Causes and predictors of nonresponse to treatment of intensive care unit–acquired pneumonia*

Abstract

To prospectively evaluate the predictive factors for the nonresponse to empirical antibiotic treatment and mortality in patients with intensive care unit–acquired pneumonia. A 1-yr prospective cohort of patients with suspicion of intensive care unit–acquired pneumonia. Five medical and surgical intensive care units of Hospital Clinic in Barcelona. A total of 71 patients with intensive care unit–acquired pneumonia were studied. The definition of nonresponse included at least one of the following: failure to improve the Pao2/Fio2 ratio or need of intubation because of pneumonia, persistence of fever or hypothermia and purulent respiratory secretions, worsening of pulmonary infiltrates, or occurrence of septic shock or multiple organ dysfunction not present at onset of pneumonia. Clinical assessment, including severity scores, blood and quantitative cultures of respiratory secretions, and cytokine measurements in serum and bronchoalveolar lavage at onset of pneumonia and 72 hrs after antimicrobial treatment. A total of 44 patients (62%) fulfilled criteria of nonresponse, and at least one cause of nonresponse could be determined in 28 cases (64%): inappropriate treatment in ten (23%), superinfection in six (14%), concomitant foci of infection in 12 (27%), and noninfectious causes in seven cases (16%). The remaining 16 patients with no definite cause of nonresponse presented with septic shock, multiple organ dysfunction, or acute respiratory distress syndrome. Increased levels of interleukin-6 at onset of pneumonia (odds ratio, 9.7; p = .014) was an independent predictor of nonresponse to treatment. Likewise, increased level of interleukin-6 at follow-up (odds ratio, 27; p = .001) was the only independent predictor for hospital mortality. Increased systemic inflammatory response was the main predictor of nonresponse to treatment and mortality.