Geographical and climatic factors driving regional variation in medical centres for dementia in Japan: a nationwide geocomputational analysis Authors: Yoshihiko Baba, Shuji Tsuda, Kae Ito, Shuichi Awata Published: 2026 ABSTRACT AbstractObjectivesJapan’s dementia policies have mandated establishing at least one medical centre for dementia (MCD) in each secondary healthcare area (SHA) and ensure equitable access to specialised dementia care services across regions. This study aimed to identify the geographical and climatic characteristics of MCDs and examine geographical accessibility to MCDs, and to clarify the geographical characteristics of SHAs in which MCDs have not been established, despite the policy that an MCD should be established in each SHA.DesignA geocomputational analysis using publicly available geographical datasetsMethodsWe conducted a geocomputational overlay analysis of the spatial distribution of the three types of MCDs across areas with heavy snowfall, mountainous areas, agricultural areas, peninsulas and remote islands. We then reviewed prefectural development policies and municipal dementia care pathway documents in SHAs without MCDs. Finally, catchment analysis was performed to estimate geographical accessibility to the nearest MCD in areas where SHAs without an MCD were concentrated. We used the datasets of municipalities, secondary healthcare areas, heavy snowfall, mountainous areas, agricultural areas, peninsulas and remote islands, provided by the Ministry of Land, Infrastructure, Transport and Tourism and the dataset of MCDs provided by the Ministry of Health, Labour and Welfare.ResultsOverall, 509 MCDs were identified nationwide: 109 (19.8%) in areas of heavy or extremely heavy snowfall, 12 (2.4%) in mountainous areas, 72 (14.1%) in agricultural areas, 29 (5.7%) on peninsulas and 11 (2.2%) on remote islands. Of the 344 SHAs, 18 (5.2%) did not have an MCD, encompassing 105 municipalities. The MCDs comprise three categories: 21 core, 388 regional and 100 affiliated centres. While only one core centre (4.8%) was located in an agricultural area, 46 regional centres (11.9%) and 25 affiliated centres (25.0%) were located in agricultural areas. dementia care pathways analysis revealed that these municipalities either did not designate any MCDs (41%), designated MCDs in adjacent municipalities (21%) or listed all MCDs within the prefecture (36%). Municipalities that did not designate an MCD in their dementia care pathways were primarily located in geographically or climatically disadvantaged areas, including heavy-snowfall regions and peninsulas. Catchment area analysis using 30 km driving isochrone from the nearest MCD revealed that mountainous terrain, heavy snowfall and the presence of a large lake limited catchment coverage in municipalities that did not designate an MCD in their dementia care pathways.ConclusionsWhile the nationwide expansion of MCDs has largely achieved regional coverage, gaps remain in geographically disadvantaged areas where SHAs lack an MCD. Future dementia care planning should consider geographical accessibility in addition to administrative coverage to ensure equitable access to specialised dementia services. STRENGTHS AND LIMITATIONS OF THIS STUDY The strength of this study lies in its use of open data and provision of reproducible code. The strength of this study lies in its use of open data and provision of reproducible code. The limitations of this study encompass the inability to analyse outdated or unavailable data, including those for depopulated areas. The limitations of this study encompass the inability to analyse outdated or unavailable data, including those for depopulated areas. Accessibility analysis for remote islands could not be performed. Accessibility analysis for remote islands could not be performed. Introduction The WHO has been calling on countries around the world to develop national strategies for dementia.1 Although dementia diagnosis, treatment, care and support is one of seven action plans to be included, Alzheimer’s Disease International has reported that 75% of individuals with dementia globally may be undiagnosed.2 3 A meta-analysis has revealed that rural areas have a higher Alzheimer’s disease prevalence than urban centres,4 and the shortage of diagnostic and social welfare