STATEMENT OF THE PROBLEM The ideal time for performing a tracheostomy has not been clearly established. Periods ranging from 3 days to 3 weeks have been suggested in the literature. With current operative methods, it has been established that tracheostomy can be performed with a low rate of complications. In a review of 281 tracheostomies, as well as another 2,862 cases in the literature, Zeitouni and Kost1 reported 0% mortality in their series and 0.3% mortality in the other series since 1973. The risks of prolonged endotracheal intubation—such as patient discomfort, necessitating increased sedation; sinusitis; inadvertent extubation; and laryngeal injury—have become increasingly apparent. Selection of patients who might benefit from conversion of a translaryngeal tube to a tracheostomy tube is a complex medical decision. Furthermore, different subgroups may benefit from tracheostomy at different times in their hospital course. Management of patients with a single organ failure (head injury or respiratory failure) may differ from that of the multiple injury trauma patient. With the lack of clear guidelines for selecting patients for tracheostomy, considerable variability exists in the timing of the procedure, with local practice preferences guiding care, rather than patient considerations. We initiated our review by converting the need for information about optimal timing of tracheostomy into several answerable questions: Does performance of an “early” tracheostomy provide a survival benefit for the recipients? What patient populations benefit from an “early” tracheostomy? Does “early” tracheostomy reduce the number of days on mechanical ventilation and intensive care unit length of stay (ICU LOS)? Does “early” tracheostomy influence the rate of ventilator-associated pneumonia? PROCESS Identification of References A computerized search was undertaken using Medline with citations published between the years of 1966 and 2004. By using the search words “tracheostomy” and “timing” and by limiting the search to citations dealing with human subjects and published in the English language, we identified 87 articles. From this initial search, case reports, review articles, editorials, letters to the editor, and pediatric series were excluded before formal review. Additional references, selected by the individual subcommittee members, were then included to compile the master reference list of 24 citations. Articles were distributed among the subcommittee members for formal review. A data sheet was completed for each article that was reviewed, which summarized the purpose of the study, hypothesis, methods, main results, and conclusions. The reviewers classified each reference by the methodology established by the Agency for Health Care Policy and Research of the U.S. Department of Health and Human Services. Quality of the References Class I: Prospective randomized controlled trials (PRCT) (7 references). Class II: Clinical studies in which the data were collected prospectively, and retrospective analyses that were based on clearly reliable data. Types of studies classified as such included observational studies, cohort studies, prevalence studies, and case-control studies (5 references). Class III: Studies based on retrospectively collected data. Evidence used in this class included clinical series and database or registry review (12 references). An evidentiary table was constructed using the remaining 24 references (Table 1). Additionally, a meta-analysis, including the seven class I articles, was done by the Vice Chairman of the Committee, Michael Dunham. Recommendations were based on studies that were included in the evidentiary tables.TABLE 1: Evidentiary TableTABLE 1: Evidentiary Table (continued)RECOMMENDATIONS Level I There is no mortality difference between patients receiving early tracheostomy (3–7 days) and late tracheostomy or extended endotracheal intubation. Level II Early tracheostomy decreases the total days of mechanical ventilation and ICU LOS in patients with head injuries. Therefore, it is recommended that patients with a severe head injury receive an early tracheostomy. Level III Early tracheostomy may decrease the total days of mechanical ventilation and ICU LOS in trauma patients without head injuries. Early tracheostomy may decrease the rate of pneumonia in trauma patients. Therefore, it is recommended that early tracheostomy should be considered in all trauma patients who are anticipated to require mechanical ventilation for >7 days, such as those with neurologic impairment or prolonged respiratory failure. SCIENTIFIC FOUNDATIONS The optimal timing for tracheostomy has been controversial. Laryngeal complications were common before the widespread use of endotracheal tubes with low-pressure, high-volume cuffs. With modern endotracheal tubes, it has been established that patients can be safely intubated for at least 14 days.2 In a prospective, randomized, controlled trial of 74 trauma patients who received a tracheostomy at either 3 days or 14 days, there was no significant difference in the incidence of major laryngotracheal pathology between the early and late tracheostomy groups. Significantly, because the perfusion pressure of the tracheal mucosal capillaries was ∼20–30 mm Hg, the authors were careful to limit cuff pressures to 25 mm Hg whenever possible. There have been many advantages attributed to converting a translaryngeal endotracheal tube to a tracheostomy tube in the critically ill or injured patient, although not all of them are supported in the literature. Among them are improved patient comfort; facilitation of nursing care, such as airway suction and oral hygiene; and psychologic benefit. These statements have never been tested with a large-scale prospective, randomized study. Astrachan et al.3 surveyed 60 critical care nurses, the majority of whom felt that tracheostomy simplified airway care. They also thought that patients were more comfortable and, therefore, required less sedation and restraints. Unfortunately, there are no data available from either patients or their families on the