This study was conducted in the 39 municipalities in Nara Prefecture, Japan, as a model region. Nara Prefecture is located in the center of Japan’s main island of Honshu (Figure 1A) and has both urban and rural (mountainous) areas. Nara Prefecture has a population of 1,286,651 (approximately 1/100th of the national total), a land area of 3691 km² (approximately 1/100th of the national total) [13,14], and 32% (0.4/1.25) of adults aged ≥65 years (compared to 36.4/125.52, 29% nationally; Table 1). The characteristics of municipalities, such as population and aging rate, are shown in Table 1. Figures 1B and C illustrate the geographical locations of the 5 secondary medical areas and 39 municipalities. Nanwa secondary area, in the southern part, is the most rural area with mountains. Since there seems to be a geographical difference within Nara Prefecture, it would be suitable as a model region for geographical analysis. Secondary medical areas provide general inpatient medical services, whereas municipalities or smaller regional units provide more basic medical services.

For each municipality, geographical accessibility to home medical care services was analyzed using GIS (ArcGIS Desktop version 10.8.1; ESRI Japan Corporation) in the following two scenarios: (1) actual scenario: geographical accessibility to patients from hospitals and clinics in Nara Prefecture that actually provided home medical care services, and (2) ideal scenario: geographical accessibility to patients from all hospitals or clinics in Nara Prefecture was used. The latter simulates geographical accessibility in an ideal scenario in which the number of hospitals or clinics that provide home medical care services would increase. These scenarios were different only in terms of the number of hospitals and clinics used in the analysis, as described hereafter. Hypothetical patients were generated on a GIS by using the methodology described in a subsequent section “Geographical Analysis.”

For the actual scenario, hospitals and clinics that provided home medical care services were defined using data from the KDB database of Nara Prefecture (Nara KDB) in FY2019. This is a prefecture-wide database of medical and long-term care insurance claims. To be defined as hospitals and clinics that actually provided home medical care services, hospitals and clinics were required to have at least one claim with home medical care service fees (medical service code: 114001110, 114030310, 114042110, 114042210, 114042810, 114046310, 114027710, and 114027810 in the national fee schedule) for patients aged ≥75 years in Nara KDB in FY2019. Consequently, 346 hospitals and clinics were identified in the actual scenario.

The ideal scenario used all the hospitals and clinics in Nara prefecture, that is, those providing or not providing home medical care services. The geographical data on the 1308 hospitals and clinics were obtained from the Geospatial Information Authority of Japan [15]. Dental clinics were excluded from the study. The distribution of hospitals and clinics in the 2 scenarios is shown in Figure 1.

Population data at each 500 m² mesh (ESRI Japan Corporation) were used to distribute the locations of the hypothetical patients.

First, hypothetical patients were generated on GIS. The number of patients in each municipality was that of patients aged ≥75 years who used home medical care services in each municipality, identified from the Nara KDB data analysis. The usage of home medical care was defined in the same manner as identifying the hospitals and clinics in the actual scenario. That number of patients was randomly allocated to 500 m² meshes on GIS, reflecting the distribution of the population aged ≥75 years based on the national census 2020. In other words, the hypothetical patients were allocated to each mesh in a manner that the number of patients in each mesh is proportional to the mesh population aged ≥75 years using R version 4.1.2 (The R Foundation) [16,17]. This methodology enables the simulation of accessibility, reflecting the actual geographical distribution of potential patients. Thereafter, based on the patient allocations, hypothetical patient points were created on randomly determined points within the allocated meshes on ArcGIS using the “Create Random Points” function.

Subsequently, travel time from the patients to the hospitals or clinics was analyzed using the “Find Closest Facilities” function of ArcGIS for the 2 scenarios, respectively. We assumed that the travel to the patients was from the closest hospital or clinic. However, the closest hospitals or clinics can be located outside the municipality where the patients live.

The median and 25‐75 percentiles of travel distance and time were aggregated for the patients in each municipality. Moreover, the rates of patients for whom travel distances were over 16 km were calculated since, in the Japanese universal fee table, the travel distance to provide home medical care services needs to be 16 km or less, unless an exceptional reason is present [18].

The usage rate was calculated by dividing the number of patients aged ≥75 years using home medical care services by the total population aged ≥75 years. The entire process of geographical analysis was repeated 10 times to consider the uncertainty of patient locations in each trial. The scheme of geographical analysis is visualized in Figure 2.

This study was approved by the ethics committee of the National Institute of Public Health, Japan (NIPH-IBRA number 12324‐2). The requirement for informed consent was waived because all data were anonymized.

In this study, an accessibility analysis of home medical care services in Nara Prefecture was conducted as a model region, using the KDB data to investigate the equitable distribution of home medical care services. Using the KDB, we accurately identified medical institutions that provide home medical care, estimated the usage rates of home medical care services for each municipality, and demonstrated the feasibility of analyzing geographical accessibility to home medical care services that reflect actual health care usage and clinical trends.

