International comparison of health spending and utilization among people with complex multimorbidity

Abstract Objective The objective of this study was to explore cross‐country differences in spending and utilization across different domains of care for a multimorbid persona with heart failure and diabetes. Data Sources We used individual‐level administrative claims or registry data from inpatient and outpatient health care sectors compiled by the International Collaborative on Costs, Outcomes, and Needs in Care (ICCONIC) across 11 countries: Australia, Canada, England, France, Germany, the Netherlands, New Zealand, Spain, Sweden, Switzerland, and the United States (US). Data Collection/Extraction Methods Data collected by ICCONIC partners. Study Design We retrospectively analyzed age–sex standardized utilization and spending of an older person (65–90 years) hospitalized with a heart failure exacerbation and a secondary diagnosis of diabetes across five domains of care: hospital care, primary care, outpatient specialty care, post–acute rehabilitative care, and outpatient drugs. Principal Findings Sample sizes ranged from n = 1270 in Spain to n = 21,803 in the United States. Mean age (standard deviation [SD]) ranged from 76.2 (5.6) in the Netherlands to 80.3 (6.8) in Sweden. We observed substantial variation in spending and utilization across care settings. On average, England spent $10,956 per person in hospital care while the United States spent $30,877. The United States had a shorter length of stay over the year (18.9 days) compared to France (32.9) and Germany (33.4). The United States spent more days in facility‐based rehabilitative care than other countries. Australia spent $421 per person in primary care, while Spain (Aragon) spent $1557. The United States and Canada had proportionately more visits to specialist providers than primary care providers. Across almost all sectors, the United States spent more than other countries, suggesting higher prices per unit. Conclusion Across 11 countries, there is substantial variation in health care spending and utilization for a complex multimorbid persona with heart failure and diabetes. Drivers of spending vary across countries, with the United States being the most expensive country due to high prices and higher use of facility‐based rehabilitative care.

What is known on this topic • Health systems are structured and financed differently.
• Patients with complex multimorbidity are more susceptible to poor quality of care and incur higher health care costs than other patient groups.
• International comparisons of health systems mostly focus on the inpatient setting, with limited data evaluating differences in care across different components of the health care system.

What this study adds
• This study compares health care utilization and spending across 11 high-income countries for an older person aged 65-90 years hospitalized with a heart failure exacerbation and a secondary diagnosis of diabetes, across five domains of care: acute hospital care, primary care, outpatient specialty care, post-acute rehabilitative care, and outpatient drugs.
• Compared to other countries, the United States incurred the highest average of health care spending per person due to high unit expenditures and higher use of facility-based postacute rehabilitative care.
• All countries spent a substantial amount for people with heart failure over the course of the year, which increases incrementally among individuals with more complex comorbidities.

| INTRODUCTION
As populations across the globe continue to grow and age, health systems are faced with the challenge to manage the increasing prevalence and impact of multimorbidity in older adults of their populations.
In 2016, the World Health Organization declared the rise in multimorbidity, defined as the co-occurrence of at least two chronic conditions in a given individual, a worldwide epidemic. 1 A recent National Academy of Medicine (NAM) report 2 highlighted that among people with multimorbidity, those with major chronic conditions-including heart failure, diabetes, and chronic obstructive pulmonary disease (COPD)-should be considered a priority group by health systems as they are more susceptible to poor quality of care and incur higher health care costs for several reasons. First, these individuals are usually of advanced age with long-term chronic conditions that compromise their functional and cognitive abilities, thus limiting their ability to care for themselves effectively. 1,3 Second, their treatment often includes complex management regimens, which leads to issues of polypharmacy, adverse drug events, and medication adherence. 1,4 Finally, these patients are also at a higher risk for suffering the consequences of fragmented care, such as lack of coordination and poor communication, due to their frequent and complex interactions with multiple health care providers. 5,6 Comparisons across health systems may offer new knowledge regarding cost and quality differences and create opportunities for improvement internationally. To date, we have limited evidence of cross-country variability in the management of these complex populations and the extent to which systems make different, more efficient, use of certain care settings.
Therefore, as part of the International Collaborative on Costs, Outcomes, and Needs in Care (ICCONIC), we sought to understand whether there are key differences in health care spending and utilization among a persona with complex multimorbidity across 11 countries. 7 The countries in this study include Australia, Canada, England, France, Germany, the Netherlands, New Zealand, Spain, Sweden, Switzerland, and the United States (US). Using the NAM high-need, high-cost framework, we identified a comparable persona, defined as an older person hospitalized with a heart failure exacerbation with a co-occurrence of diabetes. We then evaluated patterns of resource use across countries for 365 days following a heart failure exacerbation.
Specifically, in this study, we asked the following key questions.
(1) How does health care spending vary across different components of the care pathway-including primary care, outpatient specialty care, hospital care, drugs, and post-acute rehabilitative care-among 11 high-income countries (where data are available)? (2) To what extent do these variations in health care spending persist after taking into account the relative country-specific utilization across care settings? Finally, (3) among a more complex subset of patients hospitalized with heart failure with a comorbidity of diabetes, how does an additional comorbidity of COPD influence health care spending across countries relative to those without diabetes or COPD? 2 | METHODS

