Analgesia-Based Sedation During Mechanical Ventilation

Overview

Hypothesis: A protocolized algorithm for sedation in critically ill patients on mechanical ventilation can decrease ventilator days, costs and improve outcome. This is a multicenter observational-interventional study on critically ill patients who require mechanical ventilation for more than 48 hours, involving 13 ICU in Chile. There are two periods (groups): a descriptive phase of sedation practices, and an interventional period in which an analgesia-based, goal-directed, nurse-driven sedation is applied. Main outcome: ventilator-free days between both periods.

This is a multicenter observational-interventional study on critically ill patients who require mechanical ventilation for more than 48 hours, involving 13 ICU in Chile. Main exclusion criteria are neurologic impairment, previous chronic cardiac, liver and renal failure, second period of mechanical ventilation during hospitalization, short term expected mortality. There is an initial descriptive phase of sedation practice, involving sedative drugs, sedation level assessment, and outcome: ventilator-free days, ICU stay, costs and mortality. After a period of analysis and training, an analgesia-based, goal-directed, nurse-driven sedation protocol is applied. Fentanyl infusion is started and titrated to obtain a patient calm and cooperative or mildly sedated while on mechanical ventilation. Hypnotics and opiates i.v. boluses are allowed during the first hours of mechanical ventilation. Midazolam infusion is started if ventilatory distress continue despite fentanyl 1.8 µg/kg/min. Haldol, muscle relaxants and other sedative drugs are allowed depending on patient condition. After 48 hours of mechanical ventilation, sedative drugs are discontinued in the morning. Main outcome: ventilator-free days. Secondary outcome: ICU stay, costs, and sedation quality. At 6 moths SF-36. Each period is planned to included at least 140 patients, for a 20% difference in ventilator-free days, with 80% power and a 0.05 type I error.