Electronic nose analysis of exhaled breath to diagnose ventilator-associated pneumonia

Background: Exhaled breath analysis is an emerging technology in respiratory disease and infection. Electronic nose devices (e-nose) are small and portable with a potential for point of care application. Ventilator-associated pneumonia (VAP) is a common nosocomial infection occurring in the intensive care unit (ICU). The current best diagnostic approach is based on clinical criteria combined with bronchoalveolar lavage (BAL) and subsequent bacterial culture analysis. BAL is invasive, laborious and time consuming. Exhaled breath analysis by e-nose is non-invasive, easy to perform and could reduce diagnostic time. Aim of this study was to explore whether an e-nose can be used as a non-invasive in vivo diagnostic tool for VAP. Methods: Seventy-two patients met the clinical diagnostic criteria of VAP and underwent BAL. In thirty-three patients BAL analysis confirmed the diagnosis of VAP [BAL+(VAP+)], in thirty-nine patients the diagnosis was rejected [BAL-]. Before BAL was performed, exhaled breath was sampled from the expiratory limb of the ventilator into sterile Tedlar bags and subsequently analysed by an e-nose with metal oxide sensors (DiagNose, C-it, Zutphen, The Netherlands). From further fifty-three patients without clinical suspicion of VAP or signs of respiratory disease exhaled breath was collected to serve as a control group [control(VAP-)]). The e-nose data from exhaled breath were analysed using logistic regression. Results: The ROC curve comparing [BAL+(VAP+)] and [control(VAP-)] patients had an area under the curve (AUC) of 0.82 (95% CI 0.73-0.9). The sensitivity was 88% with a specificity of 66%. The comparison of [BAL+(VAP+)] and [BAL-] patients revealed an AUC of 0.69; 95% CI 0.57-0.81) with a sensitivity of 76% with a specificity of 56%. Conclusion: E-nose lacked sensitivity and specificity in the diagnosis of VAP in the present study for current clinical application. Further investigation into this field is warranted to explore the diagnostic possibilities of this promising new technique. (C) 2015 Elsevier Ltd. All rights reserved.