Clinical Epidemiology

Introduction: Acute compartment syndrome (ACS) is a limb-threatening complication of extremity trauma that requires timely diagnosis to prevent irreversible muscle and nerve injury. Current diagnostic methods are invasive, intermittent, and operator-dependent. We evaluated the feasibility of a novel, Bluetooth-enabled electrical impedance myography (EIM) device (mAlert, Myolex, Inc., Brookline, MA, USA) for continuous, noninvasive detection of ACS-related tissue changes. Methods: Ten Yorkshire swine underwent anterior tibial compartment monitoring using three ACS models: albumin infusion (ALB, n=3), femoral artery and vein ligation (LIG, n=3), and combined albumin infusion plus ligation (ALB+LIG, n=4). Resistance (R), reactance (X), and phase (P) were measured every minute across 1 to 199 kHz alongside continuous intra-compartmental pressure (ICP) monitoring. Group differences in normalized impedance trends were evaluated using the Kruskal Wallis test with Dunn post hoc correction. As a proof-of-concept human study, nine healthy volunteers wore the device for up to five days to assess electrode durability and signal stability. Tissue ischemia was validated using pimonidazole immunohistochemistry. Results: ALB infusion produced progressive, frequency-dependent decreases in R, X, and P, whereas LIG produced consistent increases in R and X across frequencies. The ALB+LIG model generated mixed responses, reflecting the competing effects of edema and ischemia. Normalized phase slopes differed significantly among groups (H=6.14, p=0.046), with post hoc testing showing significant divergence between the ALB and LIG models (p=0.041). Control limbs remained stable throughout monitoring. Pimonidazole staining confirmed hypoxic injury in the intervention limb. In the human pilot study, three participants completed five days of monitoring, demonstrating sustained signal acquisition, while electrode degradation limited data collection in the remaining participants. Conclusions: This preliminary feasibility study demonstrates that wearable EIM can continuously detect model-specific physiological changes associated with ACS in a large-animal model. These findings support further development and clinical evaluation of wearable EIM as a non-invasive monitoring technology for early ACS detection in trauma patients.

Importance: Adverse drug events in older adults are a substantial public health burden, yet spontaneous reporting systems detect them poorly owing to underreporting and the lack of a defined population. These limitations are of particular concern for older adults, who are underrepresented in pre-approval trials yet at elevated risk owing to polypharmacy, multimorbidity, and age-related changes in drug metabolism. Objective: To develop and apply an active, claims-based pharmacovigilance framework using sequential target trial emulation to detect adverse drug event signals in older adults, with atorvastatin as the initial application. Methods: Using Medicare fee-for-service claims (2017-2019), we studied statin-naive beneficiaries aged 65 years or older following myocardial or cerebral infarction. We emulated up to 14 daily sequential trials from the discharge date, classifying patients as initiating atorvastatin (A1), initiating a different medication (A2), or no new medication (A0); the primary contrast was A1 versus A2. For each trial, incident outcomes were ascertained and classified into 552 outcomes based on the Clinical Classifications Software Refined categories. Per-protocol effects were estimated over a 6-month follow-up period using Fine-Gray regression models weighted by the inverse probability of treatment and censoring, treating death as a competing risk, with the false discovery rate controlled via the Benjamini-Hochberg procedure. A signal was declared when the q-value was 0.10 or lower and the subdistribution hazard ratio (sHR) was 1.20 or greater in any prespecified analytic stratum (sensitivity analyses used thresholds of q 0.20 or lower and sHR 1.20 or greater). Results: Of 70,130 eligible patients, 39,948 initiated atorvastatin (A1) and 19,182 initiated another new medication (A2); after weighting, baseline characteristics were closely balanced. After excluding outcomes with sparse cell counts, 295 outcomes were analyzed; five met the primary signal detection criteria: valve disorders (sHR 1.71, 1.20 to 2.43); sprains and strains (sHR 1.79, 1.26 to 2.54); general sensation/perception symptoms (sHR 1.23, 95 percent CI 1.11 to 1.36); abnormal findings without diagnosis (sHR 1.55, 1.18 to 2.05); and prediabetes (sHR 1.71, 1.24 to 2.36). In the sensitivity analysis, we additionally detected posthemorrhagic anemia, hemorrhagic stroke, varicose veins, and other circulatory and skin conditions. Conclusions: An active, claims-based framework using sequential target trial emulation detected both expected and previously unrecognized adverse drug event signals following atorvastatin initiation in older adults, offering a systematic alternative to passive surveillance that can be extended to other commonly prescribed medications.

