Article Text
PDF
Abstract
Objectives To examine the relative differences in injury hospitalisation cost estimates from hospital administrative data when using (1) hospital reimbursement based on National Efficient Price (NEP) versus local hospital costings methods, (2) inflation factors based on Consumer Price Index (CPI) versus health group index and (3) different healthcare funder’s perspectives.
Design Retrospective population-based cohort study of linked Queensland Hospital Admitted Patient Data Collection dataset and National Hospital Cost Data Collection data.
Setting All admitted injury-related care episodes occurring within a major trauma hospital in Queensland during 1 January 2012 to 31 December 2017.
Outcome measure Total in-patient hospitalisation cost associated with an episode of care.
Results Injury without catastrophic/severe complications or comorbidities was the most frequently occurring Australian Refined Diagnosis-Related Groups (AR-DRG) over the 6-year period, while rehabilitation with catastrophic complications or comorbidities was the most expensive ($37 938, 95% CI $36 067 to $39 809). Among the top 10 AR-DRGs, seven had NEP-based cost estimates substantially higher than the hospital-reported costs, with differences varying between 2.6% and 43.0%. CPI-inflated costs were significantly lower than health group index-inflated estimates, with observed differences between 7.7% (95% CI 6.9% to 8.7%) and 11.9% (95% CI 10.8% to 13.1%) for the same AR-DRG. Finally, cost estimates were significantly higher for care funded by private health insurers compared with care funded by either the public insurer or compulsory third-party injury insurers, with differences varying significantly between 8.4% (95% CI 7.2% to 30.1%) and 55.0% (95% CI 53.3% to 56.9%) for the same AR-DRG. Care funded by compulsory third-party injury insurers, however, incurred the highest cost for the most expensive AR-DRGs.
Conclusion There were considerable discrepancies in cost estimates for common injury-related hospitalisations depending on type of costing method used, inflation metrics applied and healthcare funder’s perspective adopted. These factors need to be considered when evaluating hospital cost in Australia’s health system using administrative data.
- Health Care Costs
- HEALTH ECONOMICS
- TRAUMA MANAGEMENT
Data availability statement
Data may be obtained from a third party and are not publicly available. The data that support the findings of this study are available from each data custodian, including the Queensland Health Department, Motor Accident Insurance Commission and Queensland Office of Industrial Relations. Restrictions apply to the availability of these data, with these restrictions imposed on the research team for the current study, so are not publicly available.
This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See: http://creativecommons.org/licenses/by-nc/4.0/.
Statistics from Altmetric.com
STRENGTH AND LIMITATIONS OF THIS STUDY
This was a state-wide population-based retrospective cohort data linkage study.
This study focused on the 10 most frequently admitted injuries across the major trauma hospitals in the state of Queensland, Australia
Only public hospitals costs (and not private hospital cost) were available for the study cohort.
Introduction
Injury causes a significant burden on the Australia’s health system, with over half a million injured people hospitalised annually.1 Due to increasing pressure on health systems, robust estimation of the costs of injured persons is needed to ensure that economic evaluation of interventions and models of care that seek to improve injury outcomes are accurate.
With the rise of digital transformation in health, underpinned by the National Digital Health Strategy2 and the enactment of the Data Availability and Transparency Act 2022, which establishes a new, best practice scheme for sharing Australian Government data,3 as well as the 2020–2025 National Health Reform Agreement,4 evaluating Australia’s health system using electronic administrative data is becoming realised.
There are important considerations, however, that govern the usefulness of administrative datasets for economic evaluations due to the complex nature of health systems (eg, in Australia, although public hospitals receive the majority of funding (92%) from Commonwealth and State governments, ownership and management of public hospitals are primarily the responsibility of local Hospital and Health Service boards).5 First, public hospital care is funded predominantly through Activity-Based Funding (ABF), where similar episodes of care are funded according to the Australian Refined Diagnosis-Related Groups (AR-DRG) classification.6 The ABF model uses the National Efficient Price (NEP) as a reference point for reimbursement, where an episode of care is allocated a national weighted activity unit (NWAU) as determined by the Independent Health and Aged Care Pricing Authority (IHACPA). Hence, NEP payments are made against the NWAU that measures hospital activity expressed as a common unit.7 8 Given the scarcity of hospital-reported cost data for economic evaluation of health intervention in Australia, using the NEP provides a straightforward and easily accessible approach to estimate in-hospital cost. However, the NEP approach can lead to discrepancies between how much funding hospitals commit to specific treatments (hospital-reported cost) and how much they are reimbursed for delivery of these services.8 These discrepancies can be substantial and may have severe implications on the conclusion drawn from economic evaluations.
