Differences in health care spending and utilization among older frail adults in high‐income countries: ICCONIC hip fracture persona

Abstract Objective This study explores differences in spending and utilization of health care services for an older person with frailty before and after a hip fracture. Data Sources We used individual‐level patient data from five care settings. Study Design We compared utilization and spending of an older person aged older than 65 years for 365 days before and after a hip fracture across 11 countries and five domains of care as follows: acute hospital care, primary care, outpatient specialty care, post–acute rehabilitative care, and outpatient drugs. Utilization and spending were age and sex standardized.. Data Collection/Extraction Methods The data were compiled by the International Collaborative on Costs, Outcomes, and Needs in Care (ICCONIC) across 11 countries as follows: Australia, Canada, England, France, Germany, the Netherlands, New Zealand, Spain, Sweden, Switzerland, and the United States. Principal Findings The sample ranged from 1859 patients in Spain to 42,849 in France. Mean age ranged from 81.2 in Switzerland to 84.7 in Australia. The majority of patients across countries were female. Relative to other countries, the United States had the lowest inpatient length of stay (11.3), but the highest number of days were spent in post–acute care rehab (100.7) and, on average, had more visits to specialist providers (6.8 per year) than primary care providers (4.0 per year). Across almost all sectors, the United States spent more per person than other countries per unit ($13,622 per hospitalization, $233 per primary care visit, $386 per MD specialist visit). Patients also had high expenditures in the year prior to the hip fracture, mostly concentrated in the inpatient setting. Conclusion Across 11 high‐income countries, there is substantial variation in health care spending and utilization for an older person with frailty, both before and after a hip fracture. The United States is the most expensive country due to high prices and above average utilization of post–acute rehab care.

Patients also had high expenditures in the year prior to the hip fracture, mostly concentrated in the inpatient setting.
Conclusion: Across 11 high-income countries, there is substantial variation in health care spending and utilization for an older person with frailty, both before and after a hip fracture. The United States is the most expensive country due to high prices and above average utilization of post-acute rehab care. What is known on this topic • Health systems spend different amounts caring for patients.
• Older persons with frailty are more likely to incur high levels of spending as compared to other older populations.
• International comparisons of health systems mostly focus on the inpatient setting

What this study adds
• This study compares health care utilization and spending across 11 high-income countries for an older adult with frailty across five domains of care, including acute hospital care, primary care, outpatient specialty care, post-acute rehabilitative care, and outpatient drugs.
• The United States is the most expensive country due to high prices and above average utilization of post-acute rehab care.
• Across 11 high-income countries, there is substantial variation in health care spending and utilization for an older person with frailty, both prior to and after a hip fracture.

| INTRODUCTION
A key challenge faced by many health systems is how to best design services to provide care to a small number of high-need high-cost (HNHC) patients. One important group of HNHC patients is older adults with frailty. Frail older adults are weak, often have multiple complex medical needs, and often require assistance for daily activities (such as dressing, eating, toileting, mobility, etc.). Frailty is a strong predictor of poor clinical outcomes. [1][2][3][4] In addition, the frail population is much more likely to incur high levels of spending as compared to other older populations, including higher levels of potentially modifiable spending related to avoidable hospitalizations. [5][6][7] As the world population ages, and we see trends of increased longevity in older people, the incidence of frailty is expected to rise. Therefore, it is critical for health systems to identify ways to optimize care their care. One way to do this is by examining how care patterns for older patients with frailty vary across systems and, importantly, understanding how best practices can be applied from one health system to another.
A reliable marker of frailty among older adults is hip fracture, 8 which accounts for the majority of fractures related to fragility globally. 9 By 2050, the annual incidence of hip fracture worldwide is expected to rise over 6 million. 10 Hip fracture is also highly associated with physical and mental disability, high mortality, and increased costs, thus requiring considerable health care resources from different parts of the health system. [11][12][13][14] As hip fractures almost always require a hospital admission and usually require surgery, the vast majority will be recorded in hospital admissions data and can thus serve as a robust and reliable tracer condition to explore differences in resource use across health systems. 8 As part of the International Collaborative on Costs, Outcomes and Needs in Care (ICCONIC), we explored cross-national variations in care trajectories and resource use for frail elders across health systems in 11 countries, which have different models of health care provision and reimbursement as follows: Australia, Canada, England, France, Germany, the Netherlands, New Zealand, Spain, Sweden, Switzerland, and the United States. We made use of hip fracture in patients older than age 65 years as a tracer condition for frailty in order to identify a comparable set of patients across the 11 health systems. Making use of patient-level datasets linked across multiple care settings-spanning primary care, specialty services, acute hospital care, and post-acute care-we explored the variations in utilization and costs of health services across care settings and health systems in the 365 days before and after a hip fracture. Our study focuses on the following three questions: (1) how do patterns of spending and utilization of care for hip fracture patients differ across care settings in health systems that are structured and financed differently; (2) how do patterns of spending and utilization of care for these patients differ from patterns of spending and utilization in the 365 days prior to the hip fracture by country; and finally (3) to what extent do we observe notable differences in the total amount of spending and utilization of care for hip fracture patients across health systems?

