Incidence and microbiological characteristics of ventilator-associated pneumonia assessed by bronchoalveolar lavage and endotracheal aspirate in COVID-19 patients: COV-AP study

ABSTRACT

Introduction: Diagnosis of ventilator-associated pneumonia (VAP) in COVID-19 patients remains challenging. Also, the lack of gold standard for microbiological sampling undermines clinical judgement and management. We studied incidences of microbiologically-confirmed VAP comparing endotracheal aspirate (ETA) and bronchoalveolar lavage (BAL) in COVID-19 patients. Etiological agreement between ETA and BAL was then assessed. Method(s) Single-center prospective cohort study (NCT04766983). Patients were enrolled within 48 h from intubation;surveillance ETA ( ETASURV) was performed twice weekly. ETA ( ETACX) and BAL ( BALCX) samples were collected upon VAP suspicion (Johanson's criteria). CDC definitions were used for microbiological confirmation. ETA-BAL agreement (interrater reliability and Cohen's kappa) and clinical/microbiological data were assessed for the first episodes of suspected VAP per patients. Result(s) Ninety intensive care (ICU) patients enrolled from 01/2021 to 05 06/2022, of which 26 females (28.9%);median age was 60 [52-66] years. In-ICU mortality was 30/90 (33.3%), median length of stay in survivors 19 (10-32) days. Fifty-three patients (58.9%) had >= 1 episode of suspected VAP after 6 [5;10] days from ICU admission. ETASURV were available in 52 cases, 2 [1;3] days before VAP suspect, and tested positive in 28 (53.8%). ETACX and BALCX resulted positive in 35 (66.0%) and 29 (54.7%) of episodes. Main microbiological results are displayed in Fig. 1, panel A. Etiological agreement between techniques is shown in Fig. 1, panel B. Incidence rate of VAP suspicions per 1000 ventilator-days was 60.2 (95% CI 43.9-76.4), while incidence rates of microbiologically-confirmed VAP were 27.4 (18.3-36.5) with ETACXand 18.9 (95% CI 12.0-25.8) with BALCX, respectively. Conclusion(s) We observed different incidence of VAP in COVID-19 ICU patients depending on sampling method. Etiological agreement between techniques yielded limited interrater reliability. The potential clinical impact needs further studies.