Background: Acute kidney injury (AKI) is common among critically ill patients. Continuous renal replacement therapy (CRRT) is often preferred to intermittent modalities in hemodynamically unstable patients; however, premature circuit clotting and shortened filter lifetime may reduce treatment efficiency and increase clinical burden. In Japan, multiple commercially available hemofilters are used for CRRT, yet randomized evidence directly comparing specific hemofilters in the intensive care unit (ICU) setting remains limited.

Methods: The FILTER CHOICE Trial is a multicenter, randomized, open-label, parallel-group study conducted across five ICUs in Japan. Adult patients with Kidney Disease: Improving Global Outcomes (KDIGO) stage 3 AKI requiring CRRT will be randomized (1:1) to receive either a polymethyl methacrylate hemofilter (Hemofeel® CH-1.8W) or polysulfone hemofilter with embedded hydrophilic NV polymer (Hemofeel® SNV-1.0). The assigned hemofilter will be used for up to three consecutive sessions (maximum 72 h per session). The primary endpoint is hemofilter (circuit) lifetime during the first CRRT session after randomization. Secondary endpoints include hemofilter lifetime in subsequent sessions, vasopressor- and ventilator-free days, renal recovery, major adverse kidney events, mortality, biomarkers, hemodynamic indices, and 6-month health-related quality of life. The target sample size is 88 patients (44 per group).

Conclusions: This trial will evaluate whether the choice of hemofilter (Hemofeel® CH-1.8W vs. Hemofeel® SNV-1.0) affects circuit life and clinically relevant outcomes during CRRT in the ICU, providing evidence to inform hemofilter selection in routine practice.

Logistic regression is commonly used to model binary outcomes because it restricts predicted probabilities to be between zero and one. Odds ratios are often reported in published papers, but it is challenging to interpret the effect of a change in an explanatory variable on the outcome and the magnitude of an odds ratio cannot be compared across studies because they are scaled by a factor that is undefined. One consequence of this scaling is that the magnitude of the odds ratio depends on the variables included in the model, something that does not happen in linear regression or marginal effect measures (risk differences, risk ratios). Instead of reporting odds ratios, we recommend reporting marginal effects, such as risk differences or risk ratios. This paper summarizes different ways to report results from logistic regression and makes recommendations for best practice.

Globally, health systems are underperforming across four dimensions: effectiveness, efficiency, equity, and responsiveness of the outputs produced in terms of public health and individual medical services. Despite rising health expenditures and continued scientific innovation, analysis of care cascades for hypertension and diabetes show that only a minority of affected individuals achieve clinical control across all country income groups. This underperformance stems from two structural challenges: innovation misalignment, in which delivery models have not evolved to match scientific advances and innovation, and institutional logic (i.e., the inherited structures and incentives of health systems), in which systems designed for managing acute infectious diseases, and maternal and child health struggle to manage chronic illness. Artificial intelligence (AI) provides an opportunity to address these shortcomings. Evidence from randomized controlled trials demonstrates that AI can improve cancer detection to improve effectiveness, reduce documentation burden to raise efficiency, close screening gaps in underserved populations to enhance equity, and optimize emergency department throughput to increase responsiveness. However, algorithmic bias threatens to widen health disparities if left unaddressed. This seminar presents a framework for AI’s role in health system transformation, reviews current evidence, and proposes a 6A framework in relation to critical considerations (Affordability, Accessibility, Applicability, Algorithmic bias, Agility, Accountability) for scaling AI equitably. The transformational potential of AI will be realized only if regulation keeps pace with innovation and deliberate efforts are made to ensure that all populations benefit.

BACKGROUND: Early diagnosis of valvular heart disease is critical. Recently, a palm-sized visualized phonocardiogram device (AMI-SSS01) equipped with artificial intelligence was developed.

METHODS: We initiated a three-year prospective cohort study in fiscal year 2024 (April 2024 to March 2025) to evaluate the feasibility of this method in a health checkup setting, including the completion rate and safety of phonocardiography. We collected phonocardiogram and bipolar single-lead electrocardiogram data, along with artificial intelligence-based assessments, using the AMI-SSS01. Outcomes were categorized as no abnormality, mild abnormality, requiring further examination, or requiring continued medical follow-up, based on AI/staff assessments and participants’ self-reported cardiac history. Participants requiring further examination were referred to a hospital for echocardiography. This report summarizes the findings obtained during fiscal year 2024.

