Nationwide Descriptive Analysis of Older Inpatients Eligible for Japan's Dementia‐Care Financial Incentive in Acute‐Care Hospitals Authors: Jinyan Wu, Kojiro Morita, Ayumi Igarashi, Hideo Yasunaga, Taro Kojima, Taisuke Yasaka, Yuya Kimura, Hiroki Matsui, Kiyohide Fushimi, Noriko Yamamoto‐Mitani Published: 2026 ABSTRACT ABSTRACTAimIn 2016, Japan introduced a dementia care add‐on (DCA; levels 1 and 2) to incentivize hospital‐based dementia care. However, nationwide characteristics of eligible patients, including care processes and outcomes, remain unclear. We conducted a descriptive national analysis of DCA‐eligible inpatients.MethodsWe analyzed a national inpatient database in Japan, including patients aged ≥ 65 years admitted between fiscal years 2016 and 2019 who met DCA eligibility criteria—moderate‐to‐severe dementia without severe consciousness disturbance. Patient characteristics, DCA uptake, care utilization indicators, and outcomes were summarized by year.ResultsAmong 1 046 652 hospitalizations, patients were predominantly of advanced age (mean 85.4 years) and functionally dependent, with > 70% requiring full assistance with activities of daily living at admission and discharge. Patient characteristics remained stable over time; approximately 20% were admitted for pneumonia or aspiration pneumonia. From 2016 to 2019, DCA1 and DCA2 billings increased from 6.4% to 20.8% and from 13.2% to 32.8%, respectively. Mean length of stay declined from 33.7 to 27.8 days, and in‐hospital mortality from 13.9% to 12.9%. Potentially inappropriate medication use at discharge decreased from 16.2% to 14.0%. In‐hospital fractures remained stable at 0.9%–1.0%. About half of patients admitted from home were discharged home.ConclusionsDCA‐eligible inpatients are highly vulnerable. Despite increased uptake, coverage remained incomplete. This study provides baseline data to inform dementia care policy and future evaluations. Introduction The number of older adults with dementia is projected to reach 139 million by 2050 [1, 2]. Acute‐care hospitalization is particularly challenging for this population, as dementia is associated with higher risks of adverse outcomes, including functional decline, inappropriate prescribing, prolonged length of stay (LOS), and institutionalization [3, 4, 5, 6], as well as increased in‐hospital mortality and readmission [7, 8]. Consequently, system‐level strategies are needed to improve the quality and safety of hospital care for patients with dementia. In Japan, approximately 30% of residents are aged ≥ 65 years, and one in five was projected to have dementia by 2025 [2, 9]. The government introduced the Dementia Care Add‐on (DCA) in April 2016 under the universal health insurance system to promote structured hospital‐based dementia care [10]. DCA has two reimbursement levels: DCA1, which incentivizes hospital‐wide multidisciplinary dementia care teams, and DCA2, which uses a ward‐based approach centered on dementia‐trained nurses. Both target hospitalized older adults with moderate‐to‐severe dementia; detailed eligibility criteria, staffing requirements, and reimbursement structures are provided in the Supporting Information S1. Despite nationwide implementation of the DCA scheme, the target clinical population and real‐world care processes for DCA‐eligible patients remain poorly characterized at the national level. Although several studies have evaluated its impact [11, 12, 13], comprehensive nationwide descriptive analyses capturing patient profiles, care processes, and outcomes among DCA‐eligible patients are scarce. Such baseline evidence is essential to contextualize dementia care policies and inform subsequent evaluations of their implementation. In this study, we aimed to provide a nationwide descriptive overview of DCA‐eligible inpatients in acute‐care hospitals in Japan using an administrative inpatient claims database. Specifically, we characterized patient demographics and clinical profiles, described DCA use as an indicator of care coverage, and summarized care processes and outcomes over time. Methods Data Sources We conducted a descriptive analysis using the Diagnosis Procedure Combination (DPC) inpatient database, a nationwide administrative claims and discharge database covering approximately half of all inpatient admissions in Japan [14]. It includes patient demographics; diagnoses coded using the International Classification of Diseases, 10th Revision (ICD‐10); procedures and medications during hospitalization; discharge status; Barthel Index scores [15] at admission and discharge as measures of functional dependence; and level of consciousness at admission. Diagnostic and procedural coding validity is high, and the database is widely used in clinical epidemiology and health services research [16, 