Purpose: The aim of the study was to assess the utility of pretest probability and modified clinical pulmonary infection score CPIS in the diagnosis of late-onset ventilator-associated pneumonia (VAP).
Materials and methods: In 740 adults enrolled in a multicenter randomized trial, intensivists prospectively rated the pretest probability of VAP as low, moderate, or high based on their clinical judgment. The modified CPIS was calculated without considering culture results. Ventilator-associated pneumonia diagnosis was determined by 2 adjudicators using standardized definitions. We analyzed the relationship between pretest likelihood, CPIS, and VAP diagnosis.
Results: Among the 739 patients analyzed, 14.5%, 39.6%, and 45.9% had low, moderate, and high pretest probability of VAP. Patients with high pretest probability had a lower PaO2/FiO2 ratio and a larger volume of secretions. High or moderate vs low pretest probability had high sensitivity (0.88; 95% confidence interval [CI], 0.87-0.89) and positive predictive value (0.87; 95% CI, 0.86-0.88) but low specificity (0.27; 95% CI, 0.21-0.35) and negative predictive value (0.29; 95% C,: 0.22-0.37) for the diagnosis of VAP. Therefore, 71% of patients who had a low pretest probability were actually infected (1 - negative predictive value). The area under the receiver operating characteristic curve for the modified CPIS was not significant (0.47; 95% CI, 0.42-0.53), meaning that no score threshold was clinically useful.
Conclusions: Pretest probability and a modified CPIS, which excludes culture results, are of limited utility in the diagnosis of late-onset VAP.