RESULTS: A total of 342 participants (43.6% men; mean age, 68.7 years) underwent phonocardiography without difficulties in device attachment, device malfunction, or adverse events, including contact dermatitis. Two participants were categorized as “requiring continued medical follow-up,” both of whom self-reported aortic stenosis. Twenty-three participants, who were generally older, were categorized as “requiring further examination,” and 11 of them underwent echocardiography. Among these, the artificial intelligence correctly identified all four diagnosed cases (one case each of moderate aortic stenosis, moderate mitral regurgitation, severe aortic regurgitation, and mild-to-moderate aortic regurgitation) as moderate or severe valvular heart diseases.

CONCLUSIONS: Phonocardiographic screening using this device is technically feasible in a health checkup setting. Artificial intelligence-based assessments may aid in identifying individuals who are likely to have clinically significant valvular heart diseases.

Sweden has a long history of collecting population-based administrative and healthcare data and using them for research. These data constitute a central resource for research, enabling population-wide studies with decades of follow-up, which would have been difficult to conduct otherwise. This paper provides an overview of Swedish register data commonly used in health research, along with ethical considerations and the process for accessing these data. Key national data include non-health registers managed by Statistics Sweden and the National Board of Health and Welfare for healthcare, encompassing deaths, and social care service data. National quality registers offer disease-specific information to improve patient outcomes, regional healthcare databases cover healthcare data not included in national data, and datasets maintained by other governmental organisations provide unique information, such as nationally notifiable infectious diseases, military conscription assessments of young men from 1969, crime records and road and traffic data, including accidents. Strengths of Swedish register-linkage data include near-complete population coverage, deterministic linkage using personal identity numbers and rich non-health information, including but not limited to demographic, socioeconomic and geographic information; educational history; use of social benefits such as sick leave and livelihood support; and generational and family ties. Limitations such as incomplete primary care coverage in national registers and variations in data quality across sources were noted. Overall, Swedish register data offer exceptional opportunities for health-related research, although limitations and characteristics of source data need to be understood, and ethical approval must be obtained for application in research.

Background

The incidence of pulmonary embolism (PE), which can be life-threatening emergencies, is increasing. Several recent clinical studies investigating PE have employed Japanese administrative data. This study aimed to evaluate the validity of Japanese outpatient and inpatient claims data for identifying patients with acute PE.

Methods

Claims data were provided by the RWD database. The reference standard consisted of symptomatic/asymptomatic patients with acute PE enrolled in the COMMAND VTE Registry, a cohort study of patients with symptomatic venous thromboembolism in Japan. Registry data and claims were linked deterministically using patient identifiers. The algorithms were developed based on diagnosis and treatment; anticoagulation therapy, thrombolytic therapy, and inferior vena cava filter placement. Diagnosis was defined using codes in claims (SB; syobyo, SY; syobyomei, and BU; shindangun-bunrui).

Results

The algorithm based on diagnosis defined as ‘SB or SY or BU’ showed a sensitivity of 95.0% (267/281, 95% CI, 91.8–97.2), and a positive predictive value (PPV) of 16.6% (267/1, 611, 95% CI, 14.8–18.5). Algorithms that included diagnosis and treatment information had a higher PPV, but lower sensitivity. This was partly due to the reference standard including PE patients who did not receive treatment. Consequently, subgroup analyses restricting references to those who received the corresponding treatment demonstrated a sensitivity of 84.1% (37/44) and a PPV of 74.0% (37/50) for the algorithm based on diagnosis and thrombolytic therapy defined as ‘SB or SY or BU’.

Conclusions

The algorithms derived from Japanese claims data were highly sensitive for identifying inpatients and outpatients with acute PE.

Background: Functional decline is a sequela of COVID-19, and immobility during hospitalization further increases the risk in older individuals. In Japan, all individuals with COVID-19 aged ≥ 65 years were hospitalized, irrespective of disease severity. This study aimed to describe functional decline among older patients admitted for mild COVID-19, in the context of changes in viral variants, patient severity, vaccination rollout, and treatment practices over time.

