Preventable hospitalisations in regional Queensland: potential for primary health?

Additional keywords: Aboriginal and Torres Strait Islander, chronic disease, health economics, potentially preventable hospitalisation, primary health care.

Regional and remote hospitals have high rates of admission for conditions that should have been managed with timely access to quality primary and community health services.^1,2 Australia’s National Healthcare Agreement defines these admissions as potentially preventable hospitalisations (PPHs).³ Although debate remains around PPH definitions,⁴ the Australian Institute of Health and Welfare (AIHW) routinely collects and analyses PPH data (sourced from hospital separation diagnosis and procedure codes) as a component of the National Health Performance Framework to monitor the quality and effectiveness of these services.

The PPH data are broadly classified into three types of conditions: vaccine preventable, chronic health and acute medical. The most recent AIHW report for 2014–15 found 493 347 PPHs were recorded in Australian public hospitals, accounting for 8.2% of all separations.⁵ Rates per 1000 population remain incrementally higher in regional (inner = 25.7; outer = 29.3) and remote areas (remote = 39.8; very remote = 59.1) than in major cities (23.7).⁵ The reason people living in regional and remote areas have higher PPH rates is a complex issue, but three factors are likely to contribute significantly: (1) higher levels of socioeconomic disadvantage;^6,7 (2) difficulty accessing health services;^8,9 and (3) a higher burden of disease and comorbidities.^10–14

Since 2006, the Australian Health Ministers’ Advisory Council has produced biennial reports against the Aboriginal and Torres Strait Islander Health Performance Framework. Over this time period, results have remained fairly consistent despite various improvements to coding rules. The most recent of these reports¹⁵ found the rate of PPH for Aboriginal and Torres Strait Islander people is threefold higher than for non-Indigenous Australians, and this disparity generally increases with age and remoteness. The report indicates these rates likely reflect higher underlying disease burden, limited access to early detection and management of common chronic conditions and limited non-hospital alternatives in rural and remote areas.¹⁵ This is supported, in part, by a recent Northern Territory study of 14 184 people living in remote Indigenous communities over the period 2002–11,¹⁶ which found that higher primary healthcare utilisation was associated with a 60–85% reduction in common chronic PPH, and every AU$1 invested in primary health could produce savings of AU$3.95–11.75 in hospital costs, in addition to other health benefits for individuals.

The relative paucity of primary and other community health services in regional and remote areas contributes significantly to high PPH rates, with local hospitals becoming the default service provider.¹⁷ For example, factors such as adequate physician supply, long-term physician–patient relationships and general practitioner (GP) management plans have all been found to be important for reducing PPHs.¹⁸ A recent report from the Grattan Institute identified northern Queensland as a ‘hot spot’ for PPHs, with rates at least 50% higher than the state average in every year for a decade.¹⁹ The authors of that report recommend that geographical areas such as these represent persistent and costly health inequality, and should be prioritised for state and federal government investment to improve primary and other health services. The authors also recommend using local data to assist in identifying the health priorities of local communities and in developing, testing and evaluating tailored interventions over time.¹⁹ The federal government is currently trialling the introduction of ‘health care homes’ throughout Australia,²⁰ the purpose of which is to strengthen the role of general practices and Aboriginal Community Controlled Health Services (ACCHSs) as the central care provider (rather than the hospital) by providing continuity of care for people with chronic and complex conditions. In addition to improvements in access, care processes and patient experience, it is hoped that this initiative will reduce unnecessary hospitalisations, including PPHs.

The purpose of the present study was to use local admissions data from a large regional tertiary care hospital in far north Queensland to examine PPHs (and estimated costs) as a proportion of total hospital separations and by Indigenous status. We also identify opportunities for local primary health services to reduce hospitalisations and discuss issues related to strategic intervention.

