Tag: statistics

JMIR Res Protoc. 2026 Jul 7;15:e91051. doi: 10.2196/91051.

ABSTRACT

BACKGROUND: High mortality and prevalence rates are hallmarks of chronic heart failure (CHF). Patients frequently have a much lower quality of life as a result of diminished exercise tolerance. Chinese guidelines have recommended Yiqi Fumai lyophilized injection (YQFM) for the treatment of heart failure, although there is currently inadequate evidence to support its effectiveness in increasing exercise tolerance in these patients.

OBJECTIVE: The purpose of this cohort study is to examine the relationship between the improvement of exercise tolerance in patients with CHF and the addition of YQFM to guideline-directed medical therapy.

METHODS: In total, 216 hospitalized patients with CHF with New York Heart Association (NYHA) functional classes II-IV were to be enrolled in the prospective, observational cohort trial design. The participants were divided into a YQFM group (exposed group: n=144) and a non-YQFM group (nonexposed group: n=72) at a 2:1 ratio based on real clinical medication and patient preference. Standard guideline-directed medical therapy was administered to both groups; however, the YQFM group also got a 10-day YQFM exposure. The change in metabolic equivalents measured by the Veterans Specific Activity Questionnaire was the main outcome. The 6-minute walk distance, Kansas City Cardiomyopathy Questionnaire score, NYHA functional class, N-terminal pro-B-type natriuretic peptide levels, and echocardiographic parameters were among the secondary objectives. Traditional Chinese medicine syndrome scores and the frequency of hard clinical occurrences were the exploratory objectives. A linear mixed-effects model was used to examine repeated measurement data, and propensity score weighting was used to account for baseline confounding variables.

RESULTS: The first patient was registered in June 2024, and all 216 patients had been recruited and followed up with by December 2025. Data cleaning and statistical analysis began in January 2026, with final results scheduled to be published in the autumn of 2026.

CONCLUSIONS: This study, using a prospective cohort design, intends to offer high-quality, real-world evidence for the use of YQFM in cardiac rehabilitation for heart failure. This will aid in the optimization of preventative and treatment strategies for CHF that combine traditional Chinese and Western medicine, giving an objective basis for enhancing patients’ long-term quality of life.

PMID:42413004 | DOI:10.2196/91051

Braz J Psychiatry. 2026 Jul 7. doi: 10.47626/1516-4446-2026-4868. Online ahead of print.

ABSTRACT

BACKGROUND: Brazil’s Unified Health System (SUS) mental inpatient care relies on government funding, especially in structurally unequal regions. Despite funding concerns, general hospitals manage serious mental disorders strategically.

OBJECTIVE: Determine if federal transfers cover the true cost of psychiatric hospitalizations in a Northern Brazilian general hospital.

METHODS: This cross-sectional cost analysis examined all 2023 psychiatric admissions in the Dr. Emílio Vasques Psychiatric Unit, General Hospital of Palmas, Tocantins. While the institution has 22 inpatient beds, only 11 are registered and eligible for federal funding by the Ministry of Health. Direct and indirect costs were evaluated using microcosting and absorption costing. Official national databases provided federal transfer statistics.

RESULTS: 421 hospitalizations. Annual costs was BRL 13.984.390,24 (USD 3,107,642.28). Spending was 44.6% direct and 55.4% indirect. The average hospitalization cost BRL 33.217,08 (USD 7,381.57). Federal transfers covered 12% of direct hospitalization expenditures, half the unit’s operational capacity.

CONCLUSION: Federal funding for mental inpatient care is insufficient, transferring the financial burden to state governments and compromising service viability.

PMID:42412980 | DOI:10.47626/1516-4446-2026-4868

NEJM Catal Innov Care Deliv. 2026 Feb;7(2):CAT250237. doi: 10.1056/CAT.25.0237. Epub 2026 Jan 21.

