Pattern of hospital admissions and costs associated with acute rheumatic fever and rheumatic heart disease in Australia, 2012–2017

Study design

This was a retrospective linked-data analysis of hospitalisations and costs for ARF, RHD and related complications between 1 July 2012 and 30 June 2017 (the study period).

Data from the End-RHD in Australia: Study of Epidemiology (ERASE) was used, which included RHD register, adult cardiac surgery, emergency presentation, inpatient admission and death records for people with ARF/RHD.¹² Data were available from Northern Territory (NT), Queensland (Qld), Western Australia (WA), South Australia (SA) and New South Wales (NSW) for 1 January 2001 to 31 December 2017. Victoria, Tasmania and Australian Capital Territory were excluded due to lack of RHD registers.

Study sample

People aged <65 years at 1 July 2012 with any register/hospital record of ARF/RHD in the previous 11.5 years, or newly diagnosed during the study period were included. Individuals with congenital abnormalities were excluded (Supplementary material file S1).

Earliest ARF/RHD diagnosis date across all records was determined per person. For register records, diagnosis date was symptom onset date; for hospital records, it was earliest admission date for diagnosis of ARF/RHD using International Classification of Disease 10 Australian Modification (ICD-10 AM) codes I00–I02 (ARF, principal diagnosis) and I05–I09 (RHD, all diagnoses). For RHD hospitalisations, an ERASE algorithm with high positive predictive rate for RHD reduced misclassification.^13,14

Hospitalisation for ARF, RHD or complication during the study period (on or after ARF/RHD diagnosis date) were analysed for the cohort. Hospitalisations were included for a given financial year if individuals who had ARF/RHD were alive at the midpoint. For analyses by residential location, inter-hospital transfers were considered part of a single episode of care. Only residents of the same state where data were captured were included to ensure complete data coverage, except for SA and NT which had inter-jurisdictional linkage.

Demographics and admission characteristics

Study variables included financial year of hospitalisation, diagnosis (ARF, RHD or complication), emergency admission status (inpatient admissions commencing with an emergency presentation), inpatient surgical procedure, sex, population group (Indigenous, immigrant from low or middle income country, other Australian), state of residence, geographical remoteness (based on Accessibility/Remoteness Index of Australia, ARIA+), socioeconomic status (index of relative socioeconomic disadvantage quintiles, IRDS) and RHD register notification status.

Person-level Indigenous status was assigned using the algorithm developed by the WA-based Getting our Story Right project and applied across all jurisdictions and data sets to avoid over or under-counting persons identifying as Indigenous.¹⁵

Admissions coded to ARF/RHD were identified as described in ‘Study sample’. Complications including atrial fibrillation (I48), heart failure (I50), stroke (I60–I64, I69) and infective endocarditis (I33) were identified from the principal diagnosis field. Admissions involving valvular surgeries were identified via ICD-10 AM procedure code in any field (Supplementary material file S1).

Hospitalisation patterns

Hospitalisations were aggregated by financial year of admission and stratified into paediatric (age <16 years) and adult (age 16–64 years) groups. Total persons represented by these admissions, diagnosis, emergency admission status, RHD-related surgery within admission and hospital bed days were calculated. For location-based analyses, grouping of admissions into episodes allowed for investigation of treatment outside of residential region and air transfers. Air transfers were identified using mode of admission variables from both emergency and hospitalisation datasets.

Hospitalisation costs

Hospitalisation costs were generated using ‘National Weighted Activity Unit’ calculators obtained from the Independent Health and Aged Care Pricing Authority, which estimate costs based on Australian Refined Diagnosis Related Group (AR-DRG) codes (Supplementary material file S2).¹⁶ Costs were indexed to 2016/17 Australian Dollars (AU$) and presented by financial year of admission, diagnosis, emergency admission status and jurisdiction. Sensitivity analyses were conducted to assess the implications of missing interstate surgeries (Supplementary material file S3).

Statistical analyses

Linear regression models with financial year as the predictor variable were applied to total cases, admissions, surgeries, hospital bed days and costs. Regression coefficients and 95% confidence intervals (CIs) around the financial year variable were interpreted as annual average change during the study. Z-tests were used to compare hospitalisation patterns between metropolitan and non-metropolitan residents.

Data management and cost calculations were conducted in SAS 9.4; modelling and graphics used R 4.1.1.

Ethics

Aboriginal Ethics Committees approved ERASE in WA (No. 717), SA (No. 04-16-700) and NSW (No. 1363/18). Human Research Ethics Committee approvals were also obtained from Menzies School of Health Research (No. 2016–2705), WA Health (No. 2016/29), SA Health (No. HREC/16/SAH/120), Queensland Health (No. HREC/15/QPCH/289) and NSW Health (No. HREC/18/CIPHS/15).