Inflation is another important variable that needs careful consideration when estimating healthcare cost over time because the purchasing power of a currency changes over time making it more costly to purchase the same quantity of goods and services over time.9 10 Hence, it is important to standardise all costs to a reference year to make accurate comparisons. The type of inflation metric applied (eg, the Consumer Price Index (CPI) or the health group index) can have direct impacts on the cost estimate and therefore the conclusion drawn from an economic evaluation of a health intervention.
Finally, in a mixed public-private health system like Australia, where private health insurers, Department of Veterans Affairs and compulsory third-party injury insurers (such as workers’ compensation and state-based motor vehicle third party insurers) play a role in healthcare funding,11 a recognition of the perspective of each sector is important when attributing cost using administrative data. Hospital cost for an episode of care can vary substantially depending on the funding type initiated at hospital admission (whether public, private or compensation system funder) and may have implications on cost estimates depending on the perspective of the economic evaluation. Hence, proper accounting for such differences when estimating cost from administrative health data is needed to ensure unbiased estimates of cost.
Using hospital administrative data from the state of Queensland, Australia, this study sought to examine the differences in cost estimates for injury-related hospitalisations when (1) the NEP or hospital-reported values are used, (2) different inflation metrics are used to adjust for inflation over time, and (3) different healthcare funders’ perspectives are considered.
Methods
Data sources
Data were sourced from the broader Transport and Work-Related Injury Compensation Linkage And Injury Management Study (CLAIMS),12 which is a retrospective, population-based cohort study of all persons who had an injury-related event attributable to a transport or work-related incident between January 2012 and December 2017 across Queensland, Australia. Although the CLAIMS study has a focus on transport and work-related injuries, this substudy is not restricted to only transport and work-related injuries but includes all types of injury-related hospital admissions as reported by the Queensland Hospital Admitted Patient Data Collection. NEP data were sourced from the IHACPA,13 while inflation data for the study period were sourced from Australian Bureau of Statistics.14
Study cohort
For the current study, the cohort included only public hospital episodes occurring within a person’s index (or first) hospital encounter identified within the 6-year time frame, where an encounter is defined as containing contiguous episodes of care (eg, it brings together interfacility transfers or care type changes within the same facility) for the same person.15 All episodes of care within that first injury hospital encounter were included in the analysis (including those with International Statistical Classification of Diseases and Related Health Problems, Tenth Revision, Australian Modification (ICD-10-AM) code range S00-T75 or T79), both the acute care episodes and the sub-acute/rehabilitation episodes (as assigned by the hospital) which were part of the same encounter of care. The unit of analysis, however, is an episode of care, not an encounter or patient, given AR-DRGs are assigned to each episode of care. In addition, the cohort included only episodes of care occurring within the five major trauma hospitals in Queensland,16 given some smaller hospitals operate under a different funding model and may confound the results. Figure 1 summarises the study flow chart.
Study flow chart.
Cost measure
Total in-patient hospitalisation (ie, episode) cost was the main outcome variable in this study. This is the sum of direct and overhead costs associated with an episode of care as reported by hospitals and sourced from the National Hospital Cost Data Collection database.17 In adjusting for inflation, 2022 was used as the base year and costs were reported in Australian dollars. Costs were analysed from three perspectives: health system (public insurer), private health insurer and compulsory third-party healthcare funders.
Statistical analyses
Descriptive statistics were undertaken for the top 10 injury-related AR-DRGs. Means and confidence intervals were estimated using bootstrapping technique with 1000 replications due to the non-normal cost distribution.18 Analysis of variance was used to test for differences in mean cost between groups. The NEP for each AR-DRG was estimated by first multiplying the average NEP by the NWAU for that condition for a specific year (as shown in online supplemental appendix A). After computing the NEP, the mean NEP cost over the study period (2012–2017) was estimated. Due to AR-DRG version changes over the study period, version six was used in this study as the reference version. Although the AR-DRG classification assigned to an episode of care is based on more than just a patient’s principal diagnosis,19 for ease of reporting the results, an AR-DRG will be labelled as a diagnosis. Analyses were conducted using STATA V.17.
Supplemental material
Patient and public involvement
None.