Background: Transthyretin amyloidosis cardiomyopathy (ATTR-CM) is an infiltrative cardiomyopathy and an increasingly recognized cause of heart failure. With improved survival from disease-modifying therapies, an increasing number of patients are presenting for surgery and may be at increased risk of adverse postoperative outcomes. This study reports outcomes of ATTR-CM patients undergoing surgery and evaluates the utility of the Revised Cardiac Risk Index (RCRI), a perioperative risk tool. Methods: A total of 145 ATTR-CM patients were included, among which 51 patients underwent at least one eligible surgical procedure. Preoperative risk was assessed using the RCRI, analyzed both as a categorical and as a dichotomized ({greater than or equal to}3 vs <3) variable. Postoperative outcomes included unplanned hospital admission, length of stay (LOS), prolonged hospitalization (>48 hours), and major adverse cardiac events. Models were adjusted for frailty (Clinical Frailty Scale {greater than or equal to}5) and major surgery, using multivariable, ordinal, and Firth penalized logistic regression analyses. Results: Patients were predominantly male (86%) with a mean age of 76 {plus minus} 9 years, and 61% were frail. Higher RCRI scores were associated with unplanned postoperative hospital admission (RCRI {greater than or equal to}3: adjusted OR 48.9, 95% CI 4.8-502.2) and longer LOS (RCRI {greater than or equal to}3: adjusted OR 40.7, 95% CI 4.3-382.8). RCRI {greater than or equal to}3 was also associated with prolonged hospitalization (>48 hours) in Firth penalized logistic regression, whereas frailty was not independently associated. Conclusions: In a real-world ATTR-CM cohort undergoing major non-cardiac surgery, the overall risk of adverse outcomes was low, and higher RCRI scores were associated with increased postoperative hospital admission and longer LOS, including hospitalization exceeding 48 hours. The RCRI retains prognostic utility in this high-risk cohort and may support peri-operative risk stratification.

Background: Heart failure with preserved ejection fraction (HFpEF) is a heterogeneous syndrome comprising multiple pathophysiological phenotypes. HFpEF trials have largely enrolled diverse populations and reported average treatment effects, consistently yielding neutral results that may obscure drug-specific benefits within distinct subgroups. To address this issue, we employ an interaction-based that incorporates treatment-by-variable interactions to uncover drug-specific responses. Methods: We leveraged four HFpEF clinical trials (TOPCAT, RELAX, NEAT-HFpEF, INDIE-HFpEF) and developed a framework comprising two complementary approaches. The first employed a prognostic responder model to evaluate whether conventional responder definitions reflect treatment-specific benefit or instead capture favorable clinical trajectories common to both treatment and placebo groups. The second used an interaction-based individual treatment effect (ITE) modeling to identify baseline variables that modify therapy effect, distinguishing drug-specific response from prognostic phenotypes. Results: Although the prognostic responder model demonstrated good discrimination, further analisys suggested it primarily captured a prognostic signal associated with favorable clinical trajectories common to both treatment and placebo arms. In contrast, the ITE model identified distinct, drug-specific effect modifiers across trials (cardiorenal-inflammatory for spironolactone (TOPCAT), NO-mediated anti-inflammatory for isosorbide mononitrate (NEAT-HFpEF), afterload-reducing for inorganic nitrite (INDIE-HFpEF), and anti-volume-overload for sildenafil (RELAX). Each ITE model demonstrated significance only within its own trial suggesting drug-specific signal. Conclusions: The proposed method identifies mechanism-specific effect modifiers, and uncovers clinically meaningful heterogeneity in treatment response, which is not captured by conventional MCID-based approaches. Although exploratory, these findings support phenotype-guided therapy in HFpEF and argue for phenotype-informed trial design to enhance treatment-effect detection and therapy targeting.

