Study objectives: To determine the incidence of prolonged mechanical ventilation (PMV), which is associated with increased health-care costs and risks of adverse events, and to identify its early predictors.
Design: Retrospective cohort.
Setting: A medical-surgical ICU in a university-affiliated hospital.
Patients or participants: All patients admitted to the ICU over 3 years who received mechanical ventilation (MV) for > 12 h.
Measurements: PMV was defined as MV lasting > 21 days. We recorded epidemiologic data, severity scores, worst Pao2/fraction of inspired oxygen (Fio2), presence of shock on ICU admission day, cause for MV, length of MV, ICU length of stay (LOS), and hospital LOS. PMV patients were compared to patients weaned before 21 days (non-PMV group) to determine predictors of PMV.
Results: Of 551 hospital admissions, 319 patients (58%) required MV > 12 h. One hundred thirty patients died early and were excluded. Seventy-nine patients (14%) required PMV. The non-PMV group consisted of 110 patients. Simplified acute physiology score (SAPS) II, APACHE (acute physiology and chronic health evaluation) II, therapeutic intervention scoring system, Pao2/Fio2, shock, ICU LOS, and hospital LOS differed significantly between groups. However, logistic regression identified shock on ICU admission day as the only independent predictor of PMV (odds ratio, 3.10; p = 0.001). SAPS II and Pao2/Fio2 had the nearest coefficients and were used to build the predictive model. Sensitivity analysis was performed including the 130 patients who died early, and shock remained the most powerful predictor.
Conclusions: PMV was a frequent event in this cohort. The presence of shock on ICU admission day was the only prognostic factor, even adjusting for severity of illness and hypoxemia.