Results
Between January 2012 and December 2017, 53 346 injury-related hospital episodes of care met the study inclusion criteria. These hospital episodes were for 51 686 individuals. Of those injury-related hospital episodes, injuries without catastrophic or severe complications or comorbidities were the most frequently coded (table 1). This was followed by other head injury; poisoning/toxic effects of drugs and other substances without catastrophic or severe complications or comorbidities; injuries to the forearm, wrist, hand or foot; trauma to the skin, subcutaneous tissue and breast without catastrophic or severe complications or comorbidities; injury to the shoulder, arm, elbow, knee, leg or ankle without complications or comorbidities; hand procedures; humerus, tibia, fibula and ankle procedures without complications or comorbidities; non-surgical spinal disorders without complications or comorbidities; and rehabilitation with catastrophic complications or comorbidities. Altogether, these 10 diagnoses constituted 41% of all injury-related hospital episodes. Rehabilitation with catastrophic complications or comorbidities had the longest hospital stay of 11 days (SD=8) followed by the humerus, tibia, fibula and ankle procedures without complications or comorbidities.
Top 10 injury diagnoses among the five major trauma facilities in Queensland, Australia
There were substantial differences between the NEP and hospital reported cost estimates over the 10 most frequent injury-related hospital diagnosis (table 2). While NEP cost estimates varied between $1446 to $18 751, hospital reported costs ranged from $891 (95% CI $846 to $937) to $27 674 (95% CI 26 395 to $28 952) for the same diagnosis. Of the 10 diagnoses considered, seven had NEP costs higher than the hospital reported cost, with differences varying between 2.6% (injuries to the forearm, wrist, hand or foot) to 43.0% (trauma to the skin, subcutaneous tissue and breast without catastrophic or severe complications or comorbidities). However, hospital-reported cost estimates were substantially higher than NEP cost for the three most expensive diagnoses (hand procedures; humerus, tibia, fibula and ankle procedures without complications or comorbidities; and rehabilitation with catastrophic complications or comorbidities) with the gap varying between $1454 and $8923 per episode of care.
NEP cost versus average hospitals reported cost estimates for top ten injury diagnoses in the five major trauma facilities in Queensland
Table 3 summarises inflation-adjusted cost estimates using CPI index versus health group index. Across all top 10 diagnoses, health group index-inflated costs were significantly higher than CPI-inflated costs. CPI-inflated costs ranged from $1063 (95% CI: $1009 to $1116) to $33 899 (95% CI $32 254 to $35 545), while the health group index–inflated costs varied between $1145 (95% CI $1086 to $1205) and $37 938 (95% CI $ 36 067 to $39 809) for the same diagnosis. The lowest percentage cost difference was found for other head injury diagnosis (7.7%), while the highest percentage cost difference was observed for rehabilitation with catastrophic complications or comorbidities (11.9%).
Cost estimates using CPI inflation versus health group index for top 10 injury diagnoses in the five major trauma services in Queensland
Finally, analysis of the hospital-reported cost data shows significant cost differences across episodes of care depending on the healthcare funder. Most episodes of care that were funded by the public health insurer (public patients) incurred significantly lower cost than those funded by a private health insurer (eight out of 10 diagnoses). For these eight diagnoses, while the hospital costs for public patients varied between $1404 (95% CI $1324 to $1484) and $36 240 (95% CI $34 101 to $38 378), those for patients with private health insurance costs varied between $2242 (95% CI $1150 to $3334) and $40 521 (95% CI $35 404 to $45 638) for the same diagnosis (table 4). Although patients funded by third-party funders incurred significantly lower cost than those funded by either the public or private health insurer, they incurred the highest cost for the three most expensive diagnoses: hand procedure; humerus, tibia, fibula and ankle procedures without complications or comorbidities and rehabilitation with catastrophic complications or comorbidities. The cost difference between third-party funded patients and publicly funded patients varied between 23.3% and 39.1%, while the variation between third-party funded patients and private health insurance funded patients ranged between 10.2% and 56.2% for the same diagnoses. Online supplemental appendix B shows that among all the third-party funders, those providing workers’ compensation and third-party vehicle compensation incurred the highest cost.
Hospital cost by type of funder for top 10 injury diagnoses in the five major trauma service facilities in Queensland
Discussion
This study used real world injury-related administrative hospitalisation data to demonstrate the discrepancies that might arise when conducting population-based cost analysis. While injury without catastrophic or severe complications or comorbidities was the most frequent injury diagnosis, hand procedure; humerus, tibia, fibula and ankle procedures without complications or comorbidities and rehabilitation with catastrophic complications or comorbidities were the three most expensive injury diagnoses.