Located in northern Queensland, Australia, the large public hospital provides tertiary care for over 275 000 people. Although the hospital services people living in very remote areas, it primarily caters for the large regional population in the surrounding 10 Statistical Local Areas (SLAs). Approximately 68 867 hospital separations were recorded between 1 July 2012 and 30 June 2014. Of these, 74.2% (51 087) were generated by people living in the 10 SLAs, and the results of the present study are based on these separations. The 10 SLAs ‘map’ to two larger Local Government Area (LGA) boundaries (Cairns (LGA/32070) and Yarrabah (LGA37600)). The Far North Queensland Human Research Ethics Committee approved the present study (HREC/13/QCH/131-880) and data were provided by Queensland Health.

Factors of interest were determined from raw Queensland Hospital Admitted Patient Data Collection (QHAPDC) data, including gender, age, Indigenous status, SLA, medical or unit record number (URN), length of stay (LOS), admission and discharge dates and final primary and secondary diagnoses. Socioeconomic disadvantage for each SLA was measured using the Index of Relative Socioeconomic Disadvantage (IRSD) calculated by the Australian Bureau of Statistics²¹ and based on 2011 Census data (http://www.abs.gov.au/AUSSTATS/abs@.nsf/DetailsPage/2033.0.55.0012011?OpenDocument, accessed 4 February 2018). The IRSD was used because it provides a broad measure of disadvantage by ranking areas according to a range of social and economic conditions reported at the individual and household level within each SLA. Some of the factors of interest when measuring IRSD include income, educational attainment, unemployment and unskilled occupations. Lower IRSD scores indicate areas with greater disadvantage and higher scores indicate areas with least disadvantage.

PPHs were identified from discharge diagnosis codes using International Statistical Classification of Diseases and Related Health Problems 10th Revision Australian Modification (ICD-10-AM)²² and relevant procedure codes using National Healthcare Agreement definitions.²³

Costs associated with PPHs were estimated using Australian-refined diagnosis-related groups (AR-DRG). The AR-DRGs are sourced from the AIHW National Hospital Morbidity Database (NHMD) with diagnostic groupings based upon ICD-10-AM classifications. According to the AIHW, AR-DRG is an Australian admitted patient classification system that provides a clinically meaningful way of relating patient casemix to resources required by the hospital.²⁴ National average costs for each AR-DRG were provided by the Independent Hospital Pricing Authority (IHPA), including both total cost and disaggregated into main procedures and average LOS. This study selected costs relating to the financial year 2013–14 in order to align with the hospital’s admission data.²⁵ These national average total costs for AR-DRGs were then converted into an ‘average per day cost’ by dividing the total cost by average LOS. To estimate hospital costs, this average per day cost was multiplied by the actual LOS taken from hospital records for each acute PPH patient admission. Total costs for each PPH and median patient cost, driven by LOS, were estimated.

Analyses were performed with STATA 13.1 (StataCorp, College Station, TX, USA). Records were linked to individuals using the unique URN and the number of separations per person was calculated as the number of separations per URN.

Crude rates per 100 000 population were calculated using two financial years of hospital separation data combined (1 July 2012–30 June 2014) and two financial years of population data for the same period. A 2-year period of hospitalisations was selected to reduce the effect of annual variability in separation rates. The population data were the estimated residential population sourced from the Queensland Regional Database maintained by the Queensland Government Statistician’s Office (http://www.qgso.qld.gov.au/products/tables/qld-regional-database/index.php; accessed 15 April 2015). Data were extracted in 5-year age groups by Indigenous status for Cairns (LGA/32070) and Yarrabah (LGA37600).

Age distributions for Indigenous and non-Indigenous separations were dissimilar. To enable comparisons of separation rates between groups, crude rates were directly age standardised²⁶ to the 30 June 2001 Australian standard population maintained by Australian Bureau of Statistics.²⁷

Each hospitalisation in the dataset had an allocated diagnosis-related group (DRG). To estimate the cost of PPHs, each hospitalisation was allocated a monetary figure based on the corresponding DRG. These amounts were then aggregated based on PPH categories and the total and median costs were calculated by Indigenous status.

