Daily Sepsis Research Analysis

OBJECTIVE: To identify bias in using a single machine learning (ML) sepsis prediction model across multiple hospitals and care locations; evaluate the impact of six different bias mitigation strategies and propose a generic modelling approach for developing best-performing models. METHODS: We developed a baseline ML model to predict sepsis using retrospective data on patients in emergency departments (EDs) and wards across nine hospitals. We set model sensitivity at 70% and determined the number of alerts required to be evaluated (number needed to evaluate (NNE), 95% CI) for each case of true sepsis and the number of hours between the first alert and timestamped outcomes meeting sepsis-3 reference criteria (HTS3). Six bias mitigation models were compared with the baseline model for impact on NNE and HTS3. RESULTS: Across 969 292 admissions, mean NNE for the baseline model was significantly lower for EDs (6.1 patients, 95% CI 6 to 6.2) than for wards (7.5 patients, 95% CI 7.4 to 7.5). Across all sites, median HTS3 was 20 hours (20-21) for wards vs 5 (5-5) for EDs. Bias mitigation models significantly impacted NNE but not HTS3. Compared with the baseline model, the best-performing models for NNE with reduced interhospital variance were those trained separately on data from ED patients or from ward patients across all sites. These models generated the lowest NNE results for all care locations in seven of nine hospitals. CONCLUSIONS: Implementing a single sepsis prediction model across all sites and care locations within multihospital systems may be unacceptable given large variances in NNE across multiple sites. Bias mitigation methods can identify models demonstrating improved performance across most sites in reducing alert burden but with no impact on the length of the prediction window.

BACKGROUND: Sepsis-associated acute kidney injury (SA-AKI) is a prevalent and severe complication in critically ill patients. However, diagnostic and therapeutic advancements have been hindered by the biological heterogeneity underlying the disease. Both transcriptomic endotyping and biomarker profiling have been proposed individually to identify molecular subtypes of sepsis and may enhance risk stratification. This study aimed to evaluate the utility of combining transcriptomic endotyping with protein-based biomarkers for improving risk stratification in SA-AKI. METHODS: This secondary analysis of the PredARRT-Sep-Trial included 167 critically ill patients who met Sepsis-3 criteria. Patients were stratified into three transcriptomic endotypes-inflammopathic (IE), adaptive (AE), and coagulopathic (CE)-using a validated whole-blood gene expression classifier. Eight protein-based biomarkers encompassing kidney function, vascular integrity, and immune response were measured. Predictive performance for the primary endpoint kidney replacement therapy or death was assessed using receiver operating characteristic curve analysis and logistic regression models. RESULTS: Stratification into transcriptomic endotypes assigned 33% of patients to IE, 42% to AE, and 24% to CE. Patients classified as IE exhibited the highest disease severity and were most likely to meet the primary endpoint (30%), compared to AE and CE (17% and 10%, respectively). Kidney function biomarkers showed stepwise increases with AKI severity across all endotypes, whereas non-functional biomarkers (neutrophil gelatinase-associated lipocalin [NGAL], soluble urokinase plasminogen activator receptor [suPAR], and bioactive adrenomedullin [bio-ADM]) exhibited endotype-specific differences independent of AKI severity. NGAL and suPAR levels were disproportionately elevated in the IE group, suggesting a dominant role of innate immune dysregulation in this endotype. In contrast, bio-ADM, a marker of endothelial dysfunction, was the strongest risk-predictor of outcomes in CE. The combination of transcriptomic endotyping with protein-based biomarkers enhanced predictive accuracy for the primary endpoint and 7-day mortality, with the highest area under the receiver operating characteristic curve of 0.80 (95% CI 0.72-0.88) for endotyping + bio-ADM and 0.85 (95% CI 0.78-0.93) for endotyping and suPAR, respectively. Combinations of endotyping with functional and non-functional biomarkers particularly improved mortality-related risk stratification. CONCLUSIONS: Combining transcriptomic endotyping with protein-based biomarker profiling enhances risk-stratification in SA-AKI, offering a promising strategy for personalized treatment and trial enrichment in the future. Further research should validate these findings and explore therapeutic applications.

BACKGROUND: Refractory septic shock can lead to multiorgan failure and death due to myocardial dysfunction-induced inadequate tissue perfusion. Current guidelines advocate inotropic adjuncts to norepinephrine, but the efficacy of milrinone remains understudied in this context. This study aimed to evaluate the hemodynamic changes in septic shock patients treated with adjunctive milrinone compared to those treated with a placebo. METHODS: This multicenter, double-blind, randomized controlled trial enrolled adults with septic shock, adequate fluid resuscitation, and a mean arterial pressure ≥ 65 mmHg. Eligible patients exhibited poor tissue perfusion or impaired left ventricular systolic function. Participants were randomized 1:1 to milrinone or placebo. Echocardiographic hemodynamic assessments were performed pre- and postintervention. The primary outcome was the change in cardiac output from baseline to 6 h after drug administration. The study was prospectively registered at www.clinicaltrials.gov (NCT05122884). RESULTS: Among 271 screened patients, 64 were randomized. The baseline characteristics were comparable between the groups. The milrinone group demonstrated a significantly greater change in cardiac output at 6 h (median [IQR] 0.62 L/min [-0.51 to 1.47]) than did the placebo group (0.13 L/min [-0.59 to 0.46]; CONCLUSIONS: Milrinone administration in septic shock patients improved cardiac output at 6 h, suggesting a potential benefit for patients with persistent tissue hypoperfusion despite norepinephrine.