ABSTRACT

Vitamin D testing is frequently overused across U.S. health systems, with much of it occurring outside the bounds of evidence-based clinical indications. Despite clear guidance from the Endocrine Society and the U.S. Centers for Medicare and Medicaid Services, routine screening in low-risk populations remains common and costly. Within Oregon, vitamin D testing has been identified as one of the top 15 low-value services, representing a substantial opportunity for cost savings and practice alignment. In an effort to reduce the volume of low-value vitamin D testing across a three-hospital academic health system, a multidisciplinary working group designed and implemented an electronic health record best practice advisory to fire at the point of order entry when a vitamin D test was attempted without an approved indication. The initiative involved implementing a targeted clinical decision support intervention, aligned with evidence-based diagnostic coding curated using the U.S. Centers for Medicare and Medicaid Services coverage criteria and reinforced by coordinated communication and measurement infrastructure. The proportion of indicated vitamin D tests increased from 45% (939 of 2060) in the final month pre intervention to more than 88% (1194 of 1354) at 90 days post intervention, and remained above 90% (1381 of 1531) at 1 year post go-live. Improvements were consistent across payer types, age groups, and locations of care, with the largest gains seen in outpatient and Medicaid populations. In absolute terms, the number of vitamin D tests decreased by 25.3% (to 16,530 from 22,129) when comparing the 12 months pre intervention with the 12 months post intervention. Along with this decrease was an estimated cost savings of nearly US$112,000. The intervention significantly improved ordering appropriateness while maintaining access to necessary testing. It required no additional staffing or investment in new infrastructure. Provider feedback highlighted the value of peer engagement, transparency in metric definitions, and responsiveness to edge cases. Data revealed attempts to bypass the best practice advisory using unspecified diagnosis codes. A simple, well-targeted clinical decision support tool, paired with strategic communication and disciplined measurement, can lead to meaningful reductions in low-value vitamin D testing. Sustained improvement depends on transparent metrics, engaged leadership, and a willingness to challenge entrenched ordering habits while remaining attentive to clinical nuance.

PMID:42412971 | DOI:10.1056/CAT.25.0237

NEJM Catal Innov Care Deliv. 2026 Feb;7(2):CAT250205. doi: 10.1056/CAT.25.0205. Epub 2026 Jan 21.

ABSTRACT

Emergency department (ED) crowding, constrained inpatient capacity, and long waits for outpatient cardiology delay care for patients with urgent cardiac symptoms. In 2016, the Beth Israel Deaconess Medical Center opened a non-ED-based, cardiologist-staffed Cardiac Direct Access Clinic with examination rooms, an infusion room, and six overnight observation beds to provide rapid specialty evaluation and short-stay care. Using administrative data and contribution-margin analyses, the authors assessed operational, clinical, and financial outcomes and summarized implementation strategies. Of 11,121 total patients seen in the clinic, 4239 patients – those most likely to have otherwise been sent to the ED – were admitted on the same day of referral. Of those patients, 59% were discharged home, 39% were managed in the Cardiac Direct Access Clinic’s overnight unit, and 7% ultimately were admitted to inpatient floors. Among 1467 patients discharged from the overnight unit, the 30-day return rate to the ED was 6.4%. Overall patient experience scores were higher for the clinic than for the ED – 84.7 versus 56.9. Annual labor and supply costs for the clinic totaled approximately US$1.8 million. The contribution margin derived from the clinic’s operations (US$245,000), admissions originating from the clinic (US$303,000), and inpatient capacity created (US$1.34 million) produced an estimated US$2.4 million annual contribution margin, underscoring financial sustainability. Key enablers included centralized prior-authorization teams, state approval to operate as an alternative care space for inpatient-level services, and codified diversion pathways for the ED and urgent care.

PMID:42412968 | DOI:10.1056/CAT.25.0205

J Med Internet Res. 2026 Jul 7;28:e86087. doi: 10.2196/86087.