Background: Mortality analysis traditionally focuses on the single underlying cause of death (UCD), which obscures the wider morbidity process at the end of life. Multiple causes of death (MCoD) data, recording all conditions on the death certificate, are increasingly used as a proxy for end-of-life multimorbidity, yet how accurately they represent it remains underinvestigated. We assessed whether recorded causes reflect end-of-life health conditions or rather the chain of events leading to death. Methods: Using linked Danish registers (Population, Cause of Death, Chronic Diseases, and Cancer), we studied residents aged 50+ diagnosed with COPD, dementia, diabetes, or cancer who died in 2010-2022 (ranging from 38779 to 224330 per disease cohort). We examined how often each diagnosed disease appeared on the certificate, its location and selection as the UCD, factors associated with its appearance (logistic regression), disease-specific mortality (multiple decrement life tables), and disease associations (Cause of Death Association Indicator, CDAI). Results: Cancers appeared on the death certificate far more often than chronic diseases (around 75% versus 19-58%) and were usually recorded in Part 1 and selected as the UCD, whereas chronic diseases were rarely the UCD. The odds of a disease appearing depended on factors such as age at and time since diagnosis. When a diagnosed disease was recorded, the certificate traced a coherent path to death; when it was absent, ill-defined causes became more common. The CDAI highlighted specific association pathways between diseases. Conclusions: MCoD data capture only part of the chronic disease burden present at death and should be interpreted cautiously as a proxy for end-of-life multimorbidity. They are, however, well suited to describing the pathways leading to death.

Background. Diagnosis and risk stratification in rare hematologic malignancies such as myeloproliferative neoplasms (MPNs) - polycythemia vera (PV), essential thrombocythemia (ET), and myelofibrosis (MF) - require expert review of longitudinal, heterogeneous clinical records. This process is cognitively demanding, inconsistently applied, and difficult to scale beyond tertiary centers. No automated phenotyping workflow currently exists for hematologic malignancies. Methods. A HIPAA-compliant large language model (LLM) framework for phenotyping MPN was developed to integrate (i) rule-based retrieval of bone marrow biopsy reports, clinical notes, and structured laboratory results from the electronic health record (EHR); (ii) zero-shot extraction of diagnostic and prognostic variables from unstructured text using GPT-4 Turbo; (iii) a clinician-informed source-prioritization algorithm to reconcile conflicting multi-source data; (iv) WHO/ICC-criteria-based diagnostic classification; and (v) NCCN-based risk stratification using the conventional risk model for PV, IPSET-thrombosis for ET, and DIPSS, DIPSS-plus, and MIPSS70/MIPSS70+ v2 for MF. Patients were identified via MPN-related ICD-9/10 codes; cases met 2017 WHO criteria or had a hematologist-documented diagnosis, and controls did not. The cohort was split into a prompt-development set (n = 60) and a held-out test set (n = 450; 75 cases and 75 controls per disease). Ground truth was established by independent dual-clinician chart review with consensus adjudication. LLM performance was evaluated against the ground truth: variable-level extraction using accuracy, F1 score, and Cohen's kappa; patient-level diagnostic classification using sensitivity, specificity, and Cohen's kappa; and prognostic risk stratification (among confirmed cases) using accuracy, weighted F1 score, and quadratic-weighted Cohen's kappa. Wilson 95% confidence intervals (CIs) were used for proportions and bootstrap 95% CIs with 500 resamples for F1 scores. Results. The held-out test set included 450 patients (PV: 150; ET: 150; MF: 150) with pathology reports and structured laboratory results, and 172 patients (PV: 52; ET: 55; MF: 65) with clinical notes. From pathology reports, overall variable extraction accuracy and F1 score were 99% (95% CI, 98-100) and 1.00 (0.99-1.00) for PV, 100% (99-100) and 0.99 (0.96-1.00) for ET, and 100% (99-100) and 0.99 (0.97-1.00) for MF. From clinical notes, overall accuracy and F1 score were 96% (91-100) and 0.94 (0.85-1.00) for PV, 100% (100-100) and 1.00 (1.00-1.00) for ET, and 100% (99-100) and 0.98 (0.95-1.00) for MF. Diagnostic sensitivity was 100% (95% CI, 95.1-100.0) for PV, ET, and MF; specificity was 98.7% (92.8-99.8) for PV and 100% (95.1-100.0) for both ET and MF, with Cohen's kappa of 0.99 for PV and 1.00 for ET and MF. Risk stratification accuracy was 100% with weighted F1 score of 1.00 and quadratic-weighted Cohen's kappa of 1.00 across all three diseases. A pre-specified source-ablation analysis showed that pathology reports alone were sufficient for diagnosis (sensitivity 98.7% for PV, 100% for ET, 96.0% for MF; specificity 100% across all three subtypes) but inadequate for prognostication (accuracy 69.3% for PV, 93.3% for ET, 77.3% for MF). Adding clinical notes to pathology reports recovered full prognostic accuracy of 100% across all three diseases. Conclusions. This first-in-class automated framework achieved expert-level performance for MPN diagnosis and risk stratification from real-world EHR data, establishing a foundation for scalable, standardized phenotyping in rare hematologic malignancies. Prospective, multi-site validation is warranted before clinical deployment.