services in rural regions has also been emphasised.5 6 Japan has adopted a medical and long-term care policy enabling older adults to remain in their residential communities. The same is true for dementia, as Alzheimer’s Disease International has identified that Japan is one of the countries with the most advanced national dementia plan, currently at stage 5B. The Japanese Government has formulated the Outline for the Promotion of Dementia Policies in 2019 and enacted Basic Act on Dementia to Promote an Inclusive Society in 2024, aiming to achieve a society where individuals can live with hope despite having dementia. The Medical Centres for Dementia (MCDs) assume the roles of providing differential diagnosis of dementia, acute-phase medical care for behavioural and psychological symptoms of dementia and physical complications, specialist medical consultation, liaison with relevant organisations and the organisation of training sessions.3 The three types of MCDs encompass core (kikan), regional (chiiki), and affiliated (renkei) MCDs, although the affiliated type has been translated ‘collaborative’ in the previous literature.7 Core and regional MCDs are equipped with testing equipment and hospitalisation facilities and can deal with concurrent medical conditions, including wandering and depression, as well as complications from other illnesses. MCDs are set for establishment in each geographical healthcare area delineated by the prefectural governments. Core MCDs are expected to cover their own secondary healthcare area (SHA) and additional SHAs without an MCD within the same tertiary healthcare area, which is mostly equivalent to prefectural boundaries. Regional MCDs are responsible for their own belonging SHA, comprising several municipalities. Affiliated MCDs do not fully meet the regional criteria; however, they provide diagnosis, treatment and guidance in association with core, regional or other affiliated MCDs. To address geographical disparities, the policy aimed to establish at least 1 MCD in each of the 344 SHAs, aiming for a total of 500 MCDs. Although the target number of MCDs has been achieved as of 2024, not all SHAs have MCDs; those without MCDs are imagined to be in geographically and climatically constrained areas, but it remains unknown as to which areas. According to the Ministry of Land, Infrastructure, Transport and Tourism (MLIT), heavy snowfall areas, mountainous areas, agricultural areas, peninsulas and remote islands constitute geographically and climatically constrained areas. This paper aims to investigate whether the geographical and climatic characteristics are associated with the distribution of MCDs in Japan and to evaluate how their current placement shapes access to specialised dementia care in constrained areas such as heavy-snowfall regions, remote islands and mountainous areas. Methods This study included a nationwide geocomputational analysis, using command-based geographical information system tools for reproducibility and investigation of governmental documents. First, we conducted a geocomputational overlay analysis to identify whether MCDs were situated in heavy snowfall areas, mountainous areas, agricultural areas, peninsulas and/or remote islands. Second, we reviewed the development policies of the prefectures and the dementia care pathways of the municipalities in the SHAs with no MCD. Lastly, we performed catchment analysis to determine the distance from the nearest MCD for areas without MCDs. Materials In this study, geographical constraints were defined as the presence of MCDs in mountainous and/or agricultural areas, or on peninsulas or remote islands. Climatic constraints were defined as the presence of MCDs in areas of heavy snowfall. The land numerical information (LNI) datasets are publicly available geographical datasets provided by the MLIT website (https://nlftp.mlit.go.jp/ksj/) and encompass municipalities, SHAs, heavy snowfall areas, mountainous areas, agricultural areas and peninsulas. The datasets were prepared at the national level and showed 0% error in both completeness (ie, no missing or excess records compared with the reference data in one-to-one matching) and positional accuracy (ie, no coordinate discrepancies compared with the reference sources). The data files (GeoJSON and CSV) and