Methods: The Diagnosis Procedure Combination data recorded between April 2020 and March 2022 in Japan were analyzed. The study enrolled patients aged ≥65 years hospitalized for mild COVID-19, defined as no requirement of respiratory support on admission. The incidence of exacerbation, defined as the initiation of supplemental oxygen or mechanical ventilation during hospitalization, and adverse events were described, stratified by six pandemic waves. Functional decline in the subset of patients without moderate or severe symptoms was also examined.

Results: Among 58,176 patients, the proportions of patients requiring supplemental oxygen, mechanical ventilation, intensive care unit admission, and in-hospital mortality decreased over waves (p <0.001). Possible hospitalization-related adverse events, including fracture, delirium, and functional decline, also decreased over waves (p <0.001). However, among 45,206 patients who did not experience exacerbation, the proportion of patients with functional decline remained stable over waves (p = 0.47).

Conclusion: While exacerbation was absent in most patients with mild COVID-19, a substantial proportion experienced functional decline. Careful consideration of hospital admission criteria and preventive measures may be required in older patients with mild COVID-19 to mitigate hospitalization-related functional decline.

Background: The effect of early surgery for hip fractures may vary depending on patients’ pre-fracture physical and cognitive function. We aimed to examine whether the association between early surgery for hip fracture and one-year mortality differs by patients’ pre-fracture functional status, assessed using care need levels defined by Japan’s nationally standardized certification system.

Methods: This retrospective cohort study used medical and long-term care insurance databases from Tochigi Prefecture, Japan. Inpatients with hip fractures aged ≥ 65 years between June 2014 and October 2018 were included. Patients were categorized into no care needs or one of the seven care need levels (support level 1–2 or care need level 1–5), each corresponding to specific estimated daily care minutes. Early surgery was defined as surgery on day 1 or 2 of admission, and delayed surgery as surgery between days 3 and 7. Propensity score weighting and logistic regression were used to compare one-year mortality between early and delayed surgery groups in the overall cohort and within each pre-fracture care need level.

Results: Among 922 patients in the early surgery group and 2,680 in the delayed surgery group, adjusted one-year mortality was 16.2% and 14.4%, respectively (risk difference, 1.8%; 95% confidence interval, -1.2 to 4.8%). No significant differences in one-year mortality were observed between early and delayed surgery groups within any pre-fracture care need level stratum.

Conclusions: Early surgery was not associated with reduced one-year mortality in patients with hip fractures, either overall or when stratified by pre-fracture care need level.

Administrative claims data have become central resources for epidemiology and health services research in Japan. These datasets provide large sample sizes, longitudinal coverage, and standardized coding, and are increasingly treated as “ready-made” sources for descriptive, causal, and predictive research. However, they are by-products of an insurance and reimbursement system, not purpose-built research registries. What is captured, and what is missing, is determined by institutional arrangements, fee schedules, and clinical workflows. Without understanding how these data are generated and processed, investigators risk misinterpreting variables, underestimating bias, and over-trusting sophisticated analyses.

In this article, we argue that rigorous research using Japanese claims data must begin with an explicit understanding of data provenance—how and why information enters the database, and data processing —how heterogeneous records are transformed into analytic datasets. We outline key elements of provenance for Japanese health insurance claims, including insurer types, separation of medical and dispensing claims, and the evolution of coding and reimbursement rules. We then describe typical processing steps required to convert raw claims files into analysis-ready datasets, namely parsing file structures, constructing relational databases, linking to code master tables, and building cohorts tailored to specific research questions.

By making provenance and processing clearer, researchers can better recognize limitations, design appropriate epidemiological studies, and interpret findings in light of how the data came to exist. Rather than treating claims data as neutral “healthcare data”, this perspective highlights their origins as administrative records and emphasizes that understanding their generation and transformation is a prerequisite for valid epidemiology.

Background: This study retrospectively evaluated the user experience and safety outcomes of medical chatbot service providing medicine use information to pregnant and lactating women.

Methods: The chatbot utilized in this study did not employ generative artificial intelligence (such as large language models). Rather, it was a rule-based system in which all response patterns were predefined. We collected user survey data between December 13, 2021, and August 10, 2022. User status, satisfaction, addressed consultation categories, and post-use health problems were analyzed.