Background. Clostridioides difficile infection (CDI) imposes a burden that extends well beyond the gastrointestinal tract, yet existing outcome measures only partially capture the patient experience. We used frontier large language models (LLMs) on patient and caregiver narratives at scale to describe how burden shifts with disease course. Methods. We analyzed 189 testimonials from the Peggy Lillis Foundation corpus, sorted into four cohorts with recurrence (r) and fulminant (f) severity as axes (rfCDI, fCDI, rCDI, non-rfCDI). Two independent LLMs coded eight thematic domains, four fulminant flags, thirteen emerging semantic fields, the dominant dimension, and narrative arcs. Two clinicians independently coded a subset for inter-rater reliability (PABAK, Gwet's AC1). Results. Treatment trajectory was the dominant theme in recurrent disease, whereas death and near-death dominated non-recurrent fulminant narratives. Psychological burden was near-universal in fulminant disease (98.0% in rfCDI, 97.2% in fCDI). Caregiver and bereavement content concentrated in fCDI (66.7%). Diagnostic failure was frequent across recurrent cohorts (47.6 - 56.1%). Bacteriotherapy tracked recurrence (60.2% rfCDI versus 5.6% fCDI). Financial, mental-health, and caregiver burdens were prominent and are currently unaddressed by guidelines. Human-human reliability was substantial (PABAK 0.79 for semantic fields, 0.76 for domains); arc coding was least reliable. Conclusions. Patient narratives reveal a course-dependent, multidimensional burden in CDI. Concrete gaps exist between what patients prioritize, what guidelines recommend, and what therapy access provides. Frontier-LLM coding, validated against clinicians, offers a reproducible route to translate these priorities into research, care, and policy.

Background: Black individuals have a lower incidence of atrial fibrillation (AF) than White individuals despite a higher burden of many traditional cardiovascular risk factors. Differences in left atrial (LA) structure and function by race could partly explain the observed pattern of AF risk. Methods: This analysis included 4,576 (978 Black and 3,598 White) participants from the Atherosclerosis Risk in Communities (ARIC) study, followed between 2011 and 2021. The association of selected echocardiographic measures of LA structure and function with AF incidence was evaluated with race-specific Cox proportional hazards models with adjustment for sociodemographic and clinical covariates. Additional analyses assessed whether LA measures attenuated the association between race and incident AF. Results: The analysis included 778 AF cases (113 in Black and 665 in White participants, mean age 75 years). Larger LA size and worse LA function were associated with higher AF risk in both Black and White individuals, with most associations of similar magnitude in both groups, except for a slightly stronger association of LA reservoir strain in Black than White participants (Black: hazard ratio (HR) 0.89, 95% CI 0.86-0.92 per 1% increase; White: HR 0.94, 95% CI 0.92-0.95, p for interaction = 0.01). In the overall sample, White participants showed higher AF risk compared to Black participants (HR 1.59, 95% CI 1.24-2.03). Adjustment for most individual LA measures did not attenuate the association between race and AF risk. Conclusion: Larger LA size and worse LA function were associated with incident AF in both Black and White ARIC participants. However, these measures did not explain the lower AF incidence observed among Black participants. LA remodeling appears to be an important predictor of AF risk, but it is not the primary explanation for the Black-White AF paradox.

