Multicenter, randomized, prospective trial of early tracheostomy

Objectives: Determine the effect of early (days 3-5) or late (days 10- 14)tracheostomy on intensive care unit length of stay (ICU LOS), frequency of pneumonia, and mortality, and evidence of short-term or long-term pharyngeal, laryngeal, or tracheal injury in head trauma, non-head trauma, and critically ill nontrauma patients. Study Design: Randomized, prospective. Setting: Five Level I trauma centers. Methods: Data were obtained prospectively and included Acute Physiology and Chronic Health Evaluation III score (AIII), Glasgow Coma Scale score, Emergency Room Trauma Score, Injury Severity Score, Acute Injury Score, type of endotracheal tube or tracheostomy, level of positive end-expiratory pressure, and peak inspiratory pressure. Patients were to undergo laryngoscopy for detection of injury according to the Lindholm criteria at the time of endotracheal tube or tracheostomy removal and be reevaluated at 3 to 5 months after discharge. Results: One hundred fifty-seven patients were entered, 127 to early randomization (3-5 days) and 28 to late randomization (10-14 days); however, only 112 patients with early and 14 with late randomization had completed data forms for the primary study goals. An additional 22 patients from the early entry groups were rerandomized late. Early randomization data: the AIII score was higher (p < 0.05) in the head trauma tracheostomy (65 ± 4) than in the nontracheostomy group (51 ± 4) and in the nontrauma tracheostomy (92 ± 6) than in the nontracheostomy group (68 ± 7), but was equivalent in the non-head trauma group. Glasgow Coma Scale score, Emergency Room Trauma Score, Injury Severity Score, Acute Injury Score, positive end-expiratory pressure, and peak inspiratory pressure were not significantly different in any of the groups. There were no significant differences in ICU LOS, frequency of pneumonia, or death in any of the groups after either early or late tracheostomy compared with continued endotracheal intubation. Only 83 patients underwent postextubation laryngoscopy. There were no significant differences between the groups; however, there were trends to more vocal cord ulceration and subglottic inflammation in the continued intubation group. No patient was seen in this study with late vocal cord or laryngeal stenosis; there were no tracheal-innominate artery fistulae. Seven of the patients with abnormal findings at extubation had normal 3- to 5-month postextubation laryngoscopy. Conclusion: Physician bias limited patient entry into the study. Although there were higher AIII scores in the head trauma early tracheostomy patients, there were no differences in the primary end points of ICU LOS, pneumonia, or death in any of the groups studied. Long-term endoscopic follow-up was poor, but no known late tracheal stenosis was seen.