17]. Since 2014, for patients aged ≥ 65 years, the DPC database has included the Criteria for Determination of Daily Life Independence of Older Adults with Dementia (CDED), an observer‐rated scale of dementia‐related functional independence [8]. The CDED is widely used in Japan, including for long‐term care insurance certification, and classifies patients into six categories: 0 (no dementia), I (no ADL limitation), II (mild limitation requiring partial assistance), III (moderate limitation requiring intermittent care), IV (severe limitation requiring continuous care), and M (severe psychiatric or physical symptoms requiring medical management). The CDED's ability to identify dementia against the Clinical Dementia Rating [18], validity for assessing ADL against the Functional Assessment Staging Test [19], and criterion validity against the Mini‐Mental State Examination and Hasegawa Dementia Scale–Revised [20] have been shown to be acceptable. In the DPC database, CDED is categorized as 0, I–II, and III–IV or M. Hospital‐level data were obtained from the Annual Report for the Functions of Medical Institutions, a mandatory nationwide survey of acute‐care hospitals under the Japanese Medical Care Act, which provides detailed hospital data from 2014 onward [21]. Using unique hospital identifiers, these data were linked to patient‐level DPC records. This study was conducted in accordance with the Declaration of Helsinki and approved by the Institutional Review Board of the University of Tokyo. The requirement for informed consent was waived because the data were anonymized prior to analysis. Participants We included hospitalized patients aged ≥ 65 years with a CDED score of III, IV, or M (moderate‐to‐severe dementia, consistent with DCA eligibility) admitted to acute‐care hospitals between April 2016 and March 2020 (fiscal years 2016–2019). We excluded patients with severely impaired consciousness, defined as a Japan Coma Scale score ≥ 30 [22]. To ensure each record represented a unique hospitalization, we included only the first admission per patient within the same hospital during the study period. Variables Variables were grouped into three domains: (1) demographic characteristics, (2) care‐related indicators, and (3) clinical events. Demographic characteristics included age, sex, body mass index, smoking history, level of consciousness at admission (Japan Coma Scale), Barthel Index category at admission (independence, partial assistance, or total assistance), admission route (emergency vs. elective), ambulance use, pre‐admission location (home, nursing facility, or another hospital), use of home‐ or community‐based care services before admission, Charlson Comorbidity Index [23], and treatment type (medical treatment, emergency surgery, or elective surgery). Detailed definitions are provided in Table S1. Primary diagnoses were categorized into 16 acute conditions: pneumonia, aspiration pneumonia, heart failure, cerebral infarction, kidney or urinary tract infection, intracranial injury, hip fracture, acute myocardial infarction, intestinal obstruction, bradyarrhythmia, cholangitis, cholecystitis, chronic kidney disease, type 2 diabetes, gastrointestinal inflammation, and peripheral arterial disease; all others were grouped as “other.” Hospital‐level characteristics included hospital type (university or tertiary emergency hospital) and ward type (intensive care unit, high‐care unit, acute‐care ward, or other). Care‐related indicators comprised annual proportions billed for DCA1 or DCA2, admission and discharge planning fees, rehabilitation add‐ons, pharmaceutical management fees, and potentially inappropriate medication (PIM) use during hospitalization. Based on the Japan Geriatrics Society guidelines and the American Geriatrics Society Beers Criteria, PIMs were operationally defined as antipsychotics for behavioral and psychological symptoms of dementia (BPSD) and medications for insomnia, including benzodiazepines and non‐benzodiazepine hypnotics [24, 25]. In‐hospital PIM use was described as an indicator of medication exposure. Clinical events included LOS, in‐hospital mortality, Barthel Index category at discharge, discharge destination (home, facility, or another hospital), in‐hospital fractures, and PIM prescription at discharge. PIM at discharge was assessed among patients discharged alive. In‐hospital fractures were identified using diagnosis and procedure codes, excluding fractures present at admission [26]. All variables were summarized descriptively to characterize clinical profiles and hospital care without assessing causal effects. Statistical Analysis Descriptive statistics characterized DCA‐eligible patients in FY2016–2019 and changes over time. Categorical variables are presented as counts and percentages; continuous