(1426) Accurate Sepsis Identification Using SeptiCyte RAPID vs Other Biomarkers of Systemic Inflammation

Introduction: Accurate identification of patients with sepsis can be challenging because of vague presenting clinical signs and a lack of timely microbiology results [1]. Improved methods of sepsis identification have included single, or combined, biomarkers of systemic inflammation [2]. We hypothesized that SeptiCyte® RAPID(SeptiScore), which measures two immune biomarkers to differentiate sepsis from Systemic Inflammatory Response Syndrome(SIRS), could be combined with other biomarkers to improve sepsis identification.

Methods: Performance of SeptiScore and 5 additional biomarkers (ROC-AUC) were compared to a Retrospective Physician Diagnosis (RPD) [2] for 378 patients where each panel physician adjudicated either sepsis or SIRS (sepsis, n=154; SIRS, n=224). A permutation analysis of 6 variables (SeptiScore, C-Reactive Protein (CRP), procalcitonin (PCT), lactate, white blood cell count (WBC Min & Max within 24 hrs of ICU admission), alone, or in any combination, was utilized resulting in 63 logistic regression models. Missing values were imputed using a multivariable imputation algorithm, Amelia [3] for WBC.Min (n=9), WBC.Max (n=14), PCT (n=68), lactate (n=125) & CRP (n=337). Further analysis of SeptiCyte RAPID performance included categorization of patients into five groups using a composite score whereby 1 point was given for each biomarker being abnormal or elevated:PCT≥0.50ng/mL, CRP>10mg/L, lactate≥1mmol/L, WBC < 4 OR >12x103 cells/mm3.

Results: Mean AUCs for SeptiScore, PCT, CRP, WBC Min & Max & lactate were 0.85, 0.75, 0.60, 0.60, 0.57 & 0.54, respectively. Use of all five other biomarkers in combination with SeptiScore had an AUC of 0.85 (95% CI: 0.85-0.86). A model with CRP, PCT, Lactate, WBC (Min & Max) had mean AUC (0.74, 95%CI: 0.69-0.81) compared to models with SeptiScore (mean AUC=0.85). There were 7 patients retrospectively diagnosed with sepsis that had a composite score of 0. SeptiScore AUCs were 0.83, 0.82 & 0.85 for patients with composite scores of 1, 2 or 3, respectively. The mean SeptiScore score for sepsis patients with a composite score of 4 (n=8) was 8.8 (range 0-15).

Conclusions: No combination of biomarkers outperformed SeptiScore alone or models with it at diagnosing sepsis. In this study, adding other biomarkers to SeptiScore did not increase the diagnostic certainty of sepsis.