Lack of Added Predictive Value of Portable Chest Radiography in Diagnosing Ventilator-Associated Pulmonary Infection

摘要

Introduction: The accurate diagnosis of ventilator-associated pulmonary infection (VAPI) poses an ongoing challenge. At our institution, patients in whom VAPI is strongly suspected on the basis of the Clinical Pulmonary Infection Score (CPIS) undergo diagnostic mini-bronchoalveolar lavage (mBAL) with quantitative cultures, followed by empiric antibiotic therapy in our surgical intensive care unit (sICU). We sought to determine the role of portable chest X-radiography (pCXR) in the diagnosis of VAPI. Methods: We conducted a retrospective analysis of mechanically ventilated adult (＞ 18 y of age) patients with suspected VAPI undergoing concomitant pCXR and diagnostic mBAL in a combined tertiary-care unit for trauma and surgical intensive care. Portable chest X-radiographs were evaluated in a blinded manner by surgical intensivists, critical care fellows, general surgical residents, and radiologists, and were rated as: (0) Not suspicious for pneumonia, (1) possible pneumonia, or (2) likely pneumonia. These results were compared with the microbiologic results of mBAL culture. Demographic and clinical characteristics including age, gender, white blood cell count (WBC), temperature, purulence of secretions, blood and urine culture results, and length of hospitalization were correlated with the results of mBAL. Results: Regardless of interpreter specialty or level of training, pCXR had no predictive value for VAPI. The overall sensitivity and specificity of pCXR were 77% and 74%, respectively, and its positive predictive value, negative predictive value, and receiver-operating characteristic (ROC) curve areas all had values below 50%. The inter-rater agreement (p) was 0.965, showing little discrepancy between raters. The degree of purulence on mBAL, concurrent blood stream infection, and increase in the number of days of hospitalization before diagnostic testing were correlated with an increased frequency of VAPI. The three CPIS criteria of febrile response, leukocytosis/leukopenia, and arterial oxygenation correlated poorly with the results of mBAL culture. Conclusion: Portable chest X-radiography has no added predictive value in identifying patients who should be evaluated further for VAPI. This supports the elimination of findings on chest X-radiography as defining characteristics of VAP, which accords with the U.S. Centers for Disease Control and Prevention's recent definition of VAP as but one of a number of types of ventilator-associated pulmonary infection (VAPI).