NEP cost estimates were found to be substantially higher than the hospital reported costs in most cases (seven out of 10). This suggests that using the NEP approach to estimate the cost or benefits of an intervention (such as reduced hospitalisation and length of stay) in economic evaluation can lead to biased conclusions since the cost or benefits from the intervention is likely to be overestimated. Such overestimation can vary between 2.6% and 43.0% depending on the injury-specific diagnosis being evaluated. However, among the three most expensive injury procedures (hand procedure; humerus, tibia, fibula and ankle procedures without complications or comorbidities; and rehabilitation with catastrophic complications or comorbidities), hospital-reported costs were substantially higher than the NEP estimates, suggesting that hospitals incurred more cost for treating these patients than how much they were reimbursed by the state government. The reimbursement gap varied between $1454 (hand procedure) and $8923 (rehabilitation with catastrophic complications or comorbidities) per episode of care. Using the number of hospital episodes that were related to these three diagnoses and doing a back-of-the-envelope calculation, these estimates suggest that the major trauma centres in Queensland were under-reimbursed by up to $39.9 million between 2012 and 2017 for the in-hospital treatment of injuries related to hand; for rehabilitation with catastrophic complications or comorbidities; and for injuries related to the humerus, tibia, fibula and ankle without complications or comorbidities. Although every year the IHACPA uses hospital-reported cost data as well as projections of healthcare cost inflation together with other technical adjustments to adjust the NEP and NWAUs,7 8 the acute phase of care for patients with severe injuries can be very resource intensive due to their many complex needs,20 which leads to potential gaps in hospital funding. Hence, researchers need to be cautious when using the NEP approach to estimate the cost associated with a particular health condition as that could produce biased estimate of healthcare cost.
Second, we found that there are significant variations between the CPI and health group index–inflated costs for injury patients. That is, cost estimates using the CPI inflator instead of the health group index inflator tend to underestimate the cost of hospital treatment. The difference can vary between 7.7% and 11.9% depending on the injury diagnosis being investigated. While CPI inflation reflects the average rate of change in general prices of goods and services in the economy over time, health inflation reflects the average rate of change in prices of goods and services within the health sector of the economy.14 21 Hence, CPI inflation may not accurately reflect the prices of healthcare resources specifically. For example, between 2012 and 2022, CPI inflation was between 101.0% and 127.3%, while the health group index inflation varied between 103.6% and 150.6% for the same period (see online supplemental appendix C).14 That is, during the study period, health group inflation was significantly higher than CPI inflation, which is consistent with the Australian Bureau of Statistics estimates that highlights health group inflation as the third highest contributor to price increases in the overall economy between June quarter of 2023 and June quarter of 2024 (5.7% change).22 The implication of this finding in health economic evaluation is that CPI inflation may tend to underestimate the true cost or benefit of an intervention (eg, in cost-utility analysis). Therefore, since health economic evaluation involves evaluating interventions within the health sector of the economy, it is appropriate to use the health group index inflator when adjusting for changes in the prices of health goods and services over time.
Finally, the study has shown that hospital cost for the same injury diagnosis varies significantly across patients, depending on their healthcare funder. Overall, patients funded by private health insurers tend to incur the highest hospital cost compared with patients funded by either public health insurance (Medicare) or through a third-party insurer. This finding, in part, can be explained by the extra benefits private health insurance holders enjoy while receiving treatment in a public hospital, such as access to services not covered by Medicare (eg, physiotherapy, optical and dental), choice of preferred physician, treatment scheduling and quick access to hospital procedures that are not a medical emergency.11 Intriguingly, among the three most expensive injury diagnoses (hand procedure; humerus, tibia, fibula and ankle procedures without complications or comorbidities; and rehabilitation with catastrophic complications or comorbidities), those covered by third-party insurers incurred the highest cost compared with those funded by either a private health insurer or public insurer. This finding in part may reflect the severity of injury in that although patients may have the same diagnosis code, those who sustained their injury in a transport crash or while working may be at the higher severity end (as shown in the third-party cost breakdown in online supplemental appendix B). These findings highlight the need to consider the healthcare funders’ perspective when estimating cost from hospital administrative database, because depending on the perspective of cost analysis (whether public insurer/health system, private health insurer or third-party insurer), the findings will differ. In a mixed public-private health system like Australia’s, where it is common to find patients with private health insurance who elect to use their insurance to receive treatment at a public hospital, one must be cautious when estimating the benefits or costs of an intervention using hospital administrative data.
