Clinical judgment was inaccurate for the diagnosis and treatment of ventilator-associated pneumonia in ICU patients who needed mechanical ventilation
ACP J Club. 1993 July-Aug;119:22. doi:10.7326/ACPJC-1993-119-1-022
Fagon JY, Chastre J, Hance AJ, et al. Evaluation of clinical judgment in the identification and treatment of nosocomial pneumonia in ventilated patients. Chest. 1993 Feb;103:547-53.
To evaluate clinical judgment in the diagnosis and treatment of ventilator-associated pneumonia (VAP) in patients receiving mechanical ventilation.
A cohort study in which a clinical diagnosis of VAP and the associated management plan were compared with a protected brush catheter (PBC) diagnosis of pneumonia and PBC-directed antibiotics.
A university-affiliated intensive care unit (ICU) in France.
84 patients (mean age 61 y, 73% men) who required mechanical ventilation for > 72 hours in the ICU had infiltrates on chest radiographs, had macroscopically purulent tracheal aspirates, and were suitable for flexible fiberoptic bronchoscopy. Exclusion criteria were changes in antibiotic strategy in the previous 3 days, decontamination of the digestive tract, endotracheally administered antibiotics, or previous enrollment.
Description of tests and diagnostic standard
Within 12 hours of entry and before any new antibiotics were administered, fiberoptic bronchoscopy and PBC were used to obtain secretions for quantitative cultures (diagnostic standard). Immediately after bronchoscopy, using data collected for the diagnosis of VAP, physicians made independent clinical predictions of pneumonia and planned antimicrobial therapy. 4 diagnostic categories were based on the patient's hospital course: definite bacterial pneumonia (positive pleural fluid culture, rapid activation of infiltrate, or histologic confirmation); probable bacterial pneumonia (positive PBC culture); definite no pneumonia (recovery without changing antibiotics); and probable no pneumonia (no culture growth while receiving antibiotics).
Main outcome measures
Quantitative culture results were compared with preculture, postbronchoscopy clinical decisions.
27 patients had VAP (17 definite and 10 probable); 57 patients did not have VAP (34 definite and 23 probable). No physical findings accurately predicted pneumonia. Clinicians made 408 predictions of pneumonia; 77% of these were accurate (84% accuracy for the absence of pneumonia and 62% accuracy for the presence of pneumonia). The accuracy of senior consultants, junior physicians, and residents did not differ (77%, 72%, and 78%, respectively). Antibiotic treatment was appropriate for 43 of 131 (33%) therapeutic plans for patients with pneumonia and inappropriate for 45 of 277 (16%) without pneumonia.
Clinical judgment was inaccurate for the diagnosis and treatment of ventilator-associated pneumonia in patients in an ICU who required mechanical ventilation.
Source of funding: In part, Faculté Xavier Bichat.
For article reprint: Dr. J.Y. Fagon, Hôpital Broussais, Service Reanimation Médicale, 96 rue Didot, 75014 Paris, France. FAX 33-1-4545-0545.
The diagnosis of VAP is a common clinical problem often diagnosed by sampling of lower respiratory tract secretions using PBC or bronchoalveolar lavage (BAL). In patients with suspected nosocomial pneumonia, Fagon and colleagues compared the diagnostic predictions and therapeutic plans independently formulated by a group of physicians with those resulting from quantitative culture results of PBC fluid. The results show how the commonly used nonspecific clinical and paraclinical features of pneumonia yield an accuracy rate of only 60%, consistent with previously published data. A caveat in the application of these data is that the study comes from centers with expertise in invasive respiratory diagnostics. If clinicians do not withhold antibiotics before PBC sampling, as is often the case in North America, the potential for false-negative PBCs is further increased.
These investigators showed that using PBC results can minimize false-negative and false-positive diagnoses and improve the appropriateness of antimicrobial therapy. These findings have stimulated investigators to further examine the effect of PBC or BAL on patient outcome. Timsit and colleagues (1) found that a suspicion of pneumonia using the usual clinical radiologic criteria predicted mortality but that confirmed VAP by PBC or BAL added no prognostic information. In another study examining the influence of BAL on antibiotic administration, Luna and colleagues (2) found a higher mortality associated with inappropriate empiric antibiotics according to BAL as compared with appropriate empiric antibiotics. However, use of invasive diagnostic approaches to VAP are now being tested in a large randomized trial to determine more carefully their effect on clinically important outcomes and ICU costs.
Deborah J. Cook, MD, MSc
McMaster UniversityHamilton, Ontario, Canada