
OBJECTIVES: To evaluate the impact of implementing a host-protein bacterial/viral test in the emergency department (ED) sepsis alert order set. DESIGN: Pre/postimplementation study. Host-protein testing was automatically ordered both pre- and postimplementation as part of the ED sepsis alert order set; test results were accessible to treating clinicians only postimplementation. The pre- (September 2024 to December 2024) to postimplementation (January 2025 to March 2025) transition also involved education on the host-protein test to ED clinicians. The primary outcome was patient classification into Surviving Sepsis Campaign (SSC) sepsis categories (definite/probable vs. possible vs. unlikely sepsis). SETTING: ED of an academic medical center. PATIENTS: Adults who triggered the ED sepsis alert order set. INTERVENTIONS: Host-protein bacterial/viral test (MeMed BV) automatically ordered in ED sepsis alert order set. Test result was only visible to clinicians postimplementation. MEASUREMENTS AND MAIN RESULTS: There were 893 eligible patients; 505 patients in the preimplementation cohort and 388 in the postimplementation cohort. The distribution of SSC sepsis categories differed between pre- and postimplementation cohorts (χ 2 (2) = 8.73; p = 0.013). The difference was primarily driven by a larger proportion of patients classified as unlikely sepsis (from 32.5% to 41.1%; p = 0.009) and a smaller proportion classified as possible sepsis (from 18.4% to 13.2%; p = 0.043). Among unlikely sepsis patients, host-protein test implementation was associated with ordering fewer cefepime doses (from 3.3 to 2.2; p = 0.044) and a lower hospitalization rate (from 92.1% to 79.9%; p = 0.002). CONCLUSIONS: These findings support an association between host-protein testing for patients triggering ED sepsis alerts and the distribution of possible vs. unlikely sepsis cases. Multicenter clinical utility studies are warranted.
IMPORTANCE: Cefepime is a core antibiotic in sepsis treatment, yet standard dosing yields 10- to 30-fold exposure variability in critical illness. In the early “golden hours” of sepsis, low cefepime levels (affecting up to 70% of cases) may lead to organ dysfunction progression, treatment failure, and antimicrobial resistance. Empiric dose escalation to overcome low levels can increase the risk of dose-dependent neurotoxicity. We developed and qualified a population pharmacokinetic model using patient weight and estimated glomerular filtration rate from creatinine and cystatin C that, in simulation, improved attainment of effective and safe drug levels. OBJECTIVES: To describe the Beta-Lactam Optimization and Outcomes Management (BLOOM) 1 pragmatic feasibility trial. DESIGN, SETTING, AND PARTICIPANTS: This single-center, cluster-randomized, pragmatic feasibility trial (ClinicalTrials.gov: NCT07287332) was approved with a waiver of informed consent. Participants are critically ill adults (age ≥ 18 yr), hospitalized at Mayo Clinic (Rochester, MN), who are receiving cefepime for suspected or documented sepsis. A screening tool embedded in the electronic health record (EHR) identifies eligible patients newly initiated on cefepime (≤ 1 previous dose in 48 hr) who are then randomized to care groups (clusters) that use an individualized dosing algorithm (intervention) or usual-care dosing (control). Exclusions are cephalosporin allergy, acute kidney injury, kidney replacement therapy, extracorporeal membrane oxygenation, and missing inputs required for the algorithm (weight, creatinine, cystatin C). Recruitment began on January 15, 2026, and is expected to conclude by January 2027 (target enrollment, ≥ 300 patients). INTERVENTIONS, OUTCOMES, AND ANALYSIS: For eligible patients, EHR-embedded clinical decision support prompts the cluster-assigned dosing approach, with final dose decisions made by the clinical team. Feasibility, fidelity, and exploratory pharmacokinetic and clinical outcomes will be assessed using mixed methods. CONCLUSIONS: This workflow-integrated pragmatic feasibility trial will determine whether personalized cefepime dosing can be implemented in routine practice to optimize timely sepsis therapy while maintaining safety.
