Background
Japan, the country with the oldest population, implemented a universal long-term care (LTC) insurance system in 2000. The Japanese LTC insurance system is one of the most comprehensive social care systems for older people in the world, built to assure fairness and efficient delivery of user-centered LTC services regardless of income. Japanese universal LTC system is highly decentralized, with municipality playing a key role in its operation. Municipalities operate as insurers, collect LTC insurance premiums, certify the need for LTC, provide insurance benefits and manage the LTC insurance finances. Regarding financing the LTC insurance, primary insured persons (aged 65 or over) and secondary insured people (40 to 64 years old) are contributing to 23% and 27% of total LTC budgets by paying insurance premiums. The other half of LTC budgets is covered by general tax (of which, 25% is covered by the national government, 12.5% is covered by prefectural governments and 12.5% is covered by municipal governments) [
1,
2]. Therefore, fiscal and budgetary pressure on LTC expenditure varies across municipalities depending on their local needs.
Large regional variations in healthcare utilization and spending have been documented in many countries [
3,
4]. Previous studies reported that demand factors such as demographics and health status largely explained regional variations [
5,
6]. Supply factors such as density of physicians and competition were investigated, and the impact of these variables varied according to the structural factors [
7]. Structural factors defined as political, economic, social, and organizational environments influenced regional variation [
6,
7]. Evidence from the abovementioned studies is used to address the gap between regions and contribute to the financial sustainability of healthcare system. However, to our knowledge, whether (and to what extent) there is a regional variation in LTC spending across the municipalities remains unclear.
Therefore, this study aims to examine municipality-level variations in LTC spending and clarify the drivers of such variations using national-level LTC claims open data.
Discussion
This is the first study to examine variation in LTC spending across municipalities in Japan using national LTC claims open data and other municipality-level statistics. Per-capita LTC spending among older adults was more than four times higher in the highest-spending municipalities than in the lowest. After adjusting for demand, supply, and structural factors, 84.0% of the total variance in LTC spending was explained. Demand-determined variance was remarkably high, which contributed to 85.7% of the overall R2. The proportion of severe care level among older adults was the covariate that explained most of the regional variation in LTC spending.
Older adults contribute to a portion of total LTC spending by paying insurance premiums; therefore, older adults living in municipalities with higher per capita LTC spending also bear a higher financial burden. Our results showed a great variation in LTC spending among municipalities in Japan. Since regional variation explained by demographic differences is unavoidable, we also calculated age-sex adjusted per-capita LTC spending. Following this, regional variation reduced remarkably; however, there was still considerable variation in adjusted per-capita LTC spending across the municipalities.
The finding that demand factors largely explained regional variation in LTC spending is in line with previous studies from other developed countries. Van Noort and their colleagues reported that demand factors contributed to 55% of regional variation in the usage of in-home care in Netherlands [
11]. Similar to LTC spending, demography and health explained 55–73% of regional variation in health care spending [
6,
7,
12]. The care-need level certification rate explained a great deal of the regional variation in LTC spending, despite controlling for demographic and care-need level. As a possible explanation, supplier-induced demand in the LTC market may be related to a higher care-need level certification rate [
13], because there was a strong correlation between care-need level certification rate and proportion of home care users. Thus, LTC beneficiaries living in municipalities that have an adequate supply of home care services can easily gain extensive information on these services and this may have been a link to higher care-need level certification rate. Another interpretation of this result is the health problems related to the care-need level certification rate. A Japanese study reported that a higher rate of patients (diseases of the circulatory system or cerebrovascular diseases) per 100,000 population is related to a higher care-need level certification rate [
13]. Accordingly, efforts to prevent the onset and severity of lifestyle-related diseases may help reduce per-capita LTC spending.
Our results demonstrated that the proportion of severe care-need levels (care-need levels 3–5) among older adults contributes to approximately 32.7% of the overall R
2. Therefore, to reduce the regional variations in LTC spending due to demand, a future study examining the factors associated with high care-need levels is needed. In addition, preventing the deterioration of the care level for mild and moderately disabled older adults may be linked to lower LTC spending. Previous studies have reported that in rehabilitation services [
14], additional payments for case-specific care services [
15] impact the deterioration of care level.