source code (QMD) described below are available at the GitHub repository (http://github.com/babayoshihiko/MCD_2026/). Municipalities (data/N03_okinawa_amami.geojson and data/N03_small.geojson): Japan has 47 prefectures and approximately 2000 municipalities. Municipalities (data/N03_okinawa_amami.geojson and data/N03_small.geojson): Japan has 47 prefectures and approximately 2000 municipalities. SHAs (data/A38_ okinawa_amami.geojson and data/A38_small.geojson): Also known as secondary healthcare service areas or secondary medical areas, SHAs are geographical units responsible for most healthcare service provision. However, service distribution across SHAs has been inconsistent.8 SHA boundaries are available through the LNI. SHAs (data/A38_ okinawa_amami.geojson and data/A38_small.geojson): Also known as secondary healthcare service areas or secondary medical areas, SHAs are geographical units responsible for most healthcare service provision. However, service distribution across SHAs has been inconsistent.8 SHA boundaries are available through the LNI. Heavy snowfall areas (data/A22_small.geojson): Identified under the Act on Special Measures concerning Countermeasures for Areas of Heavy Snow. The heavy snowfall areas have two levels (heavy and extremely heavy). Heavy snowfall areas (data/A22_small.geojson): Identified under the Act on Special Measures concerning Countermeasures for Areas of Heavy Snow. The heavy snowfall areas have two levels (heavy and extremely heavy). Mountainous areas (data/A24_small.geojson): Defined under the Act on Promotion of Mountainous Areas. Mountainous areas (data/A24_small.geojson): Defined under the Act on Promotion of Mountainous Areas. Agricultural areas (data/A25_okinawa_amami.geojson and data/A25_small.geojson): Defined by the Act on the Promotion of the Improvement of Basic Conditions of Agriculture, Forestry and Other Business in Hilly and Mountainous Areas. Agricultural areas (data/A25_okinawa_amami.geojson and data/A25_small.geojson): Defined by the Act on the Promotion of the Improvement of Basic Conditions of Agriculture, Forestry and Other Business in Hilly and Mountainous Areas. Peninsulas (data/A18_small.geojson): Defined according to the Peninsula Development Act. Currently, 10 peninsulas are considered to have difficulty in accessibility. Peninsulas (data/A18_small.geojson): Defined according to the Peninsula Development Act. Currently, 10 peninsulas are considered to have difficulty in accessibility. Remote islands (data/rito.csv): Japan comprises 14 125 islands, including the 5 main islands (Hokkaido, Honshu, Shikoku, Kyushu and Okinawa). Of these islands, those not connected to the five main islands by bridge are considered remote islands, and a total of approximately 420 remote islands are inhabited. MCD (data/MCD.csv): The Ministry of Health, Labour and Welfare (MHLW) regularly updates the complete list of MCDs and their addresses with no missing data. We used the 2024 list, which included 509 MCDs, for our analysis. The addresses were converted to geographical data. Data of underpopulated area from the LNI datasets were not included because underpopulated areas have been redefined several times since the available dataset, and the available dataset is different from the current status. Geocomputational overlay analysis The file sizes of the downloaded datasets were large. First, we simplified the datasets by reducing vertices and omitted unnecessary attributes [geo_jp_map_01_simplify.qmd]. We retained only 1% of vertices and file sizes were decreased to 5–10% of the original size. Second, we constructed thematic maps [geo_jp_map_02_basemap.qmd]. The online version of the map is also available at https://babayoshihiko.github.io/MCD_2026/. The maps encompassed the location of MCDs by type, SHAs with no MCDs, snowfall areas, mountainous areas, agricultural areas, peninsulas and municipalities with remote islands. Third, we performed intersect analysis to determine whether MCDs were located in geographically and climatically constrained areas [geo_jp_map_03_intersect.qmd]. In this context, an MCD located within an area was considered ‘intersecting’ and spatially joined. We calculated the proportion of the three types of MCDs present in each area and stratified the areas on the basis of these counts [geo_jp_map_04_count.qmd]. For statistical analysis, we employed Fisher’s exact test and set the significance level at 0.05, using R (V.4.5.0). Dementia care pathway document review Japan does not operate a general practitioner system, and patients are free to visit any hospital or clinic directly. Consequently, people must identify on their own which facility can provide an appropriate dementia evaluation. Since 2013, municipalities in Japan have been creating dementia care pathways, which summarise for each region when, where and what kind of medical and care services people with dementia and their families can receive.6 For SHAs that do not have an MCD, we reviewed the dementia care pathways of the municipalities within them and described the measures they use. Geocomputational catchment analysis Lastly, we conducted catchment area analysis using OpenRoute Service (https://openrouteservice.org/) [geo_jp_map_06_nearest.qmd]. The analysis focused on prefectures containing at least one SHA lacking an MCD, as well as all immediately neighbouring prefectures, to capture potential cross-prefecture service utilisation. For these selected prefectures, isochrones were constructed using the OpenRoute Service to delineate reachable areas. A 30 km travel range was selected on the basis of established benchmarks for emergency care access in rural regions.8 9 The options ‘motor vehicles’ for transportation mode and ‘distance’ for range were adopted. Distance was preferred over time because the target region solely relies on motor vehicles for transport and is subject to unpredictable weather conditions that may substantially confound and possibly invalidate time-based estimates. To determine the area reachable within the 30 km travel distance over the existing road network, isochrones were calculated outward from the MCD centroid. Geocomputational software Geocomputational analysis and visualisation were performed using R (V.4.5.0) and R packages {geojp}, {gridExtra} (2.3), {leaflet} (2.2.3), {rmapshaper} (0.5.0), {sf} (1.0–20), {tidyverse} (2.0.0) and {tmap}.10 The source codes and processed datasets used in this study are publicly available at the MCD_2026 GitHub repository.11 YB developed R package {geojp} for several studies, including this study.12 Patient and public involvement None. Results In 2024, there were 509 MCDs, with three categories: 21 core, 388 regional and 100 affiliated centres. Among the 344 SHAs, 18 (5.2%) were not covered by regional/affiliated MCDs. In particular, five SHAs in Tokyo, Ishikawa, Fukui, Shiga and Miyazaki were currently not covered by core MCDs or regional MCDs in neighbouring SHAs. Geocomputational overlay analysis The geographical characteristics of the MCDs were as follows: 109 (19.8%) in areas of heavy or extremely heavy snowfall, 12 (2.4%) in mountainous areas, 72 (14.1%) in agricultural areas, 29 (5.7%) on peninsulas and 11 (2.2%) on remote islands. Core MCDs were predominantly situated in non-geographically constrained areas, with only one located in an agricultural area (table 1, figure 1). By contrast, geographically constrained areas more frequently contained regional and affiliated MCDs. Table 1Geographical and climatic characteristics of medical centres for dementiaGeographical/climatic characteristicsTotalTypeP valueCoreRegionalAffiliatedn21388100SnowfallNot snowfall40818 (85.7%)311 (80.2%)79 (79.0%)0.6328Heavy snowfall873 (14.3%)68 (17.5%)16 (16.0%)Extremely heavy snowfall140 (0%)9 (2.3%)5 (5.0%)MountainousNot mountainous49721 (100%)382 (98.5%)94 (94.0%)0.0431Mountainous120 (0%)6 (1.5%)6 (6.0%)AgriculturalNot agricultural43720 (95.2%)342 (88.1%)75 (75.0%)0.0029Agricultural721 (4.8%)46 (11.9%)25 (25.0%)PeninsulaNot peninsula48021 (100%)369 (95.1%)90 (90.0%)0.0813Peninsula290 (0%)19 (4.9%)10 (10.0%)Remote islandNot remote island49821 (100%)385 (99.2%)92 (92.0%)0.0005Remote island110 (0%)3 (0.8%)8 (8.0%)Note: Percentages are calculated within each MCD type.MCD, medical centre for dementia. Geographical and climatic characteristics of medical centres for dementia Note: Percentages are calculated within each MCD type. MCD, medical centre for dementia. Geographical characteristics of medical centres for dementia. (A) Heavy-snowfall areas, (B) mountainous areas, (C) agricultural areas, (D) peninsulas and (E) municipalities