variables as means with standard deviations or medians with interquartile ranges, as appropriate. Patient characteristics were compared across fiscal years using Pearson's χ 2 test for categorical variables and one‐way ANOVA for continuous variables. For multi‐category variables, parent‐row p‐values represent global tests of differences in the overall category distribution across fiscal years, with missing values included as a separate category where shown. Standardized differences comparing FY2016 and FY2019 values were calculated to assess the magnitude of differences in patient characteristics; values ≥ 0.1 were considered to indicate meaningful differences. Temporal trends in binary care‐related indicators and clinical events were assessed using the Cochran–Armitage trend test, and the trend in LOS was assessed using Cuzick's nonparametric trend test. Absolute changes from FY2016 to FY2019 with 95% confidence intervals were calculated for care‐related indicators and clinical events and are presented as percentage‐point differences for percentage outcomes and as differences in days for LOS. Analyses were performed using Stata/MP version 16.0 (StataCorp, College Station, TX, USA). Results Study Population Between FY2016 and FY2019, we identified 1 146 884 hospitalizations of patients aged ≥ 65 years with CDED ≥ III. After excluding 92 145 patients with severe consciousness disturbance, 1 054 739 remained. For analyses of care‐related indicators and clinical events, except in‐hospital mortality, we excluded 8087 admissions in which patients died within 24 h of admission. The final cohort for these analyses included 1 046 652 admissions from 1494 acute‐care hospitals. As DPC participation is updated annually, all available hospitals in each year were included; 501 and 876 hospitals billed for DCA1 and DCA2, respectively. Subsequent analyses included all DCA‐eligible patients regardless of billing status. Demographic Characteristics Annual cohort size ranged from 243 862 admissions in FY2019 to 291 066 in FY2018 (Table 1), with a stable mean age (85.3–85.5 years; SD 7.5); approximately 40% were male. Emergency admissions remained stable (83.3%–84.0%), while ambulance use increased slightly (42.8%–46.3%). At admission, approximately 52% were alert, 40% had “mildly impaired consciousness,” and 8%–10% were somnolent. TABLE 1Characteristics of older adults who were eligible for the Dementia‐Care add‐ons.Patient characteristicsOverall (n = 1 054 739)FY2016 (n = 256 576)FY2017 (n = 263 235)FY2018 (n = 291 066)FY2019 (n = 243 862)Standardized difference a p * Age, years, mean (SD)85.3 (7.5)85.3 (7.5)85.3 (7.5)85.4 (7.5)85.5 (7.5)0.03< 0.001Age category, years, n (%)< 0.00165–7498 173 (9.3)24 209 (9.4)24 521 (9.3)27 106 (9.3)22 337 (9.2)0.0175–84338 360 (32.1)83 793 (32.7)85 231 (32.4)92 635 (31.8)76 701 (31.5)0.03≥ 85618 206 (58.6)148 574 (57.9)153 483 (58.3)171 325 (58.9)144 824 (59.4)0.03Male, n (%)428 498 (40.6)103 073 (40.2)106 646 (40.5)118 961 (40.9)99 818 (40.9)0.02< 0.001Emergency admission, n (%)882 446 (83.7)215 611 (84.0)220 486 (83.8)242 536 (83.3)203 813 (83.6)0.01< 0.001Ambulance use, n (%)470 355 (44.6)109 906 (42.8)117 070 (44.5)130 455 (44.8)112 924 (46.3)0.07< 0.001Smoking status, n (%)< 0.001Nonsmoker762 992 (72.3)187 592 (73.1)190 916 (72.5)210 042 (72.2)174 442 (71.5)0.04Past/current smoker156 486 (14.8)36 934 (14.4)38 787 (14.7)43 656 (15.0)37 109 (15.2)0.02Missing data135 261 (12.8)32 050 (12.5)33 532 (12.7)37 368 (12.8)32 311 (13.2)0.02Consciousness on admission, n (%)< 0.001Alert549 255 (52.1)133 534 (52.0)137 034 (52.1)152 661 (52.4)126 026 (51.7)0.01Mildly impaired consciousness412 515 (39.1)98 755 (38.5)102 654 (39.0)113 500 (39.0)97 606 (40.0)0.03Somnolent92 969 (8.8)24 287 (9.5)23 547 (8.9)24 905 (8.6)20 230 (8.3)0.04Body mass index (kg/m2), n (%)< 0.001≤ 18.4298 953 (28.3)74 844 (29.2)75 000 (28.5)81 865 (28.1)67 244 (27.6)0.0418.5–24.9508 104 (48.2)121 729 (47.4)126 643 (48.1)141 629 (48.7)118 103 (48.4)0.0225.0–29.987 791 (8.3)20 130 (7.8)22 005 (8.4)24 649 (8.5)21 007 (8.6)0.0330.0–34.911 093 (1.1)2500 (1.0)2716 (1.0)3190 (1.1)2687 (1.1)0.01≥ 35.01807 (0.2)429 (0.2)468 (0.2)504 (0.2)406 (0.2)0.00Missing data146 991 (13.9)36 944 (14.4)36 403 (13.8)39 229 (13.5)34 415 (14.1)0.01Barthel Index category at admission, n (%)< 0.001Independent48 060 (4.6)10 444 (4.1)11 864 (4.5)13 563 (4.7)12 189 (5.0)0.04Partial assistance43 123 (4.1)9520 (3.7)10 635 (4.0)12 228 (4.2)10 740 (4.4)0.04Total assistance808 747 (76.7)197 330 (76.9)200 976 (76.3)223 833 (76.9)186 608 (76.5)0.01Missing data154 809 (14.7)39 282 (15.3)39 760 (15.1)41 442 (14.2)34 325 (14.1)0.03Location before hospital admission, n (%)< 0.001Home612 630 (58.1)146 628 (57.1)153 566 (58.3)169 973 (58.4)142 463 (58.4)0.03Other hospital134 726 (12.8)32 123 (12.5)33 