Ruling In and Ruling Out Sepsis Using Likelihood Ratios of a Host Response Assay

Area of Science:

• Clinical diagnostics
• Molecular diagnostics
• Sepsis research

Background:

• Sepsis diagnosis remains challenging, often requiring differentiation from systemic inflammatory response syndrome (SIRS).
• The SeptiCyte RAPID gene expression test provides a SeptiScore to assess sepsis likelihood.
• Likelihood ratios (LR+ and LR-) offer a quantitative framework for interpreting diagnostic test results.

Purpose of the Study:

• To evaluate the clinical utility of likelihood ratios (LR+ and LR-) derived from the SeptiCyte RAPID SeptiScore for diagnosing sepsis versus SIRS.
• To assess the performance of SeptiCyte RAPID as a rule-out test using low SeptiScores and LR-.
• To assess the performance of SeptiCyte RAPID as a rule-in test using high SeptiScores and LR+.

Main Methods:

• Retrospective analysis of prospectively collected data from 889 critically ill patients suspected of sepsis across 19 hospitals.
• Quantification of LR+ and LR- as a function of SeptiScore for sepsis versus SIRS discrimination in ICU patients.
• Stratification of sepsis patients into blood culture positive (BC+) and blood culture negative (BC-) subgroups for detailed analysis.

Main Results:

• SeptiCyte RAPID demonstrated strong diagnostic performance, with median LR+ of 6.97 for Band 4 and median LR- of 0.085 for Band 1.
• A monotonic increase in likelihood ratio was observed with increasing SeptiScore, irrespective of blood culture status.
• LR+ and LR- values across SeptiScore bands provide clear clinical utility for ruling sepsis in or out.

Conclusions:

• High SeptiScores correlate with high LR+ values, supporting sepsis rule-in, potentially aiding early diagnosis and stewardship.
• Low SeptiScores correlate with low LR- values, supporting sepsis rule-out and prompting consideration of alternative diagnoses.
• The LR framework complements SeptiScore banding, offering a clinically useful approach to sepsis diagnosis.