preferred method of airway support. A common perception among critical care providers is that early tracheostomy may reduce the necessity for mechanical ventilation. One possible mechanism is that mobilization of the patient might allow improved pulmonary toilet and functional residual capacity as well as avoidance of oversedation. Decreased airflow resistance and reduced dead space after tracheostomy may also contribute to accelerate weaning. In a study of 20 patients, Davis et al.4 found decreased work of breathing per minute (8.9 ± 2.9 J/min vs. 6.6 ± 1.4 J/min, p < 0.04) and airway resistance (9.4 ± 4.1 cm H2O·L−1·s−1 vs. 6.3 ± 4.5 cm H2O·L−1·s−1, p < 0.07) after conversion of a translaryngeal tube to tracheostomy. A number of retrospective and a few prospective outcome studies have examined the effect of tracheostomy on weaning from mechanical ventilation. These studies have compared patients undergoing “early” tracheostomy to patients either continuing with translaryngeal intubation or undergoing “late” tracheostomy. There is a great deal of variability in the definition of “early” tracheostomy, varying from 2 to 10 days from the time of initial intubation. Furthermore, some of these studies have used quasirandomization methods (even/odd admission day and even/odd medical record number) or assigned patients to groups based on physician preference. Rodriguez et al. prospectively randomized (by day of admission) 106 multiple trauma patients to receive either an early tracheostomy (within 7 days) or late tracheostomy (greater than 7 days). They found a reduction in duration of mechanical ventilation favoring the early tracheostomy group (12 days vs. 32 days, p < 0.05). Not surprisingly, ICU LOS was also reduced (16 days vs. 37 days, p < 0.05), as was overall hospital LOS (34 days vs. 51 days, p < 0.05). Their conclusion was that early tracheostomy shortens days on the ventilator as well as ICU and hospital LOS. They advised considering tracheostomy for patients in the ICU who are at risk for >7 days of intubation.5 In a cohort study of 136 trauma ICU patients who were entered prospectively into an ICU database, Arabi et al.6 found that the duration of mechanical ventilation was significantly shorter (9.6 days vs. 18.7 days, p < 0.0001) when tracheostomy was performed within 7 days, as was ICU LOS (10.9 days vs. 21 days, p < 0.0001). These results were supported by the review conducted by Van Boerum et al. on 94 trauma patients who required tracheostomies. In this study, tracheostomy within 7 days of intubation resulted in a significant reduction in ventilator days (9.6 days vs. 18.7 days, p < 0.0001) and a reduction in ICU LOS (10.9 days vs. 21 days).7 Lesnik et al. retrospectively reviewed 101 blunt trauma patients who underwent tracheostomy. Patients who underwent tracheostomy within 4 days of injury had significantly fewer days of mechanical ventilation compared with those who underwent tracheostomy at 5 or more days (6.0 vs. 20.6, p < 0.001). The selection criteria for performing early tracheostomy were not given.8 In a prospective, randomized trial of 62 patients with isolated head injury, Bouderka et al. randomized patients on the fifth day to receive either a tracheostomy or a prolonged translaryngeal intubation. The total ventilation days were significantly less in the early tracheostomy group (14.5 days vs. 17.5 days, p < 0.02).9 Results in the Medical Intensive Care Unit (MICU) population mirror those in the trauma patient population. In a prospective, randomized, controlled trial of 120 MICU patients who were projected to require mechanical ventilation for more than 14 days, patients received either early percutaneous tracheostomy within 48 hours of intubation or late tracheostomy at days 14–16. Early tracheostomy was associated with reduced duration of mechanical ventilation (7.6 days vs. 17.4 days, p < 0.001) and decreased ICU LOS (4.8 days vs. 16.2 days, p < 0.001).10 Brook et al. added financial information to the clinical data in their retrospective cohort study of 90 MICU patients who underwent either early (< 10 days, mean 5.9 days) or late (> 10 days, mean 16.7 days) tracheostomy. Both duration of mechanical ventilation (28.3 days vs. 34.4 days, p = 0.005) and ICU LOS (15.6 days vs. 29.3 days, p < 0.001) were reduced, which was reflected in a lower cost of hospitalization ($86,189 vs. $124,649, p = 0.001) for the patients who received tracheostomy within 10 days.11 Therefore, it seems that tracheostomy that is performed earlier rather than later will reduce days of mechanical ventilation and, therefore, ICU LOS and overall cost. Whether early tracheostomy affects the frequency of pneumonia is less well elucidated. The studies that report frequency of pneumonia use the Center for Disease Control criteria, which are notoriously inaccurate in intubated ICU patients. In the PRCT of multiple trauma patients by Rodriguez,7 early tracheostomy resulted in a statistically significant lower incidence of pneumonia (78% vs. 96%, p < 0.05). In the study of PRCT of 120 MICU patients by Rumbak et al.,10 early tracheostomy was again associated with a lower rate of pneumonia (5% vs. 25%, p < 0.05). This conclusion was supported by retrospective reviews by Lesnik et al.,8 who also demonstrated a reduced incidence of pneumonia after early tracheostomy (19% vs. 59%, p < 0.001), and Kluger et al.12 (14% Early, 23% Intermediate, and 43% Late, p = 0.0034). Although Sugerman et al.13 failed to find a difference in mortality in their PRCT of trauma and ICU patients, a difference in Apache scores between the tracheostomy and nontracheostomy groups may account for this finding. In a PRCT of 74 trauma patients who received either an early or late tracheostomy, Dunham and LaMonica also failed to find a difference in pulmonary septic complications. In their study, however, all types of respiratory sepsis (tracheitis, pneumonia, lung abscess, and peristomal infection) were combined for analysis, which may explain their differing results.2 Unfortunately, early tracheostomy has not been found to provide a survival benefit for its recipