohort, 401 Indigenous and 325 non-Indigenous persons died, corresponding to observed mortality rates of 1,451 deaths/100,000 Indigenous person-years and 883 deaths/100,000 non-Indigenous person-years (Table). Age-specific mortality rates among Indigenous persons were highest among those 45–64 years of age (4,568 deaths/100,000 person-years; Figure 2); corresponding MRR was 2.13 (95% CI, 1.81–2.52) for Indigenous versus non-Indigenous persons (Table).

For the RHD study cohort, we estimated excess mortality rates of 1,166 deaths/100,000 Indigenous person-years and 771 deaths/100,000 non-Indigenous person-years, generating an MRR of 1.5 (Table). Excess mortality rates were highest among Indigenous persons 45–64 years of age for whom the peak excess MRR of 2.1 was observed (Table; Figure 2). Excess mortality rates applied to RHD study cohort strata estimated that 319 Indigenous and 272 non-Indigenous deaths were directly or indirectly associated with RHD (Table; Appendix Table 4). By comparison, ABS RHD-coded deaths captured 145 Indigenous deaths, less than half the expected cases (174 fewer than expected), but 300 non-Indigenous deaths, approximately the same as expected (28 more).

Accuracy of our estimates is limited by the quality of the coded information within source datasets and constrained by available data, including nonavailability of migrant population denominator information for rate calculations. The RHD mortality rates that we report also do not capture the profound effects that those deaths had on families, communities, and cultures.

After adjusting for background mortality in Indigenous and non-Indigenous populations, we found that excess deaths were higher among persons with RHD. The excess mortality method applied to the RHD study cohort estimates both direct and indirect RHD-associated deaths and reduces concerns regarding misclassified and missing cause of death arising from use of ABS RHD-coded data only. Our method is particularly useful with the Indigenous population, for whom missing ABS RHD-coded data are an issue. RHD is probably not the only underlying driver of observed excess premature deaths; rather, RHD is a potent marker of the inequities experienced by Indigenous Australians and drives excess deaths indirectly in synergy with other chronic health conditions associated with social determinants. Expected deaths among non-Indigenous persons corresponded closely to ABS RHD-coded records; however, among the Indigenous population, excess deaths were more than twice those recorded in ABS (2). Similar to other chronic illnesses (diabetes and dementia [10,14]), the burden of RHD-associated deaths in Australia is potentially underascertained when based exclusively on ABS RHD-coded records, especially among Indigenous persons, for whom cause-of-death data are missing for >10% and multiple comorbidities, along with underlying RHD, contribute to death (2). Before Australia can achieve its RHD elimination goals, improved quality of Indigenous cause-of-death data and identification of synergistic factors contributing to excess RHD-associated deaths are needed (7).