Sumeet S. Chugh

Background: Contemporary data on long-term, cause-specific mortality among stroke survivors are sparse, and much of the existing evidence derives from studies performed before the modern reperfusion era. We sought to characterize long-term cause-specific mortality among stroke survivors to discharge (differentiated by stroke subtype) when compared to in-hospital stroke decedents treated under contemporary standards of care. Methods: Using a hybrid billing-code/imaging and large language model-assisted chart-review pipeline, we identified adults hospitalized for ischemic or hemorrhagic stroke in the Observational Study of Cardiac Arrest Risk (OSCAR) cohort (Cedars-Sinai Health System, Los Angeles) between 2017 and 2020, followed through June 30, 2025. Cause of death was determined and classified into 10 categories. Results: Among 1,344 stroke patients (mean age 74.0 ± 14.9 years; ischemic n=1,036 [77.1%], hemorrhagic n=308 [22.9%]), 203 (15.1%) did not survive the index stroke. Of the 1,141 discharged survivors, 625 (54.8%) died over a median follow-up of 5.2 years. Cerebrovascular causes predominated during index stroke hospitalization (36.0%) but accounted for only 15.5% of post-stroke discharge deaths. Among ischemic stroke survivors, cardiac disease was the leading cause of death (31.6%), with sudden cardiac arrest alone accounting for 10.9% of deaths. Among hemorrhagic stroke survivors, cancer (38.9%) was the leading cause of death. Despite these divergent cause-of-death profiles, overall survival did not differ significantly by subtype (median 5.5 vs 5.8 years; P =0.90). Conclusions: In this contemporary, reperfusion-era cohort, causes of death shifted from cerebrovascular predominance in the acute phase to heart disease/cardiac arrest after ischemic stroke, and malignancy after hemorrhagic stroke. These findings argue for development of subtype-specific post-stroke surveillance, including structured longitudinal cardiac and sudden-death risk assessment for ischemic stroke survivors.

Background: Sex differences in sudden cardiac arrest (SCA) incidence and outcomes are well established, but risk stratification underperforms in women. It is important to delineate sex-specific patterns in SCA across clinical, electrocardiographic, and echocardiographic domains. Methods: We studied out-of-hospital SCA cases from the Oregon Sudden Unexpected Death Study (ORSUDS; Portland, 2002–2020; discovery) and Prediction of Sudden Death in Multi-Ethnic Communities (PRESTO; Ventura County, 2015–2024; external replication). Sex differences in 38 comorbidities, 19 ECG and 12 echocardiographic measures were estimated using multivariable-adjusted logistic regression. Female sex was the modeled dependent measure (male as reference). ORSUDS estimates were compared with those in the Framingham Heart Study (FHS), a community-based reference cohort, using FDR-adjusted Cochran Q tests. Results: Among 5130 ORSUDS cases, 1715 (33.4%) were women (median age, 70.4 years [IQR, 58.7–81.8]) and 3415 (66.6%) were men (65.6 [55.0–76.8]). Compared with men, women had higher BMI (per 5 kg/m², 1.11 [1.08–1.15]) and higher odds of thyroid disease (2.56 [2.17–3.13]), asthma (2.33 [1.92–2.86]), mood disorder (2.17 [1.85–2.50]), dementia (1.27 [1.02–1.56]), and left ventricular hypertrophy (LVH, 3.13 [2.44–4.00]). Women also had lower odds of coronary artery disease (0.61 [0.52–0.71]), atrial fibrillation (0.67 [0.56–0.79]), prolonged QTc (0.45 [0.35–0.57]), prolonged PR (0.53 [0.41–0.68]), premature contractions (0.77 [0.60–0.98]), sleep apnea (0.53 [0.43–0.65]), and reduced LVEF (0.40 [0.29–0.54]). Most findings were replicated in PRESTO (n=2719). In contrast, several of these differences were reversed in the community (FHS, n=3539;FDR-significant): prolonged QTc was more common in women, whereas LVH, dementia, heart failure, and higher BMI were more common in men. Conclusions: Sex differences among SCA cases were mostly reproducible in SCA cohorts, and distinct from sex-related patterns in the community. These results have implications for development of sex-specific SCA risk assessment and prevention strategies. Clinical Perspective: What is new?: This study is the first community-based analysis to comprehensively characterize pre-arrest sex-associated profiles across clinical, ECG, and echocardiographic domains, externally validate them, and compare them with a community reference population. Women with SCA were more strongly associated with non-cardiac comorbidities, LVH, preserved LVEF, and mitral valve abnormalities, whereas male cases were characterized by coronary and peripheral vascular disease, atrial fibrillation, prolonged PR and QTc intervals, premature contractions, and reduced LVEF. Several sex associations, including QTc prolongation, LVH, body mass index, heart failure, and dementia, differed in magnitude or direction from patterns in the community reference population. What are the clinical implications?: Our findings suggest candidate features for future sex specific risk factor analyses and potential components for sex-specific risk models for SCA. Clinicians may need to recognize the critical need to extend beyond dependence on just the value of the left ventricular ejection fraction for SCA risk prediction, especially in women.