ABSTRACT

BACKGROUND: Limited data availability and privacy constraints hinder the development of robust survival prediction models for personalized treatment. Synthetic data offers a promising solution, preserving the statistical properties of real clinical data.

OBJECTIVE: This study aimed to quantitatively assess the feasibility of using synthetic data for survival prediction by evaluating model transfer performance to real-world hospital data, with a focus on model transfer strategies.

METHODS: We developed and validated colorectal cancer survival prediction models using the National Cancer Data Center (NCDC) synthetic data (30,683 patients from 3 Korean institutions) for pretraining and real hospital data (2170 patients from Hwasun Jeonnam University Hospital) for external validation. We evaluated 3 model transfer strategies-domain adaptation, zero-shot, and ensemble-using extreme gradient boosting (XGBoost) and light gradient boosting machine (LightGBM). In total, 48 model configurations were tested, defined by the combination of algorithms (LightGBM and XGBoost), sampling technique (no-sampling, random undersampling [RUS], and synthetic minority oversampling technique combined with edited nearest neighbors [SMOTEENN]), model type (baseline, domain adaptation, zero-shot, and ensemble), and optimization objective (area under the precision-recall curve [AUPRC] and F1). The outcome was 7-year overall survival, evaluated using the AUPRC and Brier scores. Performance was compared against a hospital-only baseline using absolute values and deltas (ΔAUPRC and ΔBrier). Differences and corresponding 95% CIs were estimated on the held-out test set using 2000 bootstrap samples.

RESULTS: Zero-shot application reduced the AUPRC in most settings, and any marginal improvements observed in the remaining settings were not statistically significant. In contrast, the domain adaptation model improved AUPRC in 8/12 combinations, with 4 statistically significant gains; the best setting (XGBoost+RUS+F1 optimization) achieved AUPRC=0.5391 (Δ+0.1474; P<.001). The soft ensemble increased AUPRC in 7/12 combinations, with 3 statistically significant gains; the best setting (XGBoost+RUS+AUPRC optimization) achieved AUPRC=0.5060 (Δ+0.1258, P=.002). For calibration, Brier scores improved in most domain adaptation and ensemble combinations, with a substantial proportion reaching statistical significance.

CONCLUSIONS: When domain adaptation using local hospital data was applied, the model pretrained on synthetic data exhibited similar performance to the hospital-only baseline across various settings. This study demonstrates the methodological utility of a model transfer approach using NCDC synthetic data in a setting with limited data sharing. At the same time, it clarifies that while synthetic data can serve as a complement to local clinical data, it is not a substitute for real-world clinical models.

PMID:42412950 | DOI:10.2196/86087

JMIR Med Inform. 2026 Jul 7;14:e85015. doi: 10.2196/85015.

ABSTRACT

BACKGROUND: The increasing documentation burden on physicians is a significant contributor to burnout and decreases in care quality. Artificial intelligence (AI) has been proposed as a solution to reduce documentation burden in clinical care, but there are very limited data on its use in the inpatient and intensive care unit (ICU) environments.

OBJECTIVE: This pilot study aimed to explore the feasibility of using AI-assisted ambient scribes to capture interprofessional ICU rounds and synthesize a singular document to improve documentation efficiency and clinician satisfaction during ICU rounds. In this paper, we showcase our findings from customizing prompts for large language models (LLMs) to generate and evaluate daily progress notes from transcripts of simulated ICU cases.

METHODS: This project is divided into 2 phases. In the first phase, a randomly selected transcript of an audio recording of a simulated ICU rounds case was used to iteratively evaluate and improve the prompts for the LLMs. Multiple models (n=5) were used in phase 1, and the best-performing model (M1, based on the highest accuracy) was selected for the next phase. In the subsequent phase, 5 cases were selected and evaluated using the refined prompt and 2 models: M1 from phase 1 and M6, a technological upgrade of M1. Accuracy and error percentages were used as primary metrics. Additionally, error severity and usability were assessed using the Harm scale (adapted for potential harm risk) from the Agency for Healthcare Research and Quality and the 9-item Physician Documentation Quality Instrument, respectively.

