HCA Healthcare

Importance:Pediatric sepsis causes substantial morbidity and mortality, but population surveillance relies on administrative codes with limited and variable accuracy. Objective:To estimate US national incidence, mortality, and trends of sepsis in nonneonatal children using a Pediatric Sepsis Event (PSE) definition adapted from the 2024 Phoenix criteria for scalable electronic health record (EHR)-based surveillance using routinely captured clinical data. Design, Setting, and Participants:Retrospective cohort study of 3.9 million hospitalizations (age, >30 days to 17 years) in 2 EHR datasets: Epic Cosmos (245 health care systems, 2016-2023) and HCA Healthcare (146 hospitals, 2018-2023). Secondary datasets were analyzed to assess feasibility of implementation and face validity across heterogeneous settings. The PSE was validated through medical record reviews of 581 high-risk encounters at 3 geographically diverse hospitals. Exposures:A PSE required presumed infection with concurrent organ dysfunction using Phoenix-derived thresholds adapted for routine EHR data. Septic shock was defined as a PSE with cardiovascular dysfunction. Main Outcomes and Measures:Sepsis incidence, characteristics, and in-hospital mortality were calculated. Sensitivity and specificity of PSE for physician-adjudicated Phoenix sepsis were compared with administrative codes for severe sepsis/septic shock. National sepsis case counts and deaths in 2022 and temporal trends from 2016 to 2022 were estimated using regression models. Results:Among 3 925 809 pediatric hospitalizations from 2016 to 2023, 51 542 sepsis cases (mean age, 6.6 [SD, 6.0] years; 22 840 [44.3%] female) were identified (1.3% incidence); 37 405 (72.6%) were community onset and 31 744 (61.6%) had septic shock. In-hospital mortality was 10.1% and sepsis was present in 17.8% of hospitalizations that culminated in death. Incidence, characteristics, and mortality were broadly consistent across secondary datasets. On medical record review, the PSE definition had 69.9% sensitivity (95% CI, 58.1%-79.8%) and 93.1% specificity (95% CI, 89.6%-95.7%), with higher sensitivity than and comparable specificity with administrative codes. National estimates for 2022 were 18 231 sepsis cases (95% CI, 16 129-20 334) and 1877 deaths(95% CI, 1629-2126). Neither sepsis cases nor deaths changed significantly from 2016 to 2022 (annual change, 0.2% [95% CI, -2.2% to 2.7%] and 0.3% [95% CI, -3.1% to 3.8%], respectively). Conclusions and Relevance:An EHR-based definition for pediatric sepsis demonstrated strong validity compared with physician-adjudicated Phoenix sepsis and identified sepsis in 1.3% of pediatric hospitalizations with 10% mortality, corresponding to more than 18 000 cases and more than 1800 deaths annually in the US.

BACKGROUND:Patients with extensive ischaemic change are often excluded from endovascular thrombectomy. We aimed to synthesise the evidence from recent trials in these patients by performing a systematic review and individual patient data meta-analysis to estimate treatment benefit, including within clinical and imaging subgroups. METHODS:In this systematic review and meta-analysis, we searched PubMed and Embase for randomised trials published between March 1, 2018, and March 1, 2025, that evaluated efficacy and safety of endovascular thrombectomy compared with medical management in patients with large-core ischaemic stroke (based on an Alberta Stroke Program Early CT Score [ASPECTS] of ≤5 or estimated ischaemic core ≥50 mL) presenting within 24 h of onset. Individual patient-level data from all eligible trials were obtained. A central imaging core laboratory readjudicated ASPECTS and reanalysed ischaemic core volume. A two-stage meta-analysis with random-effects model was used to evaluate the distribution of 90-day modified Rankin Scale (mRS) scores (the primary outcome) using adjusted pooled generalised odds ratios (aGenORs). Missing data were handled by multiple imputation. Safety outcomes were all-cause mortality within 90-day follow-up and neurological worsening within 24-48 h of randomisation, reported as adjusted pooled relative risk (aRR); and symptomatic intracerebral haemorrhage within 36 h of randomisation (reported as risk difference). Subgroup analyses based on clinical and imaging characteristics were done, including subgroups defined by ischaemic core volume, ASPECTS, and time window from onset to randomisation. The meta-analysis was registered with PROSPERO (CRD420251058584). FINDINGS:We included 1886 patients (944 assigned to endovascular thrombectomy and 942 assigned to medical management) from six trials. Baseline characteristics were similar between treatment groups. At day 90, the distribution of mRS scores was improved in patients in the endovascular thrombectomy group (median score 4 [IQR 3-6]; n=940) versus those in the medical management group (5 [4-6]; n=931; aGenOR 1·63 [95% CI 1·42-1·88], p<0·0001). The endovascular thrombectomy group also had reduced mortality (292 [31·1%]) compared with the medical management group (347 [37·3%]; aRR 0·82 [95% CI 0·70-0·97], p=0·022). No significant differences were observed in symptomatic intracranial haemorrhage (ten [1·1%] of 944 vs nine [1·0%] of 942 patients; pooled unadjusted risk difference -0·17 percentage points [95% CI -1·01 to 0·67], p=0·69) or neurological worsening (197 [22·0%] of 896 patients vs 161 [17·9%] of 899; aRR 1·19 [0·87-1·62], p=0·27). Improved functional outcomes with endovascular thrombectomy were consistent across clinical and imaging subgroups, except for those with an estimated ischaemic core volume of 150 mL or greater, in whom point estimates favoured endovascular thrombectomy, particularly in the early time window (0-6 h), but wide 95% CIs limited interpretation. INTERPRETATION:Endovascular thrombectomy was associated with improved functional outcomes and reduced mortality versus medical management in patients with large-core ischaemic stroke presenting within 24 h of onset. With the exception of very extensive ischaemic changes (core volume ≥150 mL) presenting beyond 6 h, where evidence remains limited, benefit was sustained across ASPECTS and ischaemic core strata for patients presenting up to 24 h after onset. FUNDING:None.