755 (12.8)37 556 (12.9)31 292 (12.8)0.01Nursing home305 322 (28.9)76 936 (30.0)75 338 (28.6)83 194 (28.6)69 854 (28.6)0.03Other2061 (0.2)889 (0.3)576 (0.2)343 (0.1)253 (0.1)0.05Home‐based care service use before hospital admission, n (%)< 0.001Yes179 572 (17.0)45 899 (17.9)44 114 (16.8)48 588 (16.7)40 971 (16.8)0.03No862 263 (81.8)207 211 (80.8)215 708 (81.9)239 325 (82.2)200 019 (82.0)0.03Missing data12 904 (1.2)3466 (1.4)3413 (1.3)3153 (1.1)2872 (1.2)0.02Charlson Comorbidity Index, n (%)0.0272630 784 (59.8)153 507 (59.8)157 423 (59.8)173 503 (59.6)146 351 (60.0)0.003–4315 307 (29.9)76 836 (29.9)78 831 (29.9)87 223 (30.0)72 417 (29.7)0.01≥ 5108 648 (10.3)26 233 (10.2)26 981 (10.2)30 340 (10.4)25 094 (10.3)0.00Primary diagnosis, n (%)< 0.001Aspiration pneumonia111 574 (10.6)29 443 (11.5)28 166 (10.7)29 551 (10.2)24 414 (10.0)0.05Pneumonia84 366 (8.0)21 110 (8.2)20 617 (7.8)23 559 (8.1)19 080 (7.8)0.01Hip fracture81 122 (7.7)19 637 (7.7)20 480 (7.8)22 470 (7.7)18 535 (7.6)0.00Cerebral infarction65 997 (6.3)16 054 (6.3)16 986 (6.5)18 018 (6.2)14 939 (6.1)0.01Heart failure51 633 (4.9)12 269 (4.8)12 910 (4.9)14 586 (5.0)11 868 (4.9)0.00Kidney infection44 655 (4.2)10 098 (3.9)10 272 (3.9)12 820 (4.4)11 465 (4.7)0.04Intracranial injury23 304 (2.2)5435 (2.1)5920 (2.2)6301 (2.2)5648 (2.3)0.01Cholangitis20 583 (2.0)4863 (1.9)4915 (1.9)5646 (1.9)5159 (2.1)0.02Bowel obstruction13 764 (1.3)3378 (1.3)3353 (1.3)3945 (1.4)3088 (1.3)0.00Cholecystitis10 728 (1.0)2453 (1.0)2628 (1.0)3121 (1.1)2526 (1.0)0.01Gastrointestinal inflammation10 570 (1.0)2623 (1.0)2579 (1.0)2929 (1.0)2439 (1.0)0.00Chronic kidney disease9484 (0.9)1248 (0.5)1407 (0.5)3794 (1.3)3035 (1.2)0.08Bradyarrhythmia7998 (0.8)1870 (0.7)1940 (0.7)2177 (0.7)2011 (0.8)0.01Acute myocardial infarction5922 (0.6)1313 (0.5)1454 (0.6)1715 (0.6)1440 (0.6)0.01Diabetes (Type 2)5869 (0.6)1416 (0.6)1508 (0.6)1607 (0.6)1338 (0.5)0.00Peripheral arterial disease4969 (0.5)1799 (0.7)1712 (0.7)807 (0.3)651 (0.3)0.06Others502 201 (47.6)121 567 (47.4)126 388 (48.0)138 020 (47.4)116 226 (47.7)0.01Admitting hospital: academic, n (%)41 797 (4.0)9823 (4.1)10 662 (4.3)11 179 (4.3)10 133 (4.7)0.03< 0.001Admitting hospital: tertiary care center, n (%)213 874 (20.3)51 808 (21.6)53 843 (21.9)56 444 (21.8)51 779 (23.8)0.05< 0.001Ward type on admission, n (%)< 0.001ICU22 126 (2.1)5575 (2.2)5638 (2.1)5876 (2.0)5037 (2.1)0.01HCU90 561 (8.6)20 411 (8.0)22 369 (8.5)24 707 (8.5)23 074 (9.5)0.05Acute‐care wards722 405 (68.5)177 570 (69.2)179 964 (68.4)198 220 (68.1)166 651 (68.3)0.02Other wards219 647 (20.8)53 020 (20.7)55 264 (21.0)62 263 (21.4)49 100 (20.1)0.01Treatment, n (%)< 0.001Emergency operation67 512 (6.4)15 874 (6.2)16 844 (6.4)19 044 (6.5)15 750 (6.5)0.01Planned surgery38 553 (3.7)8645 (3.4)9825 (3.7)10 992 (3.8)9091 (3.7)0.02Medical treatment948 674 (89.9)232 057 (90.4)236 566 (89.9)261 030 (89.7)219 021 (89.8)0.02Abbreviations: FY, fiscal year; HCU, high‐care unit; ICU, intensive care unit; SD, standard deviation. a Standardized differences compare FY2016 and FY2019 values and indicate the magnitude of differences across the study period. Missing values were included as a separate category where shown. Owing to the large sample size, even small differences may have reached statistical significance.* p‐values indicate overall differences across fiscal years (FY2016–2019): Pearson's χ 2 test for categorical variables and one‐way ANOVA for continuous variables. For multi‐category variables, parent‐row p‐values represent global tests of differences in the overall category distribution across fiscal years. Characteristics of older adults who were eligible for the Dementia‐Care add‐ons. Abbreviations: FY, fiscal year; HCU, high‐care unit; ICU, intensive care unit; SD, standard deviation. Standardized differences compare FY2016 and FY2019 values and indicate the magnitude of differences across the study period. Missing values were included as a separate category where shown. Owing to the large sample size, even small differences may have reached statistical significance. p‐values indicate overall differences across fiscal years (FY2016–2019): Pearson's χ 2 test for categorical variables and one‐way ANOVA for continuous variables. For multi‐category variables, parent‐row p‐values represent global tests of differences in the overall category distribution across fiscal years. At admission, 76%–77% required total assistance, and 4%–5% were independent. Primary diagnoses were dominated by pneumonia and aspiration pneumonia (18%–19%), with a slight decline in aspiration pneumonia (11.5%–10.0%). Other common diagnoses included hip fracture (7%–8%), cerebral infarction (approximately 6%), and heart failure (approximately 5%). Approximately 60% were admitted from home (57.1% in FY2016 to 58.4% in FY2019), 29% from nursing homes, and 13% from other hospitals. Pre‐admission use of home‐ or community‐based services declined slightly (17.9%–16.8%). Charlson Comorbidity Index categories were stable (CCI = 2 and CCI ≥ 5 in ~60% and 10%, respectively). General acute‐care wards accounted for approximately 68% of admissions at entry, with a modest increase in high‐care unit use (8.0%–9.5%); intensive care unit use remained approximately 2%. Most patients received medical treatment (approximately 90%); emergency and elective surgeries accounted for 6%–7% and 3%–4%, respectively. University hospitals accounted for 4%–5% of admissions, and approximately 20% were admitted to tertiary emergency centers, with a slight increase over time (21.6%–23.8%). In this large sample, many p‐values indicated statistically significant differences across fiscal years, but standardized differences between FY2016 and FY2019 were below 0.1, suggesting that the magnitude of these differences was generally small. Care‐Related Indicators DCA uptake increased markedly over time. The proportion of patients billed for DCA1 rose from 6.4% in FY2016 to 20.8% in FY2019, whereas DCA2 increased from 13.2% to 32.8%, peaking at 34.4% in FY2018 (Table 2). Use of supportive reimbursement items also increased, including comprehensive drug management (69.4% vs. 76.6%), comprehensive ADL support (64.8% vs. 68.9%), and admission and discharge support (39.9% vs. 54.9%). TABLE 2Adoption of the Dementia‐Care Add‐ons, supportive care items, and in‐hospital use of potentially inappropriate medications.Care process/medication categoryFY2016 (n = 254 492)FY2017 (n = 261 256)FY2018 (n = 288 947)FY2019 (n = 241 957)Absolute change (95% CI) p for trendDementia care add‐ons and supportive itemsDementia‐Care Add‐on 16.412.116.520.8+14.4 (14.2 to 14.6)< 0.001Dementia‐Care Add‐on 213.230.834.432.8+19.6 (19.4 to 19.8)< 0.001Comprehensive drug assessment69.472.474.476.6+7.3 (7.0 to 7.5)< 0.001Comprehensive ADL support64.867.468.668.9+4.1 (3.8 to 4.4)< 0.001Admission and discharge support39.946.851.554.9+15.0 (14.7 to 15.2)< 0.001In‐hospital use of PIMsAny selected PIM a 51.854.053.452.6+0.8 (0.5 to 1.1)< 0.001Antipsychotics for BPSD30.433.033.333.7+3.2 (3.0 to 3.5)< 0.001Insomnia medication38.539.037.435.4−3.2 (−3.4 to −2.9)< 0.001Benzodiazepines (injection)13.813.913.913.7−0.1 (−0.3 to 0.1)0.460Benzodiazepines (oral)19.719.317.415.2−4.5 (−4.7 to −4.3)< 0.001Non‐benzodiazepine hypnotics (Z‐drugs)14.014.814.213.4−0.6 (−0.8 to −0.4)< 0.001Abbreviations: ADL, activities of daily living; BPSD, behavioral and psychological symptoms of dementia; CI, confidence interval; FY, Fiscal Year; PIM, potentially inappropriate medication. a Any selected PIM included antipsychotics for BPSD or medication for insomnia, including benzodiazepine and non‐benzodiazepine hypnotics. Data are presented as percentages of admissions in each fiscal year after excluding deaths within 24 h of admission. Absolute changes were calculated as FY2019 minus FY2016 values using unrounded percentages and are presented as percentage‐point differences with 95% confidence intervals. The p‐value for trend was assessed using the Cochran–Armitage test. Adoption of the Dementia‐Care Add‐ons, supportive care items, and in‐hospital use of potentially inappropriate medications. Abbreviations: ADL, activities of daily living; BPSD, behavioral and psychological symptoms of dementia; CI, confidence interval; FY, Fiscal Year; PIM, potentially inappropriate medication. Any selected PIM included antipsychotics for BPSD or medication for insomnia, including benzodiazepine and non‐benzodiazepine hypnotics. Data are presented as percentages of admissions in each fiscal year after excluding deaths within 24 h of admission. Absolute changes were calculated as FY2019 minus FY2016 values using unrounded percentages and are presented as percentage‐point differences with 95% confidence intervals. The p‐value for trend was assessed using the Cochran–Armitage test. Patterns of psychotropic medication use during hospitalization were mixed (Table 2): antipsychotic use for BPSD increased slightly (30.4%–33.7%), whereas use of medications for insomnia decreased (38.5%–35.4%), primarily owing to reduced benzodiazepine prescriptions (19.7%–15.2%). Absolute changes with 95% CIs for these care‐related indicators are shown in Table 2. Clinical Events Mean LOS decreased steadily, from 33.7 days (SD 51.0; median 21 [IQR 11–40]) in FY2016 to 27.8 days (SD 27.0; median 20 [IQR 11–36]) in FY2019 (Table 3). Crude in‐hospital mortality, including deaths within 24 h of admission, declined modestly from 13.9% in FY2016 to 12.9% in FY2019. The proportion of in‐hospital fractures remained 0.9%–1.0%. TABLE 3Patient outcomes in older adults who are eligible for the dementia care add‐on.OutcomesFY2016FY2017FY2018FY2019Absolute change (95% CI) p for trendLength of stay, days, mean (SD)33.7 (51.0)33.2 (43.5)31.5 (35.7)27.8 (27.0)−5.9 (−6.1 to −5.7)< 0.001In‐hospital mortality, n (%)35 688 (13.9)35 465 (13.5)38 190 (13.1)31 507 (12.9)−1.0 (−1.2 to −0.8)< 0.001In‐hospital fracture, n (%)1780 (0.9)1930 (0.9)2320 (1.0)1765 (0.9)+0.05 (−0.01 to 0.11)0.018PIMs prescription at discharge a , n (%)35 723 (16.2)35 768 (15.7)38 245 (15.1)29 753 (14.0)−2.2 (−2.4 to −2.0)< 0.001Antipsychotics for BPSD, n (%)19 501 (8.8)20 197 (8.9)22 400 (8.9)18 138 (8.5)−0.3 (−0.5 to −0.1)0.002Insomnia medications, n (%)22 249 (10.1)21 208 (9.3)21 725 (8.6)15 927 (7.5)−2.6 (−2.7 to −2.4)< 0.001Home discharge, n (%)72 838 (50.1)76 117 (49.9)85 958 (50.9)72 154 (51.0)+1.0 (0.6 to 1.3)< 0.001Abbreviations: BPSD, behavioral and psychological symptoms of dementia; CI, confidence interval; FY, fiscal year; PIM, potentially inappropriate medication; SD, standard deviation. a PIMs comprise antipsychotics for BPSD and medications for insomnia (benzodiazepine and non‐benzodiazepine classes). Absolute changes were calculated as FY2019 minus FY2016 values using unrounded values and are shown as differences in days or percentage‐point differences with 95% confidence intervals. p‐Values for trend were derived from Cuzick's non‐parametric test for length of stay and Cochran–Armitage tests for binary outcomes. In‐hospital mortality includes deaths within 24 h of admission; PIM prescriptions at discharge were assessed among patients discharged alive; and home discharge was assessed among patients admitted from home. Patient outcomes in older adults who are eligible for the dementia care add‐on. Abbreviations: BPSD, behavioral and psychological symptoms of dementia; CI, confidence interval; FY, fiscal year; PIM, potentially inappropriate medication; SD, standard deviation. PIMs comprise antipsychotics for BPSD and medications for insomnia (benzodiazepine and non‐benzodiazepine classes). Absolute changes were calculated as FY2019 minus FY2016 values using unrounded values and are shown as differences in days or percentage‐point differences with 95% confidence intervals. p‐Values for trend were derived from Cuzick's non‐parametric test for length of stay and Cochran–Armitage tests for binary outcomes. In‐hospital mortality includes deaths within 24 h of admission; PIM prescriptions at discharge were assessed among patients discharged alive; and home discharge was assessed among patients admitted from home. At discharge, functional dependence was similar: 72.4% required total assistance, 3.3% were independent, and 9.0% required partial assistance; 15.3% were classified as death or missing (Figure 1). The proportion prescribed PIMs at discharge declined from 16.2% to 14.0%. By medication type, antipsychotic prescriptions decreased slightly (8.9%–8.5%), whereas insomnia medications decreased more substantially (10.1%–7.5%). Absolute changes with 95% CIs for LOS and binary clinical outcomes are shown in Table 3. Distribution of Barthel Index categories at admission and discharge among inpatients eligible for the dementia care add‐on included in the analysis (N = 1 054 739). Missing Barthel Index data are shown separately at admission and discharge, and in‐hospital deaths are shown separately at discharge. All categories account for the full study population; percentages may not total exactly 100% because of rounding. Figure 2 illustrates transitions from pre‐admission location to discharge destination. Overall, 33.7%, 26.9%, and 26.6% of patients were discharged home, to nursing homes, and to other hospitals, respectively. Among those admitted from home, approximately half were discharged home throughout the study period. Location before hospital admission and discharge destination among patients eligible for the dementia care add‐on included in the analysis (N = 1 054 739). Panel A shows location before hospital admission, and Panel B shows discharge destination. All categories, including “Others” and in‐hospital death, are presented; percentages in each pie chart sum to 100%. Discussion We outlined the clinical landscape of older adults with moderate‐to‐severe dementia admitted to acute‐care hospitals between FY2016 and FY2019 during early implementation of Japan's DCA scheme. Patients were highly functionally dependent and predominantly admitted on an emergency basis, while demographic characteristics remained largely stable over time. Uptake of DCA and supportive reimbursements increased, alongside modest reductions in LOS, in‐hospital mortality, and PIM prescriptions at discharge. These findings offer baseline context for understanding the real‐world population targeted by dementia care policies and inform evaluation of care quality and policy impact. Patient Characteristics