Background: Within the VEBIS-EHR project, monthly vaccine effectiveness (VE) of COVID-19 vaccines is routinely estimated across EU/EEA countries. While VE quantifies direct protection, it does not capture the overall population benefit of vaccination campaigns in terms of severe outcomes prevented. Aim: To estimate the impact of the 2023 and 2024 autumn COVID-19 vaccination campaigns in adults aged [&ge;]65 years. Methods: We conducted a retrospective cohort study using electronic health records data from Belgium, Denmark, Italy, Navarre (Spain), Portugal, Norway and Sweden. Weekly numbers of averted COVID-19-related hospitalisations and deaths during the 12 months following each campaign were estimated using observed COVID-19-related events, vaccine coverage (VC) and interpolated weekly VE. Results: Across participating countries/regions, among adults aged [&ge;]65 years, the 2023 autumn vaccination campaign averted approximately 6,200 hospitalisations (prevented fraction [PF] 10%) compared with 2,200 (PF 12%) in 2024. Among those aged [&ge;]80 years, the number of averted COVID-19-related deaths was 811 (PF 13%) for the 2023 campaign and 156 (PF 12%) for the 2024 campaign. Impact varied across countries, reflecting differences in VC, vaccination timing and outcome occurrence. Conclusion: The 2023 and 2024 autumn vaccination campaigns resulted in substantially different numbers of averted COVID-19-related hospitalisations and deaths among older adults, with fewer events averted in 2024. These findings highlight that the impact of vaccination programmes depends not only on VC and VE but also on alignment between vaccination timing and periods of increased viral circulation.

Molecular testing in hematology requires different assays for disease subgroup identification, risk stratification and selection of appropriate treatment regimens. Yet, molecular tests are not necessarily standardized between diagnostic laboratories, resulting in varying turnaround times and potentially divergent results. To resolve this issue and enable single-assay molecular testing, we have developed a hierarchical classification framework that combines epigenetic and genetic data from whole genome nanopore sequencing (WGNS) with machine learning to determine disease entities, epigenetic subgroups (epitypes) and genetic aberrations in hematopoietic neoplasms. We curated DNA methylation data from 5,420 samples and trained a classifier allowing entity-level diagnostics featuring 21 conditions, including healthy controls, acute and chronic myeloid and lymphoid neoplasms. This classifier was subsequently combined with entity-specific epitype classifiers predicting 44 therapeutically or prognostically relevant states, followed by integration of genetic data. Benchmarking of the combined (epi-)genetic testing strategy using WGNS confirmed high accuracy in the detection of diagnostic groups and risk stratification, and identified diagnosis-defining molecular alterations that were not reported by standard-of-care work-up.

Introduction: Multimorbidity clusters have been associated with increased dementia risk. While lifestyle factors may modify dementia risk, their role in multimorbidity clusters remains unclear. Method: Data from UK Biobank was used to identify clusters of chronic conditions using latent class analysis, assess their associations with dementia risk using Cox regression, and potential moderating effects of lifestyle factors. Results: We included 465,175 participants (mean age (SD) = 56.52 (8.01), 53.87 % female). Five clusters were identified and significantly associated with increased dementia risk, with the cardiometabolic (HR = 2.14, p < 0.001) and mental health cluster (HR =1.99, p < 0.001) exhibiting the highest risk. Only moderate physical activity lowered dementia risk in the pain-dominated multimorbidity cluster (HR = 0.77, p = 0.039). Discussion: Lifestyle factors including physical activity may protect against dementia in specific multimorbidity clusters. Future research involving objective and multiple lifestyle measures is needed.