In the two financial years 2012–14, there were approximately 51 087 separations generated by 29 485 people (Table 1). Of these separations, 10.7% (5488) were classified as potentially preventable. Half of all PPHs were for women (51.7%) and approximately one-third were for those aged ≥65 years (33.1%). Indigenous people represented 25.9% of all PPHs despite accounting for only 12.9% of the Cairns and Yarrabah LGA populations. Potentially preventable hospitalisation rates were substantially higher for Indigenous people across 5-year age groups, particularly in older age groups (Fig. 1).

Table 1.  Characteristics of 51 087 hospital separations, according to potentially preventable hospitalisations (PPHs) from 1 July 2012 to 30 June 2014, local residents only
Data are given as n (%). IRSD, Index of Relative Socioeconomic Disadvantage

Fig. 1.  Potentially preventable hospitalisation (PPH) rates for 5488 hospital separations, by Indigenous status and 5-year age groups, from 1 July 2012 to 30 June 2014, local residents only.

Most PPHs were for chronic (45.3%) and acute (51.8%) conditions (Table 2). Most chronic conditions (93.2%) were for chronic obstructive pulmonary disease (COPD; n = 640; 25.7%), diabetes complications (n = 432; 17.4%), congestive cardiac failure (CCF; n = 375; 15.1%), angina (n = 322; 13.0%), iron-deficiency anaemia (n = 289; 11.6%) and asthma (n = 258l; 10.4%). Most acute conditions (95.8%) were for urinary tract infections (UTI; n = 976; 34.3%), convulsions (n = 575; 20.2%), ear, nose, throat (ENT) infections (n = 438; 15.4%), dental conditions (n = 353; 12.4%), cellulitis (n = 257; 9.0%) and gangrene (n = 126; 4.4%).

Table 2.  Potentially preventable hospitalisations (PPHs) according to Indigenous status from 1 July 2012 to 30 June 2014, local residents only
Crude rates per 100 000 population were calculated using two financial years of hospital separation data combined (1 July 2012–30 June 2014) and two financial years of population data for the same time period. The population data were sourced from the Queensland Regional Database maintained by the Queensland Government Statistician’s Office. Data were extracted in 5-year age groups by Indigenous status. COPD, chronic obstructive pulmonary disease; CCF, congestive cardiac failure; UTI, urinary tract infections; ENT, ear, nose and throat; CI, confidence interval

The top four conditions for Indigenous people were UTIs (n = 201; 14.2%), convulsions (n = 182; 12.8%), diabetes complications (n = 178; 12.5%) and COPD (n = 156; 11.0%). The top four conditions for non-Indigenous people were UTIs (n = 775; 19.1%), COPD (n = 484; 11.9%), convulsions (n = 393; 9.7%) and ENT infections (n = 355; 8.7%).

Age-standardised PPH rates were higher for Indigenous people across all PPH categories (Table 2). Overall, Indigenous people were 3.5-fold more likely to have a PPH than non-Indigenous people (age-standardised rate 3.4; 95% confidence interval (CI) 3.2–3.6). Age-standardised rates for Indigenous people were eightfold higher for vaccine-preventable PPHs (7.8; 95% CI 5.6–10.6)], fourfold higher for chronic PPH (4.1; 95% CI 3.7–4.5), almost threefold higher for acute PPH (2.7; 95% CI 2.4–3.0) and nearly eightfold higher for diabetes complications (7.7; 95% CI 6.2–9.4).

Acute conditions were highest among younger age groups and generally decreased with age, whereas chronic conditions were lowest among younger age groups and increased with age. Vaccine-preventable conditions were fairly uniform across age groups. The pattern of PPH distribution across the lifespan was similar by Indigenous status except that the intersection point where chronic conditions start to outnumber acute conditions occurred one decade earlier for Indigenous people (41–50 years) than non-Indigenous people (51–60 years; Fig. 2). Diabetes complications appeared regularly in the top four PPH conditions across the lifespan for Indigenous people and COPD appeared at a younger age (Table 3).