RATIONALE:Efficient distribution of scarce critical care resources is essential to save the most lives in times of crisis. Evidence-based practices and processes enhance clinical decision-making. OBJECTIVES:The objective of these guidelines was to develop evidence-based, rather than expert-based, recommendations for triaging critically ill patients eligible for ICU admission during times of crisis-level shortages in ICU capacity. DESIGN:The American College of Critical Care Medicine Board convened a 21-member multidisciplinary panel, comprising doctors in medicine, nursing, and law; advanced practice providers; respiratory therapists; ethicists; and patient/family representatives. The panel included two expert methodologists specialized in developing evidence-based recommendations in alignment with the Grading of Recommendations, Assessment, Development, and Evaluation (GRADE) methodology. Conflict-of-interest policies were strictly followed during all phases of guidelines development including task force selection and voting. METHODS:The panel members identified and formulated five fundamental Patient, Intervention, Comparator, and Outcomes questions. The panel conducted a systematic review for each question to identify the best available evidence, analyzed the evidence, and assessed the certainty of the evidence using the GRADE methodology. The GRADE evidence-to-decision framework was used to formulate the recommendations. Good practice statements were included to provide additional guidance. RESULTS:The panel generated one conditional recommendation and five no recommendation statements. CONCLUSIONS:Crisis-level shortages significantly disrupt patient care. Despite the role of triage in minimizing adverse outcomes, there is a lack of evidence, as opposed to expert opinion, to guide practice recommendations in the critical clinical scenarios considered by the panel.

The lack of clinical data for chronic kidney disease (CKD) prediction frequently results in model overfitting and inadequate generalization to novel samples. This research mitigates this constraint by utilizing a Conditional Tabular Generative Adversarial Network (CTGAN) to enhance a constrained CKD dataset sourced from the University of California, Irvine (UCI) Machine Learning Repository. The CTGAN model was trained to produce realistic synthetic samples that preserve the statistical and feature distributions of the original dataset. Multiple machine learning models, such as AdaBoost, Random Forest, Gradient Boosting, and K-Nearest Neighbors (KNN), were assessed on both the original and enhanced datasets with incrementally increasing degrees of synthetic data dilution. AdaBoost attained 100% accuracy on the original dataset, signifying considerable overfitting; however, the model exhibited enhanced generalization and stability with the CTGAN-augmented data. The occurrence of 100% test accuracy in several models should not be interpreted as realistic clinical performance. Instead, it reflects the limited size, clean structure, and highly separable feature distributions of the UCI CKD dataset. Similar behavior has been reported in multiple previous studies using this dataset. Such perfect accuracy is a strong indication of overfitting and limited generalizability, rather than feature or label leakage. This observation directly motivates the need for controlled data augmentation to introduce variability and improve model robustness. The dataset with the greatest dilution, comprising 2000 synthetic cases, attained a test accuracy of 95.27% utilizing a stochastic gradient boosting approach. Ensemble learning techniques, particularly gradient boosting and random forest, regularly surpassed conventional models like KNN in terms of predicted accuracy and resilience. The results demonstrate that CTGAN-based data augmentation introduces critical variability, diminishes model bias, and serves as an effective regularization technique. This method provides a viable alternative for reducing overfitting and improving predictive modeling accuracy in data-deficient medical fields, such as chronic kidney disease diagnosis.

OBJECTIVES:This study aims to investigate the rates and statistical significance of maternal and neonatal complications in subjects with a positive COVID-19 diagnosis in pregnancy in comparison to subjects without a diagnosis of COVID-19 in pregnancy. We aim to improve the literature and patient information regarding the impact of COVID-19 on maternal and fetal outcomes to help draw conclusions or guide management. STUDY DESIGN:Clinical outcomes were identified using International Classification of Diseases, Tenth Revision (ICD-10) billing codes. The control group included a sample of 15,000 patients delivered between 4/1/2019-12/31/2019. The COVID group included 10,608 patients from 4/1/2020-4/1/2022 who were confirmed COVID-19 positive within the 9 months prior to delivery. Binary logistic regression, Chi-square, and Fisher's exact test were used for statistical analysis. RESULTS:Having COVID-19 during pregnancy is significantly associated with increased risk for preterm delivery (PTD), placental abnormalities, hypertensive disorders, and neonatal intensive care (NICU) admission. Rates of maternal mortality, fetal growth restriction (FGR) and intrauterine growth restriction (IUGR) were not significantly impacted by COVID-19. While the overall rate of FGR was not impacted, patients with 2nd trimester infection are at increased risk for FGR compared to patients with 3rd trimester infection. CONCLUSIONS:This study demonstrates a statistically significant increased rate of preterm delivery, hypertensive disorders, placental abnormalities, and NICU admission for pregnancies affected by COVID-19. These findings can help guide recommendations for increased surveillance and counseling in pregnancies affected by COVID-19.