with remote islands. Note: Some remote islands have been omitted for cartographic convenience. MCD, medical centre for dementia; SHA, secondary healthcare area. Specifically, 46 regional MCDs (11.9% of regional MCDs) and 25 affiliated MCDs (25.0% of affiliated MCDs) were located in agricultural areas. In addition, 77 regional MCDs (19.8%) and 23 affiliated MCDs (23.0%) were situated in snowfall areas. Eight regional or affiliated MCDs were located on remote islands. Moreover, municipalities encompassing remote islands contained 2 core, 41 regional and 11 affiliated MCDs. Fisher’s exact test revealed that MCD type was statistically significantly associated with mountainous (p=0.0431), agricultural (p=0.0029) and remote island (p=0.0005) areas. Visual inspection of figure 1 suggests substantial spatial overlap between mountainous and agricultural areas. 18 SHAs across nine prefectures had no MCDs: 7 in Hokkaido, 8 across Honshu, 2 in Kyushu and 1 in the remote islands of Tokyo (figure 1, table 2, https://babayoshihiko.github.io/MCD_2026/FigureS1.html). Of these SHAs, 3 (2 in Nara and 1 in Saga) had access to a core MCD in the same prefecture, leaving 15 SHAs without direct support. Most unserved SHAs were characterised by heavy snowfall, mountainous or agricultural conditions, except for Eastern Saga, which lay outside these areas. In contrast, 91 SHAs had multiple MCDs. Table 2Secondary healthcare service areas with no medical centres for dementia by prefecturePrefectureTertiarySecondary healthcare areaConstituent municipalitiesGeographyMeasuresHokkaidoNorth Oshima and HiyamaFour towns (Yakumo, Oshamanbe, Imakane and Setana)Extremely heavy snowfall, mountainous, agricultural, peninsulaBelongs to Donan RegionSouth OshimaFive towns (Esashi, Kaminokuni, Assabu, Otobe and Okushiri)Extremely heavy snowfall, mountainous, agricultural, peninsula, remote islandsBelongs to Donan RegionNorth SorachiOne city (Fukahawa) and four towns (Moseushi, Chichibubetsu, Hokuryu and Numata)Extremely heavy snowfall, mountainous, agriculturalBelongs to Sorachi RegionSouth SorachiFour cities (Yubari, Iwamizawa, Bibai and Mikasa) and five towns (Minamihoro, Yuni, Naganuma, Kuriyama and Tsukigata)Extremely heavy snowfall, agriculturalBelongs to Sorachi RegionFuranoOne city (Furano), three towns (Kami Furano, Naka Furano, Minami Furano) and one village (Shimukappu)Extremely heavy snowfall, mountainous, agriculturalBelongs to Sorachi RegionRumoiOne city (Rumoi), six towns (Mashike, Obira, Tomamae, Haboro, Enbetsu and Teshio) and one village (Shosambetsu)Extremely heavy snowfall, mountainous, agriculturalBelongs to Dohoku RegionSoyaOne city (Wakkanai), seven towns (Hamatonbetsu, Nakatonbetsu, Esashi, Toyotomi, Rebun, Rishiri and Rishirifuji) and one village (Sarufutsu)Extremely heavy snowfall, mountainous, agricultural, remote islandsBelongs to Dohoku RegionTokyoTokyo remote islandsTwo towns (Oshima and Hachijo) and seven villages (Toshima, Niijima, Kamitsujima, Miyake, Mikurajima, Aogashima and Ogasawara)Remote islandsIshikawaNotoTwo cities (Wajima and Suzu) and two towns (Anamizu and Noto)Heavy snowfall, mountainous, agricultural, peninsula, remote islandsFukuiOkuetsuTwo cities (Ono, Katsuyama)Extremely heavy snowfall, mountainous, agriculturalTannanTwo cities (Sabae and Echizen) and three towns (Ikeda, Nanetsu and Echizen)Extremely heavy snowfall, mountainous, agriculturalYamanashiKyonanFive towns (Ichikawamisato, Hayakawa, Minobu, Nanbu and FujikawaHeavy snowfall, mountainous, agriculturalMCD is located in the adjacent SHA.AichiEast Mikawa (North)One city (Shinshiro), two towns (Shitara and Toei) and one village (Toyobe)Mountainous, agriculturalIntegrated with East Mikawa (South) SHAShigaKoseiOne city (Takashima)Heavy snowfall, mountainous, agriculturalNaraYesTowaThree cities (Tenri, Sakurai and Uda), three towns (Kawanishi, Miyake and Tawaramoto) and three villages (Yamazoe, Soni and Mitsue)Mountainous, agriculturalIntegrated with Seiwa SHANanwaOne city (Gojo), three towns (Yoshino, Oyodo and Shimoichi) and eight towns (Kurotaki, Tenkawa, Nosegawa, Totsukawa, Shimokitayama, Kamikitayama, Kawakami and Higashiyoshino)Mountainous, agriculturalIntegrated with Chuwa SHASagaYesEasternOne city (Tosu) and three towns (Kiyama, Kamimine and Miyaki)None (mainly urban area)MiyazakiSaito KoyuOne