Fig. 2.  Proportion of preventable hospitalisation (PPH) types according to age group in (a) Indigenous and (b) non-Indigenous people from 1 July 2012 to 30 June 2014.

Table 3.  Most common potentially preventable hospitalisations, by age group and Indigenous status from 1 July 2012 to 30 June 2014, local residents only
Percentages are based on age group totals. COPD, chronic obstructive pulmonary disease; CCF, congestive cardiac failure; ENT, ear, nose and throat

Total PPH represented 5488 separations and an estimated AU$32.7 million, which was 10.7% of total hospital separation expenditure over the 2-year period (Table 4). The top three most expensive PPHs, in terms of total cost for all separations, were UTI (AU$4.9 million), COPD (AU$4.9 million) and diabetes (AU$4.5 million). Associated median costs (and median LOS) per separation types were: AU$3542 (1 day) for UTI, AU$5627 (3 days) for COPD and AU$6332 (3 days) for diabetes.

Table 4.  Potentially preventable hospitalisations (PPHs), according to diagnosis-related group costs and Indigenous status from 1 July 2012 to 30 June 2014, local residents only
Median costs refer to median costs per episode. LOS, length of stay; UTI, urinary tract infection; PID, pelvic inflammatory disease; COPD, chronic obstructive pulmonary disease; CCF, congestive cardiac failure; ENT, ear, nose and throat

Total PPHs in the non-Indigenous population represented 4068 separations and AU$23.8 million, which was 9.7% of total hospital costs for this group. The top three most expensive PPHs, in terms of total cost for all separations, were UTI ($4.0 million), COPD ($3.8 million) and CCF ($2.4 million).

Total PPHs in the Indigenous population represented 1420 separations and AU$8.9 million, which was 14.9% of total hospital costs for this group. The top three most expensive PPHs, in terms of total cost for all separations, were diabetes ($2.2 million), COPD ($1.1 million) and UTI ($1 million). Overall, the median cost per separation type tended to be slightly higher for Indigenous people, with diabetes the most notable with a median cost (LOS) of AU$11 546 (3 days) for Indigenous people compared with AU$4887 (2 days) for non-Indigenous people.

Chronic and acute conditions accounted for 97% of all PPHs to this large public hospital in regional north Queensland. Most chronic conditions were for COPD, diabetes complications, CCF, angina, iron-deficiency anaemia and asthma. Most acute PPH conditions were for UTIs, convulsions, ENT infections, dental conditions, cellulitis and gangrene. Although UTIs, dental conditions, cellulitis and gangrene present acutely, they often reflect a chronic underlying condition, such as diabetes. If this was taken into consideration, chronic conditions would account for 76.5% of all PPHs. Age-standardised rates for Indigenous people were consistently higher than those for non-Indigenous people, with diabetes complications appearing regularly in the top four PPH conditions across the lifespan and chronic conditions such as COPD appearing at a younger age. The financial costs associated with PPH were substantially higher for Indigenous people, and justify investment in strategic, collaborative, evidence-based primary health interventions aimed at addressing health inequalities experienced in northern Queensland.

PPH rates were found to be consistently higher for Indigenous people across the lifespan. These findings are consistent with a recent national report against the Aboriginal and Torres Strait Islander Health Performance Framework²⁸ that found these disparities also generally increased with remoteness. A recent Northern Territory study²⁹ found PPH rates per 100 000 population were 11 090 for Aboriginal people and 2779 for non-Aboriginal people for the 8-year period from 1998–99 to 2005–06. These rates are somewhat higher than those observed in the present study, but the Northern Territory study was based on hospitalisation data for the entire Northern Territory covering large areas considered to be remote or very remote, and remoteness has been shown to be consistently associated with higher PPH rates.⁵ We attempted to remove these geographical differences by restricting the present study to only those people considered local residents surrounding the hospital. Although we observed relatively low rates of vaccine-preventable PPHs in our population, the rate of these PPHs for Indigenous people was nearly eightfold higher than for non-Indigenous people. This is consistent with national reports.⁵ Aboriginal Medical Services and other health services in the area have invested considerable effort in recent years to improve historically low vaccination rates. In addition to existing clinic-based vaccination, services have engaged in strategies including education programs, outreach home visits and targeting of priority groups within the community.