| DATA AND METHODS
Our methodological approach to examine variations in health systems utilization and spending combines two existing approaches that are relatively novel for international comparison of health systems. First, we proposed to use linked patient-level data to examine the entire care pathway, rather than focusing only on care in the hospital setting.
Second, our unit of analysis is a specific type of HNHC patient, which we termed a patient persona, whom we followed throughout the system to record instances of utilization and associated spending over the course of a year. This approach builds on the use of clinical vignette methodologies that have been used by other projects to examine resource use in the inpatient setting 15 and by international organizations to examine variations in clinical practice. 16

| Data
We use linked patient-level data from 11 countries as follows: Australia, Canada, England, France, Germany, the Netherlands, New Zealand, Spain, Sweden, Switzerland, and the United States, accessed by members of the ICCONIC collaborative. Datasets included linked data across different domains of care, including primary care, outpatient specialty care, acute hospital care, post-acute rehabilitative care, outpatient pharmaceuticals, home health care, and long-term care. Specific details of each dataset used can be found in Table 1 of Appendix S1. Countries ability to collect comprehensive data across each domain for health care utilization and spending categories varied (Table 2 of Appendix S1).
The representativeness of the population for each dataset is found in Table 2 of Appendix S1. Data in three countries-New Zealand, Sweden, and Switzerland-covered their entire population.
Data in three other countries were from specific regions-Australia (New South Wales), Canada (Ontario), and Spain (Aragon). Data in the remaining five countries were large, regionally diverse samples, including in England, France, Germany, the Netherlands, and the United States. The proportion of patients covered in each dataset varied across countries, from 3% in Spain (Aragon) and 7% in England to 100% in New Zealand, Sweden, and Switzerland. Data from most countries were from 2016 to 2017, except for Spain, Sweden, and Switzerland, which is from 2015 to 2016. Australia (2012)(2013)(2014)(2015)(2016) and England (2014-2017) used data for a longer time period to allow for more observations given the size of the sample.

| Sample selection
Using the framework from the National Academy of Medicine report "Effective Care for High-Need Patients" as a starting point, we selected a patient persona that is representative of a frail older person. A frail older person was one of five priority populations identified as being among the most expensive to care for, have substantial health care needs, and are particularly vulnerable to poor-quality care. 17 The other priority populations were a person with a progressing, advanced illness; a person with complex multimorbidity; a young person with a major disability; and children with complex needs.
Our starting point was to define comparable group of patients to make up the hip persona for identification. We focused on patients older than 65 years across all systems, who were admitted to hospital with a primary diagnosis of hip fracture, which can be identified using the International Classification of Diseases-10th revision (ICD-10) diagnostic codes, as defined by the World Health Organization: S72.0, S72.1, and S72.2. These three diagnostic codes all represent fractures in the upper part of the femur, although each code represents a different type of fracture, which may require different procedures to treat.
As our group of analysis, we focus on the patients with this diagnosis who received one of three procedures: total hip replacement, partial hip replacement, or osteosynthesis (or pinning), which we identified with the relevant procedure codes in each country. The data from Spain and the Netherlands were not coded using ICD-10 codes. Spain relied on ICD-9 codes, and the Netherlands used comparable diagnostic codes available in the insurer data used for this study with help from clinical experts in the country. To advise on the selection of national codes and final group for analysis, we consulted with an international advisory board composed of national and international advisors from clinical, health policy, and research backgrounds (Table 3 of Appendix S1). Appendix S1. To identify and follow the hip fracture persona across their pathway of care over a period of a year and establish the excess utilization and spending associated with the hip fracture, we required 3 years of patient-level data. One year of data were used to identify all relevant hip fracture patients using the characteristics outlined above. The follow-up year was used to measure the service use and spending incurred by each patient, across all care settings, from day 1 of hospitalization for the 365 days that follow. A look-back year was used to establish a look-back period of 365 days for comparison and to establish baseline utilization and spending prior to hip fracture ( Figure 3 of Appendix S1). We used the look-back year a "baseline" year across all patients and countries. Given that all spending data is age-and sex-adjusted, we inferred that any additional dollar of spending observed in the year following the index hospitalization compared to the look-back year is largely attributable to the hip fracture event per se and related complications.