Results: The study included 1,000 records, comprising 419 (41.9%) cases of pregnant women and 581 (58.1%) cases of lactating women across Japan. The overall satisfaction was 93.9%, with no significant difference between pregnant and lactating women (P = .866). Headache was the most frequently consulted category followed by cold symptoms and hay fever symptoms. High satisfaction levels were observed across the consultation categories. None of the participants indicated post-use health problems.

Conclusion: A medical chatbot service providing medicine use information to women during pregnancy and lactation can achieve high satisfaction and safety.

This narrative review summarizes the concepts of low- and no-value care (LNVC), the drivers behind them, and potential strategies for de-implementation, all with a particular focus on influential conceptual and empirical research. Low-value care (LVC) refers to healthcare services for which the costs outweigh the potential benefits, while no-value care (NVC) is a narrower definition that refers to healthcare services that provide no net clinical benefit for patients (regardless of the costs). A growing global body of research shows that LNVC is widespread and imposes substantial societal costs. Mitigating the provision of such care would reduce avoidable healthcare expenditures, improve the quality and safety of treatment by minimizing overdiagnosis and overtreatment, and improve overall health by reallocating resources to high-value care. LNVC arises from interrelated factors at the system, individual-provider, and consumer-demand levels, including system incentives, industry promotion, patient expectations, and physicians’ knowledge, habits, and fears. Despite increasingly intense efforts to address LNVC, several challenges persist in terms of identification, measurement, and intervention. Multicomponent interventions targeting healthcare providers have been shown to be effective at reducing LNVC from the supply side; such interventions may include recommendations from professional societies, quality metrics for LNVC, quality-improvement initiatives, clinical decision-support tools, and payment reforms (e.g., global budgets, prior authorizations, coverage policy). Given the uneven provision of LNVC, targeted strategies centered on high-volume providers could improve the efficiency of interventions. Importantly, eliminating NVC represents a clear and actionable starting point to facilitate coordinated, multi-level efforts and collaboration among a diverse range of stakeholders.

Background: The incidence of end-stage kidney disease (ESKD) caused by chronic kidney disease is increasing worldwide. Patients with ESKD often exhibit significant premature aging and have a higher risk of premature mortality than the general population. Here, we conducted a prospective cohort study, the Kumamoto Hemodialysis Cohort Study, to explore the mechanisms of the premature aging phenotype exhibited by these patients using biological markers and highly detailed clinical information in a real-world setting.

Methods: This multicenter prospective cohort study targeted adult outpatients receiving maintenance hemodialysis therapy in Kumamoto Prefecture, Japan. After enrollment, the patients were followed up for 10 years. Data were collected from the electronic medical records of medical facilities that participated retrospectively and prospectively from January 1, 2017. Clinical information was collected from electronic medical records, medical chart reviews, and patient-completed questionnaires. Serum and plasma samples were collected every 3 years. Thus far, 1,241 patients have been enrolled in this study, and we found the median age of the participants was 69 years, and 67.5% were male. Among the primary causes of kidney failure, diabetic nephropathy accounted for 38.6% of all cases, followed by glomerulonephritis (27.2%) and nephrosclerosis (9.9%). The median hemodialysis vintage was 5.8 years.

Conclusion: This study contributes to the characterization of a premature aging phenotype together with morbidity and mortality of patients with ESKD, thereby providing insights into the prevention of the onset of premature aging-related diseases in patients receiving maintenance hemodialysis therapy.

Background: This study aimed to validate self-reported fractures in the Japan Multi-Institutional Collaborative Cohort (J-MICC).

Methods: Of the 10,332 participants who were followed up for 10 years, 1,137 individuals with self-reported fractures were included. Among them, 201 individuals were randomly selected, and 192 individuals with available medical records were analyzed. A false-positive fracture was defined as a self-reported fracture without a clinical diagnosis or record. We calculated the positive predictive value (PPV) stratified by the number of years from injury to response. Multivariate logistic regression analysis was used to elucidate the factors influencing false-positive fractures and to calculate the odds ratios (ORs) with 95% confidence intervals (CIs). The dependent variable was false-positive fractures, and the independent variables included factors such as fracture site, age, sex, body mass index, education, multisite fractures, multiple fractures, and number of years from injury to response.