Aspects of primary healthcare that have been found to be important for reducing PPH include adequate physician supply, continuity of care, GP management plans, subsidised community health services and higher utilisation.^2,16,30 A recent preliminary report released for consultation on primary care services funded through the Medicare Benefits Schedule found that funding for urgent after-hours services in Queensland is far less in regional and remote areas like Far North Queensland than urban areas such as Brisbane.³¹ Although there appears to be broad agreement that these types of services may increase access to primary healthcare, their role in reducing PPH is currently undetermined. Indeed, considerable concern exists among professional groups that these services may undermine the quality and safety of primary health through the provision of costly and fragmented care often by underqualified medical practitioners.^31–33

Four recent publications have reviewed available evidence regarding primary health interventions and PPH, most of which has been focused on chronic disease-specific approaches (e.g. diabetes and COPD).^2,18,30,34 Overall, evidence for specific interventions to reduce PPHs is weak. Australian initiatives show promise rather than conclusive evidence.³⁰ The international review of 49 reports from high-income countries concluded that strong primary care (adequate primary care physician supply and continuity of care embodied in the ‘medical home’ approach, including self-management support as described in the Wagner Chronic Care Model³⁵) reduces avoidable hospitalisations for chronic conditions.¹⁸ The authors of that review also found that there is currently a lack of evidence for other interventions, such as disease-specific management programs. Although there are numerous examples of care coordination programs in Australia, their effect on reducing PPH rates is inconclusive and requires further investigation.² A recent report from the Grattan Institute identifying ‘hot spots’ with persistently high PPH rates recommends developing, testing and evaluating tailored interventions for local areas to address the health inequities that lead to higher PPH rates.¹⁹ The authors of that report include a collaborative approach in their strategy, whereby primary health networks are supported financially and otherwise by federal and state governments to lead targeted intervention research with primary and other health services within their jurisdiction. However, the federal government currently funds primary healthcare and the state governments fund hospital care. Consequently, there are likely to be considerable financial challenges ahead in achieving the strategic development (and sustained financing) of the innovative interventions required.

The present study has identified opportunities to reduce PPHs in this large regional hospital in northern Queensland. We found that 76.5% of all PPHs were either chronic conditions (COPD, diabetes complications, CCF, angina, iron-deficiency anaemia and asthma) or likely related acute conditions such as UTIs, dental conditions, cellulitis and gangrene. As such, we suggest targeting these specific conditions because they are likely to be conducive to trials involving collaborative primary healthcare interventions and potentially deliver the most cost savings and individual benefit. One-third of all PPHs in the present study were for people aged ≥65 years. Targeting programs at high-risk age groups is therefore justified. Several promising projects are currently underway in the region aiming to improve the integration of care for elderly people living with complex conditions. The local region is not without other strategies to address conditions favourable to primary healthcare management, although historically many of these strategies have been implemented without proper consideration for effective evaluation. The local public hospital currently provides a suite of services that have been shown to reduce PPH for specific conditions,¹⁹ including a specialist diabetes out-patient clinic, physiotherapy and rehabilitation clinics for cardiac and pulmonary conditions and nurse-led case management of people with complex health needs. Initiatives such as the integrated electronic medical record and telehealth have also been evolving to enhance access, communication and continuity of care between health services. However, the results of the present study indicate that more could be done, especially for Aboriginal and Torres Strait Islander people living in the region.