Abdominal aortic aneurysm (AAA) patients in the ICU represent a heterogeneous, high-risk population with mortality risk evolving across distinct clinical phases. Existing prognostic tools rely largely on Cox proportional hazards (Cox PH) nomograms with narrow predictor sets and single time horizons, leaving the value of modern machine learning, extended features, and external generalizability uncharacterized. We extracted an ICD-coded AAA ICU cohort from MIMIC-IV v2.2 (858 patients with complete six-predictor admission data: age, BUN, sepsis, antihypertensive use, anion gap, mean SpO2) using a 24-hour admission window. An extended feature set added hemodynamic, laboratory, and comorbidity variables, with feature selection via LASSO and SVM-RFE intersection. Six models (Cox PH, logistic regression, random forest, gradient boosting, XGBoost, MLP) were trained on a 70% split and evaluated at 7-, 14-, and 28-day horizons using ROC-AUC, C-index, Brier score, calibration, and SHAP. External validation used a harmonized eICU-CRD cohort. In-hospital mortality was [~]11.8%. On the six-predictor set, logistic regression led at 7 days (AUC 0.866) and 14 days (AUC 0.872), with XGBoost competitive. Extended features yielded modest gains; random forest achieved the best 28-day AUC (0.892). The MLP consistently underperformed. Discrimination declined monotonically with longer horizons. External validation showed expected attenuation (best 7-day AUC 0.771). SHAP consistently identified anion gap, BUN, and age as top contributors. We conclude that regularized linear models excel under data scarcity, while tree ensembles gain advantage as features and horizons expand. External results motivate local recalibration before deployment.

Background: Cardiovascular disease (CVD) remains the leading cause of mortality among older U.S. adults, yet the contribution of neighborhood-based structural racism remains inadequately quantified. This study quantifies the association between the Structural Racism Effect Index (SREI) and CVD mortality among adults aged {greater than or equal to}65 years, evaluating how this relationship varies across U.S. geographic regions to identify key areas for intervention. Methods: This ecological study applied a hierarchical Bayesian spatiotemporal framework to 2017-2020 Centers for Disease Control and Prevention (CDC) Wide-Ranging Online Data for Epidemiologic Research (WONDER) data to estimate the association between SREI and CVD mortality across 3,007 U.S. counties. SREI was modeled continuously and categorically, adjusting for sociodemographic covariates. Population attributable fractions (PAF) and attributable deaths (AD) quantified the potentially preventable burden and its spatial disparities. Results: From 2017 to 2020, approximately 2.79 million CVD deaths were observed, with significant spatial clustering (Moran's I = 0.35, p < 0.001). Each standard-deviation increase in SREI was associated with 13% higher CVD mortality (IRR: 1.13, 95% CrI: 1.12-1.15). A positive dose-response gradient was observed across SREI quartiles, with mortality 24% higher in the highest quartile than in the lowest (IRR: 1.24, 95% CrI: 1.20-1.28). The PAF was 6.94% (95% CrI: 6.13-7.73), corresponding to 193,472 potentially preventable deaths. High exceedance probabilities (>0.95) were concentrated in the Southeast, Appalachia, and the Midwest. Conclusions: Structural racism is a spatially patterned, dose-dependent predictor of older adult CVD mortality, underscoring the need for public health monitoring and neighborhood-based upstream interventions where disease burden is concentrated. Keywords: Structural Racism Effect Index; Neighborhood disadvantage; Cardiovascular Disease Mortality; Bayesian Spatiotemporal Analysis; Population Attributable Fraction; Health Disparities; Health Equity.

There are no published data on cancer screening by women using medically assisted reproduction (MAR). Such data would aid interpretation of the cancer incidence and risk profiles for this group. Using linked population-based Australian health registries and administrative datasets, we compared organised publicly funded cervical and breast screening episodes for women who received one of three types of MAR and matched women who did not between 1991 and 2016. We modelled the proportion of women screened in the three years before and after first MAR treatment, adjusting for age, remoteness, parity, socio-economic disadvantage, cancer history, and uptake of the other screening program. After adjustment, a greater proportion of women who received MAR than women who did not had cervical screening before MAR (77.3%-84.1% vs 57.5%-62.0%, depending on treatment) and after MAR (77.0%-78.5% vs 68.1%-68.3%). Contrastingly, breast screening estimates were 7.6%-9.6% vs 9.3%-10.5% before MAR and 11.0%-15.0% vs 12.8%-14.9% after MAR.

