Ruling out septic arthritis risk in a few minutes using mid-infrared spectroscopy in synovial fluids.

Objectives: The aim of this study was to show the usefulness of a mid-infrared fibre evanescent wave spectroscopy point of care device in the identification of septic arthritis patients in a multicentre cohort, and to apply this technology to clinical practice among physicians.
Methods: SF samples from 402 patients enrolled in a multicentre cohort were frozen for analysis by mid-infrared fibre evanescent wave spectroscopy. The calibration cohort was divided into two groups of patients (septic arthritis and non-septic arthritis) and relevant spectral variables were used for logistic regression model. Model performances were tested on an independent set of 86 freshly obtained SF samples from patients enrolled in a single-centre acute arthritis cohort and spectroscopic analyses performed at the patient's bedside.
Results: The model set-up, using frozen-thawed SFs, provided good performances, with area under the curve 0.95, sensitivity 0.90, specificity 0.90, positive predictive value 0.41 and negative predictive value 0.99. Performances obtained in the validation cohort were area under the curve 0.90, sensitivity 0.92, specificity 0.81, positive predictive value 0.46 and negative predictive value 0.98. The septic arthritis probability has been translated into a risk score from 0 to 4 according to septic risk. For a risk score of 0, the probability of identifying a septic patient is very low (negative predictive value of 1), whereas a risk score of 4 indicates very high risk of septic arthritis (positive predictive value of 1).
Conclusion: Mid-infrared fibre evanescent wave spectroscopy could distinguish septic from non-septic synovial arthritis fluids with good performances, and showed particular usefulness in ruling out septic arthritis. Our data supports the possibility of technology transfer.
Trial Registration: ClinicalTrials.gov, http://clinicaltrials.gov, NCT02860871.
(© The Author(s) 2020. Published by Oxford University Press on behalf of the British Society for Rheumatology. All rights reserved. For permissions, please email: journals.permissions@oup.com.)