IMPORTANCE: Emergency tracheal intubation in critically ill adults is a high-risk procedure. OBJECTIVES: To evaluate associations of ketamine vs. etomidate with cardiovascular collapse and other peri-intubation outcomes. DESIGN, SETTING, AND PARTICIPANTS: Secondary analysis of individual-participant data from the DEVICE and PREOXI trials, treating induction medication as an observational exposure. The cohort included 2357 adults intubated in emergency departments or ICUs; 1822 received etomidate, 312 ketamine, and 223 other induction regimens. MAIN OUTCOMES AND MEASURES: The primary outcome was cardiovascular collapse, defined as new or increased vasopressor use, nadir systolic blood pressure below 65 mm Hg, or cardiac arrest between induction and 2 minutes after intubation. Secondary outcomes included severe hypoxemia (oxygen saturation < 80%) and procedural outcomes. RESULTS: In weighted analyses, cardiovascular collapse occurred in 14.0% of etomidate recipients and 20.4% of ketamine recipients. In weighted, covariate-adjusted models, ketamine was associated with higher odds of cardiovascular collapse (adjusted odds ratio [aOR], 1.80; 95% CI, 1.24–2.62) and new or increased vasopressor use (aOR, 1.51; 95% CI, 1.02–2.23), nominally lower odds of severe hypoxemia (aOR, 0.50; 95% CI, 0.28–0.88), and shorter modeled time to successful intubation (hazard ratio, 1.16; 95% CI, 1.02–1.33). The latter two associations were not significant after multiplicity adjustment. Estimates for systolic blood pressure below 65 mm Hg and cardiac arrest were too imprecise for reliable inference. Successful intubation on the first attempt and 28-day in-hospital mortality did not clearly differ between ketamine and etomidate. CONCLUSIONS AND RELEVANCE: Compared with etomidate, ketamine was associated with higher odds of cardiovascular collapse, primarily reflecting new or increased vasopressor use. Associations with severe hypoxemia and intubation time did not remain significant after multiplicity adjustment. Because induction medication was not randomized and residual confounding remains possible, these associations should not be interpreted as causal.
OBJECTIVES: To investigate whether neutrophil-to-lymphocyte ratio (NLR), a simple marker of immune dysregulation, is associated with early organ dysfunction progression in emergency department (ED) patients with sepsis. DESIGN: Exploratory secondary analysis of observational clustering cohort. SETTING: Single-center ED. PATIENTS/SUBJECTS: Among 999 patients with adjudicated infection, 364 fulfilled Sepsis-3 criteria for sepsis or septic shock and had available Sequential Organ Failure Assessment (SOFA) scores at admission and 24 hours. INTERVENTIONS: None. MEASUREMENTS AND MAIN RESULTS: Organ dysfunction trajectory was defined according to the evolution in SOFA score during the first 24 hours (ΔSOFA) and categorized as improving (ΔSOFA ≤ −2), stable (ΔSOFA −1 to +1), or worsening (ΔSOFA ≥ +2). Of the 364 patients, 135 (37%) were classified as improving, 187 (51%) as stable, and 42 (12%) as worsening. NLR differed significantly across trajectory groups, with median values of 12 (7–21), 12 (7–22), and 23 (10–37), respectively ( p = 0.015). In multivariable logistic regression adjusted for age, lactate concentration, and baseline SOFA score, NLR remained independently associated with worsening organ dysfunction trajectory (odds ratio [OR] 5.55, 95% CI, 1.86–18.18; p = 0.003). Patients with worsening trajectories more frequently required ICU admission (26% vs. 5% vs. 4%; p < 0.001) and experienced higher 28-day mortality (36% vs 17% vs 17%; p = 0.013). CONCLUSIONS: NLR was independently associated with early worsening of organ dysfunction in ED patients with sepsis. Beyond its prognostic value, NLR may provide complementary information to conventional severity assessment by reflecting immune dysregulation.
Background: The PrOcess Evaluation of Trials In Critical care (POETIC) framework was developed to support systematic evaluation of Context, Fidelity, Dose, Reach, and an integrative assessment of Quality of Delivery (QoD). Its practical use, however, requires explicit guidance on selecting measures, constructing QoD, assessing changing context, and governing prospective use. Methods: We undertook a retrospective comparative synthesis of process evaluations embedded in four randomized U.K. critical care trials: Biomarker-guided antibiotic stewardship in suspected ventilator-associated pneumonia (VAPrapid-2), Provision of Psychological Support to People in Intensive Care (POPPI), Sedation AND Weaning In CHildren (SANDWICH), and Alpha-2 agonists for sedation to produce Better outcomes from critical illness (A2B). We selected them because they informed POETIC’s development or application and provided sufficiently detailed quantitative and qualitative evidence. We mapped the reports, protocols, and process evaluation findings to the five dimensions; compared recurring and contradictory patterns, quantitative-qualitative agreement, measurement and interpretation problems, and mechanism-specific limitations; and translated our learning into guidance for applying the framework. RESULTS: The relevance and measurement of POETIC dimensions varied according to how each intervention was expected to work. Contextual influences changed over time and differed between sites. Combined assessments of QoD required explicit decisions about component selection, weighting, thresholds, aggregation, and missing data. VAPrapid-2 showed why a separate Dose measure and a QoD composite were not meaningful for a one-off diagnostic recommendation; POPPI used a weighted site composite; SANDWICH summarized delivery of four intervention components, staff training, and patient Reach; and A2B combined four equally weighted indicators in a 0–12 site score. This cross-trial learning informed practical guidance on selecting measures, assessing context over time, integrating quantitative and qualitative findings, constructing QoD, and governing prospective use. Conclusions: POETIC is best used as a mechanism-aligned, mixed-method scaffold rather than a requirement to measure every dimension or calculate a universal composite. Prospective use is feasible when outputs, recipients, review times, permitted actions, and safeguards are prespecified. The framework and accompanying toolkit support reproducible assessment while protecting trial integrity.