Glucagon-like peptide-1 receptor agonists (GLP-1RAs) have been linked to heterogeneous, potentially pleiotropic effects across organ systems, motivating outcome-wide comparative risk profiling in real-world data. A central challenge in such analyses is \emph{residual bias} that remains after adjustment for observed confounders, which can distort effect estimates and mis-calibrate uncertainty. We present distributional diagnosis and calibration (DC), which uses panels of negative control outcomes (NCOs) to diagnose residual bias and calibrate uncertainty. DC evaluates null behavior via $p$-value uniformity and empirical coverage across NCOs, and uses the empirical distribution of NCO effect estimates to calibrate confidence intervals for prespecified primary outcomes. DC is modular: it can wrap around commonly used causal inference methods and operates directly on summary statistics, supporting collaborative research under data-sharing constraints. Using electronic health records from a large U.S. clinical research network (152.7 million patients), we compared GLP-1RAs with sodium--glucose cotransporter~2 inhibitors across 15 prespecified outcomes spanning cardiovascular, mental health, and genitourinary domains using four causal estimators. Across outcomes and methods, DC diagnostics revealed substantial and method-dependent residual systematic error. DC calibration attenuated systematic error signals observed in negative controls and yielded more stable, better-calibrated estimates for clinical outcomes, supporting DC as a practical strategy to strengthen the credibility of real-world comparative effectiveness research.

Statistical prediction models for binary outcomes are becoming increasingly popular. One significant challenge is calibrating these models to suit the characteristics of a target population that is structurally different from the original population. Calibration is especially challenging when there is no training data available from the target population. To address this problem, we propose a novel calibration method, SimCal, which uses synthetic data generated from the model development data in conjunction with marginal statistics from the calibration cohort. We show that expert judgment modeling (EJM) may be used for calibration if cross-sectional data from the target population are available comprising expert judgments about the potential outcome and the covariates. We describe three alternative calibration approaches when calibration data are lacking: similarity-binning averaging (SBA), adaptive calibration of predictions (ACP), and Elkan calibration. In a simulation study, we compare SBA, ACP, Elkan calibration, and SimCal. R code for applying these methods is provided from the re-analysis of data on coronary artery disease. We illustrate all 5 calibration approaches with a real data set for predicting functional outcome after stroke and all approaches but EJM in the re-analysis of the Cleveland Clinic data. None of the approaches performed convincingly well in all situations. SimCal performed well when model parameters were correctly specified. EJM failed on the stroke data. Further research is urgently required for calibration in the absence of calibration data.

Background: Transition timing to post-acute rehabilitation after ischemic stroke is heavily influenced by non-clinical factors, introducing potential systemic disparities in care access. We evaluated the association between insurance payor status and acute hospital length of stay (LOS) prior to inpatient rehabilitation discharge among critically ill stroke patients. Methods: Using the MIMIC-IV database, we identified ICU-admitted adults with ischemic stroke discharged to inpatient rehabilitation (n=1,285). The primary outcome was hospital LOS prior to rehab transfer. Multivariable log-transformed linear regression evaluated the association with insurance payor (Medicare, private, other/unknown; reference: Medicaid), adjusting for demographics, diagnostic-code counts (medical complexity), and ICU LOS (acute illness severity). Results: Median hospital LOS before rehab discharge was longest for Medicaid patients (13.2 days) compared with private insurance (11.0 days) and Medicare (9.5 days). In the adjusted model, Medicare insurance was associated with a significantly shorter transition time to inpatient rehabilitation, corresponding to a 13.5% shorter acute hospital stay (adjusted LOS ratio 0.87; 95% CI: 0.79-0.96; p=0.005) relative to Medicaid. Private insurance demonstrated a descriptive trend toward shorter LOS that did not achieve statistical significance (adjusted LOS ratio 0.93; 95% CI: 0.84-1.02; p=0.122). Other and unknown payor categories showed no significant differences. Conclusions: Insurance payor status serves as an independent predictor of acute care transition timing for stroke patients requiring inpatient rehabilitation. The prolonged acute stays observed among Medicaid beneficiaries suggest significant non-clinical, administrative bottlenecks in post-acute placement, underscoring the critical need for standardized, streamlined insurance approval pathways to ensure equitable neurological recovery.