On the supply side, the number of LTC facilities per 1000 LTC beneficiaries explained 0.3% of the overall R
2, and was positively associated with higher per-capita LTC spending. This association is consistent with previous studies, presenting a cost underestimation of home and community care since no benefits for informal care are captured in the Japanese LTC insurance system [
16]. One Canadian study reported that home care is significantly less costly than residential care even when informal caregiver time is valued at replacement wage [
17]. Thus, checking if there is an excessive provision of LTC facility services among municipalities may help reduce LTC expenditure. In addition, one possibility of admission to LTC facility may be that the family members may not be able to take care of seniors at home. The current Japanese LTC system can only provide insurance benefits in kind, including in-home services (e.g., home visits/day services and short-stay services/care) and services at facilities; and do not include cash benefits or other direct benefits for family caregivers. Studies are warranted to investigate whether additional in-home services (especially more sufficient short-stay services/care), as well as cash benefits or other direct benefits for family caregivers, could help older people with LTC needs stay at home if they want.
Of the structural factors, higher financial capacity indexes and unemployment rates were correlated with higher LTC spending. Municipalities with higher financial capacity indexes have more residents with higher incomes, leading to better access to LTC services and higher LTC spending [
12,
15]. However, to locate and use LTC services, employees have to reduce their workload. Therefore they are less likely to access LTC services [
12]. Likewise, the cost of taking time off of work is incurred by family members when looking for caregiver services.
Our study has several limitations. First, we used aggregate data at the municipality level; thus, caution is needed before applying our results to individuals to avoid ecological fallacy. Second, our study was not able to identify the uses of cross-municipal LTC services, which may have caused a bias in assessing the regional variations in LTC spending. Since many urban older adults enter LTC facilities in surrounding rural areas, the LTC spending is reimbursed by urban municipalities despite receiving services in rural areas. Therefore, the density of LTC facilities in a municipality—the supply—may be related to the needs of the surrounding urban areas. Third, the supply-driven factors are generally undesirable, and therefore, it is helpful to control for as many of these as possible. Care market competition (i.e., Herfindahl–Hirschman Index), labor (i.e., the density of nursing staff), and the average length of stay in LTC facilities may explain regional variations [
6]; however, we could not adjust for these variables in this study. Fourth, we used care-need level as a proxy of health status; however, morbidity was not considered owing to data availability, even it is a sign of population health. Finally, the cross-sectional design cannot differentiate between cause and effect.
This study is the first attempt to examine variation in LTC spending using small area analysis. Since municipalities play a crucial role in LTC system in Japan, older adults in the same municipalities are more homogeneous in character than in larger areas such as prefecture. Consequently, our study displayed a wider variety of LTC spending across municipalities, making it easier to holistically identify and assess the issues of municipalities from the view of needs, supply, and structure. Regional variations could be a sign of inequity in access to LTC services and the inefficient and excessive use of LTC services [
6]; however, we would like to stress that our study does not aim to quantify inefficiencies. We examined the relative importance of demand and supply factors as drivers of regional variations in LTC spending. Second, our study presented the extent to which predictors reduce regional variations. Furthermore, even after controlling for the age-sex distribution, there were considerable regional variations in LTC spending, and most were driven by the proportion of severe care levels among older adults. Thus, policies to reduce health disparities may be an effective way to reduce regional variations in LTC spending.
Conclusions
In summary, we used national LTC claims open data, which cover all municipalities in Japan, to assess regional variation in LTC spending and identify its drivers. Our results revealed a large variation in LTC spending, despite adjusting for age and sex distribution across different municipalities. Adjusting for demand, supply, and system factors, 84.7% of the total variance in LTC spending was explained. Therefore, taking a closer look at municipalities from the demand, supply, and structural side is a necessary and effective way to reduce variation in LTC spending.
Publisher’s note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.