city (Saito), five towns (Takanabe, Shintomi, Kijo, Kawaminami and Tsuno) and one village (Nishimera)Mountainous, agriculturalCurrently calling for MCDMCD, medical centre for dementia; SHA, secondary healthcare area. Secondary healthcare service areas with no medical centres for dementia by prefecture MCD, medical centre for dementia; SHA, secondary healthcare area. Dementia care pathway analysis Care pathway documents showed significant variations in terms of how municipalities lacking an MCD described access to dementia services. Of the 18 unserved SHAs with 105 municipalities, 40 (38.1%) municipalities provided their dementia care pathways online. Two prefectures (Nara and Saga) depended on their core MCDs to cover neighbouring SHAs. In Hokkaido, dementia care pathway documents denoted the prefecture’s eight self-designated medical regions rather than SHAs, reflecting the distinct organisation of psychiatric resources. Tokyo’s unserved SHA was limited to the remote island areas (the Izu and Ogasawara archipelagos), where substantial geographical separation from the mainland necessitated distinctive care arrangements and reliance on mainland-based specialist services. No MCD was located within the islands. Dementia-related diagnostic support and consultation functions for these areas were served by the Tokyo Metropolitan Geriatric Medical Center as the designated MCD for these areas. In Ishikawa, Fukui and Shiga, the SHAs lacking MCDs were situated in areas with various geographical constraints, including heavy snowfall, mountainous terrain and, in the case of Ishikawa, peninsular topography and remote islands. In these prefectures, dementia care pathway documents frequently did not specify an MCD or listed facilities in adjacent cities or across Lake Biwa. Other prefectures had unique organisational decisions. Yamanashi assigned an MCD located outside the relevant SHA to cover Kyonan, and Aichi’s Eastern Mikawa Region served as a unified urban–rural block split into two SHAs. In Miyazaki, the unserved SHA was actively recruiting an MCD during the study. Across all municipalities in MCD-lacking SHAs, 42%, 21% and 36% of dementia care pathway documents did not state any MCD, referred to MCDs in adjacent municipalities, and listed all MCDs within the prefecture, respectively. Municipalities that mentioned no MCDs were frequently small and had only one Community General Support Centre, also known as Comprehensive Community Support Centre. Geocomputational catchment analysis The pathway analysis revealed that comprehensive catchment analysis was undertaken for Ishikawa, Fukui and Shiga, where each had MCD-lacking SHAs and was neighbouring Toyama, Gifu, Kyoto and Nara prefectures. Access patterns were evaluated across seven prefectures using a 30 km isochrone by car from the nearest MCD (figure 2, https://babayoshihiko.github.io/MCD_2026/FigureS2.html). Comprehensive geographical analysis of regions with fewer medical centres for dementia. MCD, medical centre for dementia. Among the 344 SHAs, 18 (5.2%) were not covered by regional/affiliated MCDs. In particular, five SHAs in Tokyo, Ishikawa, Fukui, Shiga, and Miyazaki were currently not covered by core MCDs or regional MCDs in neighbouring SHAs. In Ishikawa, the Noto Peninsula was located beyond the 30 km catchment of all surrounding MCDs, including those in neighbouring Toyama. Likewise, in Shiga, the Kosei Region lay outside the 30 km catchment owing to Lake Biwa serving as a major natural barrier. In Fukui, mountainous terrain and heavier snowfall contributed to limited catchment coverage in Okuetsu and Tannan. Discussion We conducted a geocomputational analysis of MCDs with publicly available and own datasets. Among the 509 MCDs, 21 were core centres among 47 prefectures, leaving 26 prefectures without any core MCD. The largest uncovered remote islands region is in Tokyo. The Tokyo Metropolis contains an SHA situated in remote islands. For these islands, Hiroo Hospital, a regional core hospital for island areas, manages emergency admissions and diagnostic support via image transmission. However, the Tokyo Metropolitan Geriatric Hospital and Institute of Gerontology, which serves as an MCD, provides dementia-related support. In prefectures outside Tokyo, MCDs may also offer coverage for remote