| Country datasets
The ICCONIC collaborative uses patient-level data from 11 countries: Australia, Canada, England, France, Germany, the Netherlands, New Zealand, Spain, Sweden, Switzerland, and the United States. Specific details of each dataset used can be found in Appendix Table A1.
The representativeness of each dataset is found in Appendix

| Sample selection of heart failure persona
Using the high-need, high-cost framework identified by NAM, we selected a patient persona that is representative of an "older person with complex multimorbidity." This group of patients was identified by NAM as a population thought to especially benefit from improving the value of care and reducing inefficiencies in care provision. The NAM framework was constructed by a team of clinicians, academics, economists, and policy makers. 2 To identify a specific persona that is reflective of the older person with complex multimorbidity, we used a consensus decision making process among members of the ICCONIC collaborative, which included physicians, policy makers, data scientists, statisticians, and health economists. We defined the patient vignette with specified demographic and clinical criteria that were common enough across countries to allow for adequate sample selection. We then built a construct that reflected these requirements, which were identified as an older person, aged 65-90 years old, who was hospitalized with a heart failure exacerbation and who had a comorbidity of diabetes at the time of hospitalization. As a sub-sample, we also identified patients who had an additional comorbidity of COPD in a subset of countries that were able to do so.
We required at least 2 years of data. The first identification step was to identify all patients in year 1 who were hospitalized with a primary diagnosis of congestive heart failure with International Classification Code-10 of I50.x, as defined by the World Health Organization. Patients were included in the sample only once from their first recorded hospitalization in year 1. Given the lack of comprehensive longitudinal data across most countries, we were unable to know if the hospitalization was the first hospitalization related to heart failure or not. We then identified the subset of patients who at the time of admission also had a diagnosis of diabetes, including International Classification of Diseases, Tenth Revision (ICD-10) codes: E11.x, E12.x, E13.x, and E14.x. Of note, type 1 diabetes (E10.x) was excluded from these analyses a priori in order to identify a more uniform population across countries. Finally, we identified subsets of patients that had an additional comorbidity of COPD, including those with ICD-10 codes: J41.x, J42.x, J43.x, and J44.x. We also identified the prevalence of other chronic conditions using Elixhauser comorbidity definitions. 8 All countries used ICD-10 classification codes, except Spain (which used ICD-9 codes) and the Netherlands (which used a set of codes given by the data supplier that matched the ICD-10 diagnostic codes).

| Defining spending and utilization categories
Across countries, we tracked spending and utilization across five We then started from day 1 of hospitalization and followed patients across the five domains, where data were available across countries, for the 365 days that followed or until date of death if the patient did not survive the full year (see Appendix Figure A1).

| 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 five age groups (65-69, 70-74, 75-79, 80-84, and 85-90 years-last age cohort consisted of 6 years) stratified by sex. While all countries have expenditure data, the cost accounting methods used to estimate expenditure differ across countries, in part due to the differences in payment systems adopted (Appendix Table A7). For example, some countries are able to report direct spending from incurred costs (those with full costing systems) while others provide information on reimbursement for specific episodes (e.g., diagnosis-related groups) or an unweighted average unit prices. There are also differences in payment systems within countries across the different sectors. Finally, the reporting and imputation of capital investments or indirect costs also varies by system.
In order to reliably compare spending, we applied the Organiza- pathway-specific PPPs, using data from the collaborative, and also health-specific PPPs.
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 were 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. We then compared age-sex standardized results across each category of spending and utilization. Our primary persona was the person hospitalized with heart failure and with a concomitant diagnosis of diabetes. Across a subset of countries with complete data, we also performed an additional comparison that compared the relative total spending across the five domains of care of patients hospitalized with heart failure who did not have a diagnosis of COPD or diabetes with the primary persona (those with a secondary diagnosis of diabetes) and a more complex persona (those secondary diagnoses of both diabetes and COPD).  3.2 | Differences in spending across countries  Figure A2).