Despite the strengths of our study, there were some limitations. First, our analysis is limited to injured patients admitted to only public hospitals and not private hospitals. Although 12% of our analysis included patients who elected to use their private health insurance at a public hospital, data on private hospital admission costs would provide greater understanding of the dynamics of cost between public and private hospital admissions for the same diagnosis. Second, since the AR-DRG classification groups all health conditions, not just acute injury, it may be that both an acute injury and non-acute injury can be assigned the same AR-DRG; however, our cohort selection has only captured episodes of care related to acute injury. This may impact the interpretation of results, considering that those with more chronic conditions may have a longer hospital stay and more complications and hence may incur more hospital costs than those with acute injury. Our findings should therefore be interpreted with caution. Moreso, our analysis excluded public hospital episodes that had no cost data, which could bias our estimates. That said, due to the comparative nature of this study, such bias would be minimal. Thirdly, since the aim of our study was to demonstrate the pros and cons of using administrative health data to estimate in-hospital care costs, we did not consider other costs associated with injuries. These other excluded costs may be related with productivity loss, absenteeism, presenteeism, general practitioner visits, medication use, care givers time and compromised health-related quality of life. Hence, our cost estimates represent a lower bound estimate of the overall burden of injury on society. Finally, the dataset used in this study is limited to 2017; hence, our cost estimates may not reflect current cost for the episodes of care examined in this study. That said, having adjusted for inflation and standardised cost to 2022 values, the extent of any biases could be minimal.
In conclusion, this study has discussed and examined some important considerations when estimating the cost of hospitalised injury using administrative health data from the Australian health system. It has demonstrated the importance of considering the variety of ways in which costs can be calculated and the different perspectives which can be taken, as these decisions have a dramatic impact on resultant cost estimates. With the increasing availability of linked electronic administrative health data through digital health reforms and given the ongoing significant burden of injury to the community, methods for estimation of costs are important considerations to enable accurate examination of the cost-effectiveness of injury prevention and treatment strategies that will improve patients’ outcomes while saving cost to the health system. While this study utilised administrative injury data as a use-case to demonstrate how different costing methods can generate differences in cost estimates, the findings are largely applicable to other health conditions or diseases. For example, the use of linked administrative health data for chronic disease studies has gained popularity in Australia in recent years.23–25 Hence, the use of such administrative health data should be guided by clearly defined methodologies and study perspective to enable accurate estimation of healthcare costs.
Data availability statement
Data may be obtained from a third party and are not publicly available. The data that support the findings of this study are available from each data custodian, including the Queensland Health Department, Motor Accident Insurance Commission and Queensland Office of Industrial Relations. Restrictions apply to the availability of these data, with these restrictions imposed on the research team for the current study, so are not publicly available.
Ethics statements
Patient consent for publication
Ethics approval
The study was approved by the Royal Brisbane and Women’s Hospital Human Research Ethics Committee (HREC/2018/QRBW/55 (39383)).
Acknowledgments
Computational (and/or data visualisation) resources and services used in this work were provided by eResearch, Research Infrastructure, Queensland University of Technology. This study benefited from the excellent data linkage work by the Statistical Services Branch and Statistical Analysis and Data Linkage Unit at Queensland Health. We would also like to acknowledge the support from the core funders of the Jamieson Trauma Institute: Motor Accident Insurance Commission, Metro North Health and Queensland University of Technology, which enable Clifford Afoakwah, Jacelle Warren, Shahera Banu, Michael Schuetz and Kirsten Vallmuur to dedicate part of their time to this study.
Footnotes
Contributors CA, JW VM and KV conceptualised and designed the study. JW and KV acquired the data. CA analysed and interpreted the data. CA drafted the manuscript. CA, JW VM, SB, MS, SMM and KV provided critical revision of the manuscript for important intellectual content. CA is the guarantor for this study. All authors reviewed and contributed to the manuscript and approved the final version.
Funding The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors.
Competing interests The Jamieson Trauma Institute received research funding support from the Motor Accident Insurance Commission (state-based regulator).
Patient and public involvement Patients and/or the public were not involved in the design, conduct, reporting or dissemination plans of this research.
Provenance and peer review Not commissioned; externally peer-reviewed.
Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.