The findings of this study revealed that in certain municipalities within the Nanwa area, a rural secondary medical area with a declining population, the travel distance to a health care patient’s home exceeded 16 km, and the travel time exceeded 30 minutes in some cases. These results highlight disparities among municipalities in terms of geographic accessibility to home medical care services. Its supply in the Nanwa area needs to be increased from the perspective of geographic accessibility.

In addition, the usage rates of these services were lower in rural municipalities in the Nanwa area. To promote the use of home medical care services, improving accessibility in these municipalities is necessary. Geographical accessibility—that is, the distance from service providers to patients—could affect service usage. Naruse et al [7] reported that travel distance can affect the usage of home nursing care services. Although geographical accessibility would also likely affect the usage of home medical care services, the extent of the effect has not been clear since this study did not quantify the effect by controlling other factors, such as socioeconomic status, care quality, and regional characteristics. Further studies with additional information beyond medical claim data are required to clarify this point.

In recent years, the government has promoted the use of home medical care to meet the increasing medical demands within each area, enabling local people to live their lives in areas they have always lived in [3]. Our results showed that a disparity occurred in geographical accessibility to home medical care based on the current practice in the actual scenario; however, in the ideal scenario, no municipality was observed in which the travel distance to any patient was longer than 16 km, except for Nosegawa. Geographical accessibility is much better in this scenario. For example, the average travel time was less than 15 minutes, except for Nosegawa. These results indicate that the disparity in terms of geographical accessibility would have been mostly resolved if all hospitals and clinics had provided these services. Furthermore, our results indicated that the potential benefit would be the largest in the Nanwa area, the most depopulated area. Therefore, if the number of hospitals or clinics providing these services is increased by a policy (eg, through dispatching health care professionals or funding), this will be effective in ensuring sufficient geographical accessibility for patients in the Nanwa area (specifically, Kawakami, Totsukawa, Kurotaki, and Kamikitayama). That is especially necessary in Kawakami and Kurotaki, since no hospitals or clinics were providing services, and the usage rates were low in these municipalities. However, it is neither feasible nor efficient for all hospitals and clinics to provide services because of limited resources. Further analysis using GIS could clarify how residents’ geographical accessibility will change when a specific hospital or clinic starts to provide home medical care services.

Notably, while the travel distance was longer in Nosegawa than in the other municipalities in both scenarios, the usage rate was the highest. The high usage rate could have been achieved by factors other than geographical accessibility. Further study, including factors other than geographical accessibility, is required to clarify the reason.

Our study is among the few that have analyzed geographical accessibility to home medical care services using GIS and KDB. The Nara KDB used in this study is an integrated database of medical and long-term care insurance claim information covering all residents aged 75 years and older, which matches the primary population of home medical care patients. This methodology enabled us to conduct a geographical analysis that reflected actual health care usage. Furthermore, our results are important when considering the optimal allocation of medical resources related to home medical care services and equal accessibility to these services. Japan is the most aging country in the world [1]. Home medical care services will be more important in other countries, and ensuring accessibility to these services will be of greater importance in the future. Geographical analyses using nationwide health care claims databases have been conducted in other countries for health care fields, such as drug prescription and pregnancy outcomes [11,12]. Likewise, our methodology can be applied to other countries for home medical care services, as those services will be more important as the aging of the population progresses.

The KDB is a comprehensive database that includes medical claims information for home medical care patients and details the provision of medical institutions, enabling a geographical analysis that reflects actual health care usage. However, this study had some limitations. First, it did not consider the capacity of each hospital or clinic because it focused on geographical accessibility. Further studies are required to consider its capacity to be more informative for optimal resource allocation. Second, a hospital or clinic was regarded as providing home medical care services if at least one claim regarding services from a hospital or clinic was identified. Some hospitals and clinics included in the actual scenario might not routinely provide home medical care services. Third, the disparity in geographical accessibility within each municipality is beyond the scope of this study. Such disparities may exist particularly in municipalities with lower geographical accessibility. Fourth, the KDB was used to analyze home medical care patients aged ≥75 years; therefore, home health care patients aged <75 years were not included. In the future, using the NDB or KDB and applying the developed methods, the analysis could be expanded to include pediatric home medical care patients and those under the age of 75 years. Finally, this study only used KDB data from FY 2019 to conduct geographical analysis while excluding the impact of the COVID-19 pandemic. The usage of home medical care has been increasing over the years, and the usage patterns could also change over the years. By applying the methodology established in this study, a nationwide analysis using the most recent data, free from the influence of the COVID-19 pandemic, can provide valuable insights into the future of home medical care.

This study conducted a geographical accessibility analysis of home medical care services using Nara Prefecture as a model region with GIS and KDB. The results revealed that geographical accessibility was lower in depopulated municipalities in the southern part, such as Totsukawa, Kawakami, Kamikitayama, and Nosegawa, which have lower accessibility than urban areas. However, the results also showed that the disparity could be partly solved in the ideal scenario. The differences in travel time between the scenarios were especially larger in the depopulated areas in the southern part, such as Totsukawa (32.6 vs 5.8 min), Kawakami (30.1 vs 11.8 min), Kurotaki (21.3 vs 5.2 min), and Kamikitayama (20.7 vs 3.5 min). These results highlight the necessity of increasing supply in the southern areas.