| Ethics approval
There were wide differences in spending related to primary care services, with as low as $421 per person in Australia to as high as $1557 per person in Spain. There was less variation across countries related to outpatient specialty care with one notable exception in the United States, which had significantly more spending at $9021 per person. Spending related to drugs was also the highest in the United States ($6118 per person), while it was the lowest in England ($809 per person).

| Spending per unit
Across almost the different domains of care, the United States spent more per unit of health service across four sectors (primary care, specialty care, drugs, and post-acute rehabilitative care) than other countries (Figure 3). Only Switzerland spent slightly more per person in acute hospital care than the United States, which was a close second.

| Relative spending increases by patient complexity
Among countries that had comprehensive data across all five care domains, as the complexity of the patient increased (those with diabetes and/or COPD relative to those without diabetes or COPD), the mean spending increased (Figure 4). The majority of the increase was related to increases in inpatient spending for heart failure patients with diabetes and COPD relative to those without these comorbidities. Across all four countries in this analysis, we observed increased levels of spending across all CHF personas as the level of complexity (additional comorbidities of diabetes and COPD) increased (Appendix Figure A5).
(D)  In addition, these differences may in part be related to what each health system covers. For example, countries like the Netherlands, Spain, France, Sweden, and Canada, all have much more generous long-term care coverage than the United States and England. 12 In the United States, Medicare, which is the primary insurance for people aged 65 years and older, does not cover long-term care services and support (unless dually enrolled and covered by Medicaid). In England, the National Health Services (NHS) does not cover long-term care (also referred to as social care). 12 Long-term care is a separate government service provided at the local level and means tested.
Prior work has found that nearly 70% of adults who survive till age 65 years will require some form of long-term care services and support. 13 In countries like the United States, the majority of people rely on family and unpaid caregivers for this support. 13 In other countries, however, long-term care benefits exist and are covered by insurance, which includes caregiver support and nursing care at home.
Some of this care is also likely shifted into the health care system, which takes the form of prolonged hospital days and prolonged courses of rehabilitative care in the United States. Our results suggest that policy makers in countries with limited long-term care services, such as the United States, should consider strategies to improve access to affordable long-term care services, especially given that caring for patients at hospitals and rehabilitative facilities is much more expensive than caring for them in residential settings.
Key cross-country differences were also observed in the number  We would like to thank the Swiss Federal Statistical Office for the provision of the linked data and in particular Antoinne Lieberherr, Patrick Schwab, and Katharina Weiss, who served as competent contacts at all times.

DISCLAIMERS
The results from New Zealand are not official statistics. • ResidEhpad (long-term care in residential facilities).

Germany
• Administrative data of a large, nationally active health insurance with more than 8 m enrollees (BARMER) (includes utilization/costs of all sectors that are paid by health insurance).

The Netherlands
• Zilveren Kruis insurance data (nationwide), which has about 30% of market share in the country.

New Zealand
• The Integrated Data Infrastructure.
• The National Minimum Dataset (hospital admissions data).
• The pharmaceutical collection (medication dispensing data).
• The National Non-Admitted Patient Collection (outpatient data).

Spain
• Base de datos de usuario (National Health Service users dataset including insurees admin data).
• Conjunto Mínimo Básico de Datos (administrative data for hospital discharges and outpatient contacts).
• Sistema de Informaci on Hospitalaria (outpatient visits to specialized care).
• Facturaci on Recetas (billing files of over-the-counter prescriptions).

Sweden
• The national patient registry (inpatient and outpatient specialized care).
• The national prescription drug registry (outpatient pharmaceuticals).
• The national mortality registry.
• The national registry for interventions in municipal health care (enrollment in home medical care).
• The national registry of measures for the elderly and people with disabilities (long-term care).
• Regional administrative registers of primary care consumption for the regions of Stockholm, Jönköping, Norrbotten, Skåne, and Västra Götaland.

Switzerland
• Medical statistics dataset of the Federal Statistical Office (FSO), including hospital admissions records.
• Patient data from hospital-based outpatient care dataset of the FSO.
• Short-and long-term care facility records dataset of the FSO.

United States
• Medicare fee-for-service data, 20% sample of all patients aged 65 years or older.