Summary Background: Spatial variation in heat-related mortality remains poorly understood, particularly at fine geographical scales. We conducted a nationwide small-area study to examine the association between spatial variation in heat-related mortality and environmental, demographic, health, and socio-economic factors. Methods: We obtained daily all-cause mortality data for people aged [&ge;] 65 years during the summers of 2011-2023 and linked them with municipality-level daily temperature estimates from the ERA5-Land reanalysis dataset. We applied a two-stage Bayesian hierarchical model to estimate small-area heat-related mortality and assess the contribution of community characteristics to spatial variability. Findings: Heat-related mortality showed marked geographical differences, with the highest rates in southern and southeastern Italy. Across municipalities, the relative risk at the 90th temperature percentile, relative to the minimum mortality temperature, ranged from 1.06 to 1.33. The heat-attributable fraction exceeded 6% in several southern municipalities, while excess mortality surpassed 8 deaths per 1,000 inhabitants in parts of the Po Valley, Tuscany, Apulia, and Sicily. National heat-attributable mortality peaked in 2022, with an estimated 17,828 deaths (95% credible intervals: 17,339, 18,285) among older adults. Municipalities with higher average temperatures, less green space, higher obesity prevalence, and more residents aged [&ge;] 85 years had higher heat-related mortality. Educational attainment and employment were among the strongest modifiers of spatial variation. Interpretation: Our findings highlight substantial small-area differences in heat-related mortality across Italy and identify socio-economic deprivation as a key determinant of vulnerability. Heat is likely to disproportionately affect disadvantaged communities, reinforcing the need for adaptation strategies addressing social inequality. Funding: Imperial College Research Fellowship; Italian Ministry of Health PNC (CUP J55I22004450001); NIHR Imperial Biomedical Research Center (BRC NIHR203323).

Universities provide a key opportunity to deliver social prescribing, a care pathway that aims to connect people with non-medical forms of support within the community to address their social, emotional, and practical needs. However, it is unclear whether students in the UK are aware of social prescribing and whether it would be an acceptable form of support. We surveyed 775 university students across the UK who completed a questionnaire measuring awareness and perceptions of social prescribing. We described awareness and attitudes and used logistic regression to explore how they differed according to individual characteristics. We found an awareness-attitude paradox. Only 25% of students were aware of social prescribing, but attitudes were overwhelmingly positive once explained: 97% thought it could support mental health and wellbeing; 95% believed universities should offer it; and 89% would accept social prescribing if offered by a healthcare professional. Students who were older, postgraduates, and had English as their first language were among those with higher odds of being aware of social prescribing, but positive attitudes were more evenly reported across the sample. Our findings indicate that implementation efforts should prioritise awareness-raising and clear referral pathways, rather than increasing students' willingness to engage with social prescribing.

Abstract Introduction. Culture data guide orthopedic-infection management, yet the organism spectrum, resistance, and no-growth fraction are reported inconsistently and mostly within proprietary registries. We characterized these in a public, reproducible dataset. Methods. Retrospective cross-sectional study using MIMIC-IV version 3.1, a de-identified single-center US database. Episodes with an International Classification of Diseases diagnosis of prosthetic joint infection (PJI) or native osteomyelitis were identified; organism-spectrum and no-growth analyses were restricted to the 46% with at least one deep musculoskeletal culture (tissue or bone, synovial or joint fluid, implant sonication), so the benchmark describes culture-sampled, not all, coded episodes. Proportions carry exact 95% CIs; variation was tested by logistic regression with Benjamini-Hochberg control, and an out-of-fold logistic model quantified how well no-growth was anticipated by structured data. Results. Of 7697 episodes (median age, 60 years; 35.5% female), 1089 were PJI, 5715 native osteomyelitis, and 893 other device infection. Among 7700 deep specimens (3560 episodes; 2603 patients), 35.7% showed no growth (patient-clustered 95% CI, 34.0%-37.3%). The fraction was higher in PJI than osteomyelitis (48.6% vs 26.6%) but rose with sampling intensity (24.5% to 50.7%), indicating differential ascertainment. S. aureus led (32.5%; 43.3% methicillin-resistant), and PJI was less often polymicrobial than osteomyelitis (adjusted OR, 0.44). No-growth was weakly anticipated by structured data (out-of-fold AUROC, 0.63). Conclusions. About one-third of deep specimens from code-defined orthopedic infection showed no growth. This specimen-level fraction differs from a criterion-confirmed culture-negative-infection rate and depends on sampling intensity; it is released as a re-runnable benchmark on identical open data, not a transferable rate.