RESULTS: Iterative improvements to the prompt increased accuracy and reduced errors during phase 1. In phase 2, M1 and M6 achieved accuracies of 69% and 80%, respectively (P=.04). Overall, errors of omission were most common (mean 15.5%, SEM 2.7%), followed by partial errors (mean 7.2%, SEM 0.92%) and then errors of commission (mean 2.6%, SEM 0.7%). The error severity of both models was low (µ=0.61 vs 0.53; P=.10), with most errors categorized as having potential for no harm to low harm. Both models performed well on the 9-item Physician Documentation Quality Instrument assessment, with the M6 model outperforming the M1 (35.8 vs 38.3; P=.06).

CONCLUSIONS: Our findings demonstrate the feasibility of integrating AI-assisted scribes for ICU documentation. Both prompt improvements and technological advancements in LLMs are noted to be helpful. This study lays the groundwork for future research into AI applications in ICU settings, paving the way for broader improvements in health care documentation.

PMID:42412948 | DOI:10.2196/85015

Int J Qual Health Care. 2026 Jul 7:mzag094. doi: 10.1093/intqhc/mzag094. Online ahead of print.

ABSTRACT

BACKGROUND: As the elderly population in Korea rapidly increases, issues related to lonely death have also emerged as a social concern. The purpose of this study is to identify factors influencing the subjectively perceived risk of lonely death among older adults living alone.

METHODS: This study employed a cross-sectional design using secondary data from the seventh wave (2024) of the Seoul Senior Citizen Survey. A total of 955 older adults aged ≥65 years living alone in Seoul were included in the analysis. Descriptive statistics, Rao-Scott chi-square tests, F-tests, and complex sample multinomial logistic regression were conducted using complex sample procedures.

RESULTS: Significant differences across the lonely death risk groups (low, moderate, and high) were observed in relation to gender, education, household income, economic satisfaction, housing satisfaction, frequency of in-person contact, eating alone, satisfaction with social relationships and social/cultural activities, depression, history of falls, instrumental activities of daily living, subjective health, and preparedness for dying alone. Factors significantly associated with the subjectively perceived risk of lonely death included low housing satisfaction, eating alone, dissatisfaction with social relationships, and preparedness for dying alone. The perceived risk of lonely death among older adults living alone varies according to a range of social, economic, and environmental factors.

CONCLUSION: The results of this study may serve as foundational resources for future research and policy development aimed at preventing lonely death.

PMID:42412537 | DOI:10.1093/intqhc/mzag094

Med Teach. 2026 Jul 7:1-12. doi: 10.1080/0142159X.2026.2681971. Online ahead of print.

ABSTRACT

BACKGROUND: The recent advances in Generative Artificial Intelligence (GenAI), from task-specific assistants to autonomous agentic artificial intelligence (AI) are changing how research is conceived, conducted, and written. Across this spectrum AI can now assist with literature searches and synthesis, protocol drafting, statistical analysis, and manuscript preparation, particularly in computational domains. Yet AI outputs remain error-prone, opaque, and carry real stakes for patients, learners, and equitable outcomes, making strong foundational research skills more important than ever.

PURPOSE: This article offers practical guidance for medical educators responsible for research training in an AI-augmented environment.

TIPS: Drawing on published work on biomedical research competencies and emerging scholarship on AI in medical education, and our own experience, twelve tips are organized around three themes: understanding the changing AI landscape, protecting non-delegable human responsibilities, and teaching new AI-era competencies.

CONCLUSIONS: AI-augmented research does not reduce the need for research education; it changes which skills deserve the most attention. Medical curricula should now emphasize critical appraisal, ethical reasoning, verification of AI outputs, and assessment strategies that distinguish independent mastery from AI-assisted performance.