Results: The average age at the time of injury was 69.6 years (range, 50–80 years), and 48 individuals (25.0%) were male. Fractures were reported in the spine (41 individuals, 21.4%), upper limbs (57 individuals, 29.7%), lower limbs (71 individuals, 37.0%), and other locations (23 individuals, 11.9%), respectively. The PPV was 85.4% (164/192 individuals; 95% CI, 79.7–89.7) and was > 90% if the period from injury to response was within 3 years. False-positive fractures were significantly associated with a period from injury to response >3 years (OR, 3.46; 95% CI, 1.43–8.37).

Conclusions: Self-reported fractures are useful outcomes; however, recall periods >3 years may increase the risk of misclassification.

Background: This study aimed to evaluate the association of the initial body weight-stratified cefazolin antimicrobial prophylaxis (AMP) dose and surgical-site infection (SSI) in major orthopedic implant surgery.

Methods: This nationwide, retrospective cohort study used data from the Diagnosis Procedure Combination database. We included patients (age ≥18 years) who underwent major orthopedic implant surgery between July 2010 and March 2022. Patients underwent surgery with an initial cefazolin dose of either 1 g (1 g cefazolin group) or 2 g (2 g cefazolin group). Primary outcome was in-hospital SSI. Restricted cubic spline functions were used to estimate the nonlinear association of body weight-stratified cefazolin dose with the outcome and between cefazolin dose per weight and the outcome.

Results: Among the 408,487 participants, 394,731 and 13,756 were in the 1 g and 2g cefazolin groups, respectively. The 1 g cefazolin group, but not the 2 g cefazolin group, showed a J-shaped relationship between body weight and SSI, which increased continuously beyond 50 kg, in an increased risk of SSI with body weight. A reverse J-shaped relationship was observed between cefazolin dose per weight and SSI, with an increased risk of SSI at <20 mg/kg.

Conclusion: An increased risk of SSI was observed with 1 g cefazolin, but not with 2 g cefazolin, as an initial AMP with higher body weight among adults weighing >50 kg.

Background

In 1986, a strategy of vaccinating infants at 2, 3, and 5 months of age was implemented in Japan to prevent the vertical transmission of the hepatitis B virus. This strategy was revised in 2013 to vaccinate infants at birth, 1 month, and 6 months of age. We aimed to evaluate the effect of this policy change on the vaccination status of infants born to mothers with positive hepatitis B surface antigen.

Methods

We identified infants born between April 2009 and March 2021 from the JMDC Claims Database. Administration of hepatitis B immunoglobulin at birth was used to identify infants who received preventive measures against vertical transmission. The timing of hepatitis B vaccine administration within 12 months of birth was compared between old (born between April 2009 and September 2013) and new (born between October 2013 and March 2021) groups.

Results

Among the 77,797 and 510,584 infants born before and after October 1 2013, respectively, 190 (0.24%) and 732 (0.14%) were analyzed. The proportion of infants receiving at least one dose of the vaccine increased from 63% to 96% (p<0.001). The proportion of infants who received timely first, second, and third doses of vaccines changed from 54% to 90% (p<0.001), 46% to 59% (p=0.002), and 39% to 31% (p=0.034), respectively.

Conclusions

The change in strategy against the vertical transmission of hepatitis B virus resulted in an increased rate of receiving the first dose of the vaccine and a decreased rate of timely completion of the recommended doses.

The DeSC database, launched in 2020 by DeSC Healthcare Co., Ltd. is a commercially available Japanese administrative medical claims and health checkup database. Recently, long-term care insurance (LTCI) claims data—covering care-need levels, monthly care benefits, service utilization, and cost information—have been integrated into the database. We showed that the distributions of the seven care-need levels were highly similar between the DeSC database and the official report by the Ministry of Health, Labour and Welfare. This similarity supports the representativeness and reliability of the DeSC database for epidemiological analyses using care-need status, particularly in the Advanced Elderly Medical Service System. This integrated dataset bridges critical information gaps and offers substantial advantages for health services research and epidemiological studies through individual-level linkage of medical, pharmacy, and LTCI claims. This study provides an overview of the LTCI claims data and available data items in the DeSC database and discusses future research prospects, with a particular focus on the linkage between medical and LTCI claims.