islands. Ishikawa, Fukui and Shiga prefectures each contain SHAs in heavy-snowfall areas without MCDs. Catchment analysis revealed that the Noto SHA in Ishikawa prefecture experiences access-related challenges owing to its peninsular geography, whereas the Kosei SHA in Shiga prefecture encounters similar constraints owing to the presence of Lake Biwa. In Miyazaki prefecture, the centre is currently being recruited. The regions concerned are predominantly rural and mountainous. In Saito city, a core city in the region, dementia care pathways encompass two memory clinics that are not designated as MCDs. By contrast, dementia care pathways in neighbouring municipalities direct patients to MCDs located elsewhere within the prefecture. Kaneko et al developed the rurality index as an indicator for measuring regional disparities in general healthcare provision in Japan.13 This index incorporates population density, distance to secondary and tertiary hospitals, alongside remote islands and heavy-snowfall areas as covariates. Our research findings also reveal numerous commonalities, including the significance of remote islands, but not snowfall areas. Regarding remote islands, although Tokyo’s islands were validated as a challenge in this study, islands outside Tokyo did not pose a substantial challenge. Relatively large islands, including Sado and Oki Islands, contain MCDs. Our finding that agricultural areas or mountainous areas, which overlap to a large extent, may be covariates for geographical constraints, contradicts previous findings. Moreover, in such areas, even within urban centres, up to 50% of home care visits may not be performed,14 suggesting the significance of remote care services. Regarding the peninsula, a modest impact was noted. In particular, areas in the Noto Peninsula in Ishikawa are >30 km by car from the nearest MCD. This considerable distance underscores the pre-existing geographical vulnerability of the Noto Peninsula, which has been severely compounded by the 2024 Noto Peninsula earthquake,14 15 making reliable access even more crucial. A survey has revealed that approximately one-third of older adults reported partially destroyed, destroyed on a large scale or destroyed homes in the disaster.16 The large lake (Lake Biwa) also demonstrated a similar trend. Japan’s urban master plans adopt a Location Normalisation Plan framework to promote compact city development and disaster mitigation in response to decreasing birth rates and population ageing.17 SHAs lacking MCDs are located in rural areas, where accessibility represents a key challenge. A choice must be made whether older adults with dementia be guided to designated residential zones or support be provided for them to continue living in their residential communities. Comprehensive surveys of SHAs lacking MCDs are warranted. The strength of this study lies in its use of open data and provision of reproducible code. While R is frequently employed in medical statistics, its capabilities for geographical analysis are also being improved. The novelty lies in conducting analysis from a geographical perspective within healthcare policy. The limitations of this study encompass (1) the inability to analyse outdated or unavailable data, including those for depopulated areas; (2) accessibility analysis for remote islands could not be performed; (3) pathway documentation was not available online for more than 50% of municipalities; (4) population density was not considered in the analysis of MCD distribution and accessibility; and (5) the analyses were based on the 2024 MHLW list of MCDs. However, changes between the 2024 and 2025 lists were limited and are unlikely to materially affect the findings of this study. Conclusion This nationwide geocomputational analysis confirms that significant geographical and climatic constraints, particularly heavy snowfall, peninsulas (eg, Noto) and major water bodies (eg, Lake Biwa), limit equitable access to MCDs in Japan. These specific topographic barriers cause several constrained SHAs falling outside the effective 30 km catchment zone. To mitigate the enduring disparities in specialised dementia care, policy initiatives must henceforth prioritise location-specific strategies and the urgent deployment of remote early diagnosis and tele-consultation systems.