Initiatives to reduce PPHs specifically for Indigenous people are still rare.³⁰ Much of the work is performed by the local Aboriginal Medical Services using Medicare Item 715 as remuneration for annual health checks.³⁶ These are designed to promote early detection, diagnosis and management of many conditions associated with PPHs. There are also specific remuneration arrangements for GP management plans for chronic diseases. Aboriginal Medical Services also initiate programs targeting areas of high community need, although these are often challenging for the organisation to fund and resource. These initiatives may target specific health issues such as influenza vaccination, diabetes, oral health and sexually transmissible infections, or be focused on broader upstream issues such as maternal and child health, social and emotional well-being and alcohol and drug support. Future success in lowering PPH rates is likely to involve collaborative design and implementation of multifactorial programs targeting the common conditions and high-risk subgroups with complex needs identified in the present study. Programs should include: (1) early detection, early treatment and symptom management; (2) creation of supportive environments; (3) self-management support; (4) service delivery and coordination; (5) local liveability; (6) socioeconomic opportunity;^37,38 and (7) respectful, community-led collaboration.³⁹

Key to lowering PPH rates in northern Queensland is likely to be further development of primary healthcare services, including Aboriginal Medical Services. A promising reform in this area is the federal government’s introduction of ‘health care homes’ throughout Australia. Clinicians from participating primary healthcare services will become responsible for the overall coordination of chronic disease management of their patients, regardless of where the treatment occurs.^20,40 In exchange for this, practices will receive an annual complexity-based payment of between AU$591 and AU$1795 per patient, and a single grant to support training and establishment.²⁰ Whether this model will work in rural and remote areas is yet to be seen. Evidence from implementation studies of the Medical Home in the US has demonstrated some improvement in preventative care service delivery for people with chronic conditions, although results are mixed and insufficient for service provision (including accessibility and coordination of care), clinical outcomes, in-patient utilisation⁴¹ and patient experience.⁴² Diversity in the primary care setting has challenged implementation of the Medical Home care model, indicating a need for setting-specific approaches to address the healthcare needs of distinctive population subgroups.

The present study has numerous strengths and also limitations. Analysis was based on a 2-year QHAPDC dataset for the major regional public hospital servicing a discrete area in northern Queensland. The dataset was complete and analysis was carefully restricted to local residents living in 10 SLAs surrounding the hospital. However, the results are subject to error inherent in any secondary administrative dataset. For example, errors in defining the local population within SLAs immediately around the hospital may have occurred from incorrect information collected relating to residential postcode and suburb. Similarly, PPHs used in our analyses were defined using National Healthcare Agreement definitions.²³ These include relevant ICD-10-AM discharge diagnosis and procedure codes, both of which have been shown to have limited accuracy.^43,44

Some controversy exists as to whether current PPH conditions truly reflect hospitalisations that are potentially preventable. As such, PPH definitions are frequently revised and are recommended to be used as performance indicators only.⁴ In addition, accuracy in identifying Indigenous status in Australian public hospitals has been shown to vary by geographical location, with only 93% of Indigenous and 98% of non-Indigenous separations correctly identified in outer regional settings such as ours.⁴⁵ Consequently, the true number of Indigenous separations in our dataset and age-standardised estimates are likely to be higher than reported here. The costing analysis was essentially a ‘top-down’ approach using national average costs for AR-DRG and applying to the hospital by adjusting for actual LOS. This is appropriate for analysis at a population level. However, the costs are approximations. A more refined method would be, for example, a microcosting approach to directly access the hospital’s medical and financial records to cost each acute admission. However, this approach was not practicable, nor deemed necessary, for this population-level analysis.

The results of the present study suggest that opportunities exist to improve local hospital and primary health services to reduce hospitalisation for common preventable conditions in regional north Queensland. In particular, community-level health services need to be more responsive to the needs of local Indigenous families. Future research should be visionary and committed to improvements over the long term. Initiatives need to involve formal coordinated collaboration between primary health networks, Queensland Health and ACCHSs supported by secure funding and other resources from federal and state governments. Trialling and prospective evaluation of these initiatives would provide timely evidence regarding efficiency and cost-effectiveness for both local and national use.

The authors declare they have no competing interests.