Predictors of treatment failure following early antibiotic discontinuation in culture-negative, ventilator-associated pneumonia: an observational study.

Abstract

Introduction: Early antibiotic discontinuation in clinically suspected ventilator-associated pneumonia (VAP) may lead to infection relapse/recurrence and increase mortality. This study aimed to evaluate the incidence and potential predictors of treatment failure with this approach.

Methodology: A retrospective observational study was conducted between September 2014 and November 2016 in a mixed intensive care unit. We included clinically suspected VAP patients whose quantitative sputum cultures from endotracheal aspirate were negative, allowing antibiotic discontinuation within 24 hours. Patients were monitored for signs and symptoms of recurrent VAP. Incidence and risk factors for treatment failure, defined as pneumonia recurrence, were determined using univariate logistic regression analysis and receiver operating characteristic (ROC) curves.

Results: Forty-three patients met the inclusion criteria. The incidence of treatment failure among culture-negative VAP following early antibiotic discontinuation was 27.9% (12 patients). There were no significant differences in procalcitonin levels, leukocyte counts or body temperature between the two groups, except for the modified clinical pulmonary infection score (mCPIS) (5.42 ± 2.19 versus 3.9 ± 1.54, p = 0.014). Procalcitonin levels at VAP diagnosis and antibiotic cessation both showed low predictive capacity for treatment failure (AUC 0.56, CI 95% 0.36-0.76 and AUC 0.57, CI 95% 0.37-0.76, respectively). However, combining mCPIS with procalcitonin improved the predictive value for treatment failure (AUC 0.765, CI 95% 0.56-0.96).

Conclusions: Early antibiotic discontinuation may lead to a high incidence of treatment failure among culture-negative VAP patients. Procalcitonin alone should not guide antibiotic discontinuation decisions while combining mCPIS and procalcitonin enhances predictive accuracy for treatment failure.