Impact of distinct definitions of acute lung injury on its incidence and outcomes in Brazilian ICUs: prospective evaluation of 7,133 patients*.

OBJECTIVES: Evaluation of prevalence and outcomes of acute lung injury in a large cohort of critically ill patients in Brazil and comparison of predictive receiver operating characteristic curve mortality of American European Consensus conference definition with new Berlin definition of acute respiratory distress syndrome. DESIGN: A 15-month prospective, multicenter, observational study. SETTING: Fourteen medical ICUs in Espirito Santo, a state of Brazil. PATIENTS: Mechanically ventilated patients who fulfilled American European Consensus conference criteria of acute lung injury or Berlin definition of acute respiratory distress syndrome. INTERVENTIONS: Clinical and respiratory data were collected for 7 consecutive days and on the 14 and 28 days. Twenty-eight day mortality, hospital mortality, and predictive receiver operating characteristic curve mortality were calculated. MEASUREMENTS AND MAIN RESULTS: Of 7,133 patients, 130 patients (1.8%) fulfilled criteria for acute lung injury (American European Consensus conference) or acute respiratory distress syndrome (Berlin definition). Median time for diagnosis was 2 days (interquartile range, 0-3 d). Main risk factors were pneumonia (35.3%) and nonpulmonary sepsis (31.5%). Mean age was 44.2 ± 15.9 years, and 61.5% were men. Mean Acute Physiology and Chronic Health Evaluation II score was 20.7 ± 7.9. Mean PaO2/FIO2 was 206 ± 61.5, significantly lower in nonsurvivors on day 7 (p = 0.003). Mean mechanical ventilation time was 21 ± 15 days. Length of ICU stay was 26.4 ± 18.7 days. Twenty-eight-day mortality was 38.5% (95% CI, 30.1-46.8); hospital mortality was 49.2% (95% CI, 40.6-57.8). Predictive 28-day mortality area under the receiver operating characteristic curve for American European Consensus conference definition was 0.5625 (95% CI, 0.4783-0.6467) and for the Berlin definition 0.5664 (95% CI, 0.4759-0.6568; p = 0.9510). CONCLUSIONS: In our population, prevalence of acute lung injury was low, most cases were diagnosed 2 days after ICU admission, and Berlin definition was not different from American European Consensus conference definition in predicting mortality. There are still several problems with the global epidemiology, definition, and mortality predictive indices that should be added to the classification of this still lethal syndrome to improve its predictive mortality power in the future.