Background: Identification of in-hospital cardiac arrest (IHCA) through manual chart abstraction is the gold standard, but its time-consuming and resource-intensive, which naturally limits its practicality. Diagnostic codes provide an accessible automated alternative, but prior work has shown this method of extraction has poor sensitivity and poor positive predictive value. More accurate and scalable automated methods are needed to support IHCA surveillance and research. Methods: We developed large language model (LLM)-based systems that incorporated physician-defined clinical knowledge through guideline-enriched prompting, supervised finetuning, and a self-correction procedure. The systems identified IHCA events and their locations from electronic health record notes. Performance was evaluated at the encounter level against physician-adjudicated chart review and compared with ICD-code–based identification. Results: Compared with an ICD-code–based algorithm (precision 0.68, recall 0.91, F1 score 0.78), the supervised fine-tuned LLM offered substantially better performance for IHCA identification on a curated evaluation corpus, reaching a precision of 0.88, a recall of 1.00, and an F1 score of 0.94. When tested on a larger, unselected validation cohort of 45,525 encounters, a separate LLM configuration—relying on guideline-enriched prompting and a self-correction procedure rather than fine-tuning—achieved a precision of 0.79 for encounter-level IHCA identification, meaning most encounters flagged by the model were genuine IHCA cases. The fine-tuned LLM underperformed on this cohort due to its inability to self-correct its decisions, an issue that can be mitigated by having a separate model perform the correction step. Conclusion: Explicit integration of clinical knowledge through prompting and supervised fine-tuning enabled accurate identification of IHCA events and locations from routine clinical documentation. In a two-pass workflow, these systems could prioritize encounters for clinician adjudication, reduce manual screening requirements, and support more scalable IHCA surveillance and registry development.

BACKGROUND: At least 50% of individuals who suffer sudden cardiac arrest (SCA) have warning symptoms before their SCA, but these are not sufficient to predict imminent SCA (ISCA). We hypothesized that combining symptoms with clinical profile could effectively predict ISCA. METHODS: In 2 community-based studies of SCA in Oregon and California, survivors of SCA who had experienced warning symptoms and called 911 (n=364) were compared with control subjects (n=313) who also called 911 for similar symptoms but did not have SCA. Symptom information was obtained directly from interviews as well as prehospital emergency records. We used machine learning to identify predictive combinations of symptoms and clinical features in this case-control study. The area under the receiver operating characteristic curve (AUC) and 5-fold cross-validation were used to assess model performance and stability. RESULTS: Warning symptom-clinical profile combinations were successfully identified. The combination of chest pain and known heart failure predicted ISCA among males, while chest pain and known coronary artery disease predicted ISCA among females (AUC=0.788). With dyspnea, both history of coronary artery disease and heart failure were predictors (AUC=0.745). In the absence of chest pain or dyspnea, seizure-like symptoms, coronary artery disease, and heart failure were predictors (AUC, 0.775). 5-fold cross-validation produced consistent results. CONCLUSIONS: Combinations of warning symptoms and clinical features distinguished individuals with SCA from individuals without SCA with good accuracy (AUCs, 0.745–0.788). These findings have implications for clinical triage of ISCA warning symptoms in emergency settings and warrant investigation in larger studies.

Background: Although cardiovascular events increase sudden cardiac arrest (SCA) risk, the impact of recurrent events on subsequent SCA risk in a contemporary population is unknown. This study assessed whether patients with a first‐time acute coronary syndrome (ACS) or heart failure (HF) hospitalization who experience a recurrent cardiovascular event have increased SCA risk. Methods: The OSCAR (Observational Study of Cardiac Arrest Risk) is a prospective cohort with adjudicated SCA outcomes. The current study followed up patients who survived a first ACS or HF hospitalization (the index ACS or HF cohorts) for recurrent cardiovascular events and SCA. Recurrent event was a time‐dependent variable in Cox models predicting SCA. Findings were validated in the FHS (Framingham Heart Study). Results: Among 2946 patients in the index ACS cohort, incidence of SCA was higher following a recurrent ACS event than without (3.70 versus 1.28 per 100 patient‐years). A recurrent ACS event was associated with higher risk of SCA (adjusted hazard ratio [HR], 3.15 [95% CI, 2.06–4.83]; P P P =0.0002); the association was not significant for recurrent HF (HR, 1.49 [95% CI, 0.73–3.03]; P =0.27). Conclusions: Recurrent events were associated with higher risk of SCA; dynamic clinical trajectories of recurrent cardiovascular events may inform prevention of SCA.