Baseline descriptive statistics

During the study, 36.9% (n = 791) of 2142 paediatric patients had one or more ARF/RHD-related hospitalisations, this was true for 24.2% (n = 2761) of 11,387 adult patients (Table 1). Hospitalised children were predominantly Indigenous (85.5%), >60% resided in remote/very-remote regions and >60% were from the most disadvantaged SEIFA category. Most hospitalisations were new cases, with 55.2% not on the RHD register. Over one-half of adults hospitalised were non-Indigenous and resided in major cities/inner regional areas, and only 7.6% were registered.

Hospitalisation patterns and temporal trends

The number of children hospitalised for ARF/RHD-related treatment increased 9.3% annually (95% CI: 4.7–13.0%, P < 0.001), from 191 in 2012/13 to 262 in 2016/17 (Fig. 1). On average there were 296 paediatric hospitalisations per year, this increased 6.2% annually (95% CI: 2.4–9.6, P = 0.001) and average bed days/child increased 5.1% (95% CI: 3.7–6.3, P < 0.001, Fig. 1, Supplementary material Table S1). Each admitted child was hospitalised an average 1.2 times for 10.6 bed days per year (Table S1).

Fig. 1.

Paediatric and adult inpatient hospital utilisation for ARF, RHD and related complications between 2012/13 and 2016/17.

The mean number of adults hospitalised was unchanged at 1442 annually; however, admission numbers increased by 1.7% annually (95% CI: 0.1–3.4%, P = 0.03) and bed days increased by 3.5% (95% CI: 2.9–4.0%, P < 0.001, Fig. 1, Table S1). Each admitted adult was hospitalised an average 1.5 times for 13.4 bed days per year (Table S1).

Frequencies and rates of RHD-related surgery were unchanged over the study period (Fig. 1, Table S1). For adults, 154 valvular surgeries were conducted interstate during the study period, with 147 NT residents having surgery in SA.

Differences by residential location

Almost 90% of paediatric and 70% of adult episodes were publicly funded in metropolitan areas; this proportion was higher in non-metropolitan areas (97.7% paediatric, 88.6% adult, Table 2). Episodes initiated via emergency admission were more frequent for non-metropolitan than metropolitan residents (79.2% vs 50.0% paediatric; 56.7% vs 40.7% adults, both P < 0.001). Non-metropolitan residents received proportionally fewer surgeries than metropolitan residents (1.8% vs 15.8% paediatric; 15.0% vs 25.3% adults; both P < 0.001). Among non-metropolitan residents, one-third or more were hospitalised outside of their residential region. Almost one-fifth of paediatric episodes involved air ambulance transfers; among adults this was recorded for 8.0% of episodes.

Pattern and trends in hospital costs

Costs of treating ARF/RHD totalled AU$130.6 m over 5 years, with AU$17.6 m for paediatric and AU$113.0 m for adult hospitalisations, and 72.2% of all hospitalisation costs (AU$81.6 m) were associated with emergency admissions (Fig. 2, Table S1). The average cost per admission was AU$11,872 per child and AU$15,670 per adult (Fig. 2, Table S1).

Fig. 2.

Total paediatric and adult inpatient costs for ARF, RHD and associated complications for 2012/13 to 2016/17 (a) paediatric patients <16 years and (b) adult patients aged 16–64 years. Subset of admission costs by diagnosis, procedure and emergency admission status are shown (non-additive).

Emergency admissions accounted for 79.0% of paediatric hospitalisation costs (AU$13.9 m of AU$17.6 m); admissions principally for ARF were 71.6% of costs (AU$12.6 m, Fig. 2, Table S1). Total paediatric costs were unchanged annually, however, the subset of costs for emergency admissions increased 11.8% annually (95% CI: 0.7–24.0%, P = 0.04). Costs for RHD admissions decreased 16.0% annually (95% CI: 2.9–27.4%, P = 0.03) (Table S1). Sensitivity analysis identified 17 valvular surgeries among NT children that were excluded from the primary analysis, representing AU$1.3 m additional costs (Supplementary material file S4). Highest paediatric ARF/RHD-related costs were incurred in NT (AU$7.6 m) and Qld (AU$5.2 m). Adult admissions for ARF/RHD incurred costs of AU$113.0 m; 60.0% (AU$67.7 m) of these costs were via emergency presentation and 69.8% (AU$78.9 m) involved valvular surgery (Fig. 2, Table S1). Adult hospitalisation costs did not change annually. The highest adult ARF/RHD-related costs were incurred in NSW (AU$33.3 m) and Qld (AU$30.7 m, Table S1).