| Analysis
Due to constraints in data sharing, each country was only able to provide aggregated data for comparison. For each of the utilization and spending categories, countries supplied aggregated data reflecting mean use and spending in seven age groups (65-69, 70-74, 75-79, 80-84, 85-89, 90-94, and 95 years and up) stratified by sex.
While all countries provided expenditure data, it is important to note that we used the perspective of the health care payer across all countries. In most countries, this is carried out either directly by an insurance or sickness fund (Germany and the Netherlands) or directly from a national form of health insurance (the United States with Medicare program, Canada, etc.). Therefore, our study does not capture full costs (as it does not account for the fixed costs of all structures within a health system). It only captures the prices actually paid for the services, which across all countries, already included the fixed costs of the system. In addition, cost accounting methods used to estimate expenditure differ across countries, in part due to the differences in existing payment systems (Table 4 of Appendix S1). For example, some countries are able to report direct spending from incurred costs (those that rely on FFS entirely), while others provide information on reimbursement for specific episodes (e.g., diagnosis-related group [DRG]) or an unweighted average unit prices. There are also differences in payment systems within countries across the different sectors. For example, pharmaceutical spending across countries reflects the amount of pharmaceutical spending in the outpatient setting and includes different amounts of out-of-pocket contributions. In the United States, this expenditure category captures Part D spending; in Australia and Sweden, these estimates include co-payments. Finally, the reporting and imputation of capital investments or indirect costs also vary across system.
In order to reliably compare spending, we first applied the Organization for Economic Co-operation and Development's (OECD) Actual Individual Consumption Purchasing Power Parities (AIC PPPs) to the expenditure data. AIC PPPs, rather than GDP PPPs, are currently used by the OECD as the most reliable economy-wide conversion rates for health expenditure. Across each country, we applied 2017 AIC PPPs to all expenditures by age groups across the seven age groups, stratified by sex.
We then performed an age and sex direct standardization using the US sample population as the reference population for all countries. For each age group and sex, all utilization and spending measures were weighted and recalculated against the US sample population weights. The totals are then calculated by weighting each individual group and sex's shares on the original country-specific total to generate total, male, and female age-sex standardized values.
Across each category of spending and utilization, we then compared age-sex standardized results.   (Table 1). The mean patient age ranged from 81.2 years (standard deviation [SD] 6.9) in Switzerland to 85.4 years (SD 7.0) in Spain. The sample was predominantly female, with the proportion of women as high as 77.1% in France and the lowest at 62.8% in Australia. Countries varied in the ability to capture secondary diagnoses in the index hospitalization, ranging from an average of 3.7 comorbidities in the United States to 1.1 in New Zealand and Canada. 18,19 In all countries but Spain, the most common diagnostic code was  (Table 1).    per visit, respectively ( Figure 3D).

| Spending differences across countries
There was little variation across countries related to total outpatient specialty care spending over the year with most countries spending around $800 per person, apart from the United States that spent $3658 per person ( Figure 2E). However, the cost per MD specialist visit was more varied, with the United States spending comparatively more ( Figure 3E). Outpatient drug spending was the highest in the

| DISCUSSION
In this study, we examined patterns of utilization and spending for an These patterns may be culminating a number of different issues in the care pathway. It is possible that patients are discharged "quicker but sicker" from the hospital in the United States because of the widely accessible post-acute care infrastructure covered by Medicare. 21,22 Other countries, like England, do not have a comprehensive provision of accessible post-acute care service and, instead, observe much longer hospital length of stay. 23 Another possibility is that US patients have less access to affordable long-term care, as it is not covered by the Medicare program. 24 This may lead to a substitution of care, where long-term care services are being provided in the post-acute setting (predominately skilled nursing facilities).
Among the countries able to provide information on postacute rehab utilization and costs, we observed that those with universal long-term care systems (such as Canada, the Netherlands,