and Background of Hospitalization Most participants were ≥ 85 years, and over three‐quarters required total assistance with ADL at admission. These findings highlight marked frailty and functional vulnerability among DCA‐eligible patients in acute‐care settings, consistent with prior nationwide analyses [12]. Pneumonia and aspiration pneumonia accounted for about one‐fifth of admissions, followed by hip fracture, cerebral infarction, and heart failure—common causes of hospitalization in patients with dementia [8, 12]. A small proportion of patients were admitted to university hospitals and tertiary emergency centers, and most received non‐surgical medical management. Hospital care likely emphasized conservative and supportive treatment rather than invasive interventions, reflecting clinical complexity and limited physiological reserve. Only about one‐fifth of patients used pre‐admission home‐ or community‐based care services. As eligibility was based on functional severity rather than diagnostic coding, some may not have had a formal dementia diagnosis or structured community support before admission [7]. Acute hospitalization may offer an opportunity to identify unrecognized dementia and unmet care needs and to initiate care planning and post‐discharge support [27]. In addition, hospitalization may contribute to functional decline, increasing the need for long‐term care after discharge [3]. Trends in Add‐On Certification The increase in patients billed for DCA1 and DCA2 paralleled national growth in certified hospitals [28, 29], suggesting DCA functioned as a policy lever to promote hospital‐based dementia care systems. The proportion receiving care‐related services also increased, reflecting both diffusion of the add‐on scheme and broader initiatives to strengthen inter‐professional geriatric care. Nevertheless, DCA1 coverage remained limited at about one‐fifth of eligible inpatients 4 years after introduction, indicating that structured multidisciplinary dementia care did not reach a substantial share of the target population. Facilities with limited staffing or access to dementia specialists may struggle to meet certification requirements [30]. In 2020, reimbursement criteria were revised to improve feasibility and dissemination [31]. To our knowledge, Japan's DCA was the first national reimbursement scheme to link additional hospital reimbursement to structured multidisciplinary dementia care. The incomplete coverage observed in this study suggests that financial incentives may promote organizational adoption of such care but may be insufficient to ensure uptake among eligible patients without adequate workforce capacity and implementation support. Ongoing monitoring of certified hospitals, DCA coverage, and care processes will be important to understand implementation. Medication Use and Prescribing Trends Antipsychotics, benzodiazepines, and Z‐drugs are associated with adverse events in older adults [32, 33]. In acute‐care settings, these risks are amplified by illness severity and environmental stress. In this study, hypnotic use during hospitalization and at discharge declined over time. Although absolute reductions were modest, they may still be clinically meaningful in this high‐risk population and align with efforts to promote safer pharmacotherapy in older adults [24, 25]. However, the proportion of patients receiving antipsychotics during hospitalization remained high, exceeding 30% throughout the study period. This level is substantially higher than that reported in outpatient or long‐term care settings (17.9%–21.3%) [34, 35], which may reflect the challenges of managing BPSD in acute‐care hospitals. The proportion of discharge prescriptions (approximately 9%) was comparable to nationwide estimates from linked medical and long‐term care databases [36]. Antipsychotics initiated during hospitalization are often continued after discharge [5]; these findings underscore the importance of strengthening early non‐pharmacological and environmental approaches in acute care, including person‐centered behavioral management, delirium prevention, staff education, and family engagement [37]. Patients' Clinical Events LOS and Mortality LOS decreased from 33.7 to 27.8 days, and in‐hospital mortality declined modestly (13.9%–12.9%), likely reflecting broader system‐level reforms in Japan, including policies promoting early discharge, functional differentiation of hospital beds, and development of a community‐based integrated care system [38]. Despite these improvements, LOS remained substantially longer than the national acute‐care average of 16 days in Japan [39], underscoring persistent challenges in discharge coordination, multimorbidity management, and post‐acute care planning for patients with advanced dementia. In‐Hospital Fractures