Scholarlink

Background <p>Diagnosing healthcare-associated infections (HCAIs) in critically ill trauma patients is difficult, because trauma-related non-septic systemic inflammatory response syndrome (SIRS) closely mimics infection and conventional biomarkers lack specificity. SeptiCyte RAPID is a fully automated host-response assay that quantifies the expression of two genes and returns a <i>SeptiScore</i> within one hour, making it suitable for bedside use. Our primary objective was to determine an optimal SeptiScore threshold for identifying HCAIs in a derivation cohort and to evaluate it in an independent validation cohort; secondary objectives were to assess diagnostic performance and to compare SeptiScore with conventional biomarkers.</p> Methods <p>This two-center diagnostic study included a prospective derivation cohort (82 samples, 44 patients) and a retrospective validation cohort (31 samples, 31 patients), comprising adult ICU trauma patients with SIRS and hemodynamic failure. SeptiScore was measured at clinical suspicion, before microbiological confirmation. Because some patients contributed several samples, the association between SeptiScore and infection was analyzed using linear mixed-effects models with a random patient intercept, and confidence intervals for diagnostic metrics were estimated using a patient-level cluster bootstrap. Diagnostic accuracy was assessed by ROC curves, the Youden index, and negative likelihood ratios (LR −), and compared with C-reactive protein, procalcitonin, and leukocyte count.</p> Results <p>Overall, 113 samples (71 trauma-related SIRS, 42 HCAIs) were analyzed. SeptiScore was higher in HCAIs (median 8.05, IQR 6.83–8.80) than in trauma-related SIRS (5.80, 5.08–7.05), and infection status remained independently associated with SeptiScore after accounting for within-patient correlation (<i>p</i> &lt; 0.001). In the derivation cohort, the area under the ROC curve was 0.79 (95% CI 0.72–0.87). A threshold of 6.7 yielded sensitivity 0.84 (0.72–0.97), specificity 0.66 (0.53–0.79), and LR − 0.24 (0.04–0.53), outperforming conventional biomarkers. In the smaller validation cohort, the pre-specified threshold showed consistent performance (AUC 0.98; 95% CI 0.94–1.00), although these estimates are imprecise given the limited number of events.</p> Conclusions <p>SeptiScore showed promising performance for distinguishing HCAIs from trauma-related SIRS and outperformed conventional biomarkers. These hypothesis-generating findings require confirmation in larger, adequately powered multicenter studies before clinical use.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Clinical evaluation of the SeptiScore biomarker for the diagnosis of healthcare-associated infections in critically ill trauma patients: a multicenter derivation and validation cohort study

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年7月10日

查看全文
  • Marwan Bouras,
  • Lisa Neutre,
  • Chloé Douarec,
  • Thomas Goronflot,
  • Cécile Poulain,
  • Alexandre Bourdiol,
  • Alexiane Blanc,
  • Charlotte Arbelot,
  • Marc Leone,
  • Marc G. Denis,
  • Antoine Roquilly

摘要

Background

Diagnosing healthcare-associated infections (HCAIs) in critically ill trauma patients is difficult, because trauma-related non-septic systemic inflammatory response syndrome (SIRS) closely mimics infection and conventional biomarkers lack specificity. SeptiCyte RAPID is a fully automated host-response assay that quantifies the expression of two genes and returns a SeptiScore within one hour, making it suitable for bedside use. Our primary objective was to determine an optimal SeptiScore threshold for identifying HCAIs in a derivation cohort and to evaluate it in an independent validation cohort; secondary objectives were to assess diagnostic performance and to compare SeptiScore with conventional biomarkers.

Methods

This two-center diagnostic study included a prospective derivation cohort (82 samples, 44 patients) and a retrospective validation cohort (31 samples, 31 patients), comprising adult ICU trauma patients with SIRS and hemodynamic failure. SeptiScore was measured at clinical suspicion, before microbiological confirmation. Because some patients contributed several samples, the association between SeptiScore and infection was analyzed using linear mixed-effects models with a random patient intercept, and confidence intervals for diagnostic metrics were estimated using a patient-level cluster bootstrap. Diagnostic accuracy was assessed by ROC curves, the Youden index, and negative likelihood ratios (LR −), and compared with C-reactive protein, procalcitonin, and leukocyte count.

Results

Overall, 113 samples (71 trauma-related SIRS, 42 HCAIs) were analyzed. SeptiScore was higher in HCAIs (median 8.05, IQR 6.83–8.80) than in trauma-related SIRS (5.80, 5.08–7.05), and infection status remained independently associated with SeptiScore after accounting for within-patient correlation (p < 0.001). In the derivation cohort, the area under the ROC curve was 0.79 (95% CI 0.72–0.87). A threshold of 6.7 yielded sensitivity 0.84 (0.72–0.97), specificity 0.66 (0.53–0.79), and LR − 0.24 (0.04–0.53), outperforming conventional biomarkers. In the smaller validation cohort, the pre-specified threshold showed consistent performance (AUC 0.98; 95% CI 0.94–1.00), although these estimates are imprecise given the limited number of events.

Conclusions

SeptiScore showed promising performance for distinguishing HCAIs from trauma-related SIRS and outperformed conventional biomarkers. These hypothesis-generating findings require confirmation in larger, adequately powered multicenter studies before clinical use.