Emergency Medicine

Peripheral IV vasopressors: very low major complication rates with short catheters

This systematic review and meta-analysis pooled 49 studies with 33,060 peripheral vasopressor catheters in critically ill adults. Across all vasopressors, minor adverse events occurred in about 2% of catheters, varying modestly by agent. Major adverse events clustered in midline catheters, with venous thromboembolism incidence around 1.4%. Short peripheral IV catheters had essentially no reported major events, with a pooled incidence of 0.0%. Peripheral vasopressor use avoided central venous catheter placement in roughly 60% of cases overall. 1

Early warning scores in the very old: REMS calibrates best, especially ≥94 years

This retrospective study evaluated five early warning scores in 50,645 ED visits by patients aged ≥80 years. Clinical deterioration within 24 hours, defined as death or ICU admission, occurred in 2.4% of visits. All scores showed fair discrimination, with AUROCs roughly 0.75–0.78; NEWS discriminated slightly best. REMS demonstrated the best calibration overall and uniquely maintained or improved performance beyond age 90, particularly ≥94 years. In older patients, oxygen supplementation, systolic blood pressure, and Glasgow Coma Scale most strongly predicted deterioration. 2

Timely ED antibiotics and fluids increase likelihood of discharge home after sepsis

This multicenter cohort included 38,568 adults hospitalized with community-onset sepsis across 67 hospitals. Timely antibiotics were defined relative to ED arrival and hypotension status, while timely fluids required ≥30 mL/kg in appropriate patients. About three-quarters received timely antibiotics and half received timely fluid resuscitation by protocol definitions. After risk adjustment, timely antibiotics were associated with a 3 percentage point absolute increase in discharge home. Timely fluid resuscitation was associated with a smaller but statistically significant increase in discharge home. 3

PT/INR and aPTT-based machine learning accurately infers anticoagulant use in acute stroke

This multicenter Japanese study derived and validated machine learning models in 30,767 ischemic and hemorrhagic stroke patients. Models used only PT/INR and aPTT to classify vitamin K antagonist, direct oral anticoagulant, or no oral anticoagulant. In temporal validation, the neural network achieved AUROC ~0.93 for vitamin K antagonists and ~0.84 for direct oral anticoagulants. Precision-recall performance and calibration were acceptable, with the neural network showing highest net clinical benefit. Probability heat maps were created to allow rapid bedside inference of anticoagulant exposure from routine coagulation tests. 4