The incidence of in‐hospital fractures (0.9%–1.0%) was relatively high compared with previous DPC‐based studies. In a nationwide cohort of patients with neurocognitive disorders or limited ADL independence, the incidence was 0.12% [40]; among patients aged ≥ 50 years with dementia‐related diseases, it was 0.59% [41]. As prior studies included patients with mild cognitive impairment, the higher proportion in our cohort (restricted to moderate‐to‐severe dementia) is clinically plausible. These findings highlight the need to strengthen inpatient safety management and fall prevention for patients with dementia. ADL and Discharge Destination More than 70% of patients remained totally dependent at discharge, indicating limited ADL recovery during acute hospitalization. This likely reflects both low baseline function and constrained rehabilitation opportunities in acute‐care settings. The proportion discharged to other hospitals was higher than the proportion admitted from other hospitals (12.8%–26.6%), and about half of those admitted from home were not discharged home, suggesting that severe frailty and cognitive impairment often necessitate continued institutional care beyond the acute phase. Previous studies in Japan showed that cognitive impairment and functional decline strongly predict inter‐hospital transfer and reduced home discharge among older inpatients [42]. Given the high functional dependence at discharge, these patterns may reflect appropriate care transitions based on patient needs and caregiving capacity rather than adverse outcomes. Acute hospitalization may therefore serve as a transition point for reassessing post‐acute care needs and adjusting care settings. Taken together, these findings highlight the importance of multidisciplinary dementia care, improved medication and inpatient safety management, and better discharge planning and care transitions for patients with advanced dementia. The indicators presented here may also support monitoring of dementia care policy implementation and quality. Limitations This study has several limitations. First, the cohort was restricted to patients eligible for DCA, limiting comparisons with those with mild or no dementia. Second, the database lacks information on care content and quality; we could not assess non‐pharmacological interventions, rehabilitation intensity, family involvement, or discharge‐planning discussions. Third, the large sample size may have made small differences statistically significant; p‐values should therefore be interpreted together with standardized differences, absolute changes, 95% confidence intervals, and clinical relevance. Finally, ADL and CDED are observer‐rated and may be subject to inter‐rater variability and misclassification. Despite these limitations, this nationwide descriptive analysis provides a comprehensive overview of real‐world clinical profiles of patients targeted by Japan's DCA during early implementation, offering context for interpreting dementia care policies and informing future evaluations. Conclusion DCA‐eligible inpatients in Japan were characterized by advanced age, severe functional dependence, and high clinical complexity. Despite increased DCA uptake during early implementation, coverage remained partial, and challenges persisted in medication management, inpatient safety, and post‐acute care transitions. Although LOS, mortality, and hypnotic use declined, persistent functional dependence and in‐hospital fractures indicate high vulnerability, while discharge patterns suggest substantial post‐acute care needs. These findings provide nationwide baseline evidence to inform dementia care policy refinement and future evaluation. Author Contributions J.W. conceived and designed the study, performed data cleaning and preprocessing, conducted the statistical analyses, interpreted the results, and drafted the manuscript. K.M. provided substantial methodological guidance throughout the study, including study conception, analytic strategy, statistical analysis, and interpretation of findings, and critically revised the manuscript. H.Y., N.Y.‐M., A.I., T.K., and T.Y. contributed to the interpretation of results and provided important intellectual input through critical review of the manuscript. K.F., H.M., and Y.K. were responsible for data extraction, management, and provision, and contributed to ensuring data quality. All authors reviewed the manuscript, provided critical feedback, and approved the final version for submission. Funding This work was supported by the JST SPRING, JPMJSP2108; the Ministry of Health, Labour and Welfare, Japan, 23AA2003 and 24AA2006. Conflicts of Interest The authors declare no conflicts of interest. Supporting information Table S1: Detailed definitions of hospital variables and patient variables.