PMID:42412521 | DOI:10.1080/0142159X.2026.2681971

J Gerontol A Biol Sci Med Sci. 2026 Jul 7:glag175. doi: 10.1093/gerona/glag175. Online ahead of print.

ABSTRACT

BACKGROUND: Early-life undernutrition may increase susceptibility to chronic kidney disease (CKD), but whether adult metabolic burden modifies this association remains unclear. We examined early-life famine exposure and CKD among middle-aged and older Chinese adults using two complementary national cohorts.

METHODS: We analyzed 7,238 participants from the 2023 to 2024 China National Health Survey (CNHS) and 8,273 from the China Health and Retirement Longitudinal Study (CHARLS). In CNHS, famine exposure was defined by birth cohort and prefecture-level cohort size shrinkage index, with age-balanced non-famine births as the reference group. Prevalent CKD was evaluated using Firth logistic regression and difference-in-differences models, including joint associations with adult meat consumption. In CHARLS, self-reported famine severity, waist circumference trajectories, and incident CKD were evaluated using Cox models.

RESULTS: In CNHS, the famine-CKD association increased with regional famine intensity, with an interaction odds ratio of 1.04 per unit increase in cohort size shrinkage index (95% confidence interval, 1.01-1.07). Severe famine exposure combined with frequent meat consumption was associated with higher odds of prevalent CKD (odds ratio, 2.09; 95% confidence interval, 1.16-3.76). In CHARLS, severe famine exposure combined with a high-stable waist circumference trajectory showed the highest risk of incident CKD (hazard ratio, 1.75; 95% confidence interval, 1.05-2.91), with an increasing trend across famine severity mainly in this trajectory group. Cumulative diabetes partly mediated this association.

CONCLUSIONS: Early-life famine exposure was associated with higher CKD risk, particularly among individuals with unfavorable adult metabolic profiles, supporting life-course metabolic vulnerability in CKD risk stratification.

PMID:42412516 | DOI:10.1093/gerona/glag175

J Gerontol A Biol Sci Med Sci. 2026 Jul 7:glag176. doi: 10.1093/gerona/glag176. Online ahead of print.

ABSTRACT

BACKGROUND: Polypharmacy is common in older adults with comorbidities. This study aims to estimate its global prevalence, temporal trends, and dose-response associations with adverse health outcomes.

METHODS: PubMed, Embase, The Cochrane Library, Web of Science and Scopus were searched from inception to June 30, 2025. Random-effects meta-analyses were used to pool prevalence estimates of polypharmacy based on mixed thresholds and relative risks. Study quality was assessed using Hoy’s method. Hazard ratios (HRs) from individual studies were synthesised using random-effects dose-response meta-analysis in studies that provided sufficient quantitative exposure information.

RESULTS: A total of 545 studies, providing 565 prevalence estimates, were included, comprising 16,620,414 older adults from 56 countries worldwide. The pooled global prevalence of polypharmacy was 50.4% (95% confidence interval [CI]: 48.2-52.7%; 95% prediction interval: 6.8-93.5%). It demonstrated a significant increasing temporal trend in polypharmacy prevalence over calendar years. Compared with non-polypharmacy, polypharmacy was associated with a significantly higher risk of multiple adverse outcomes. Dose-response meta-analyses showed a monotonic increase in risk with each additional medication for mortality (HR: 1.05, 95%CI: 1.03-1.06, P＜0.001), emergency (HR: 1.02, 95%CI: 1.02-1.03, P＜0.001), and hospital admission (HR: 1.04, 95%CI: 1.02-1.07, P＜0.001). Outcome-specific thresholds ranged from 4.55 to 5.46 medications. A pooled estimate of approximately five medications was obtained as a descriptive summary across outcomes.

CONCLUSION: Polypharmacy affects approximately half of older adults worldwide and has increased steadily over time. It is associated with an increased risk of multiple adverse health outcomes, with risks increasing progressively as medication burden rises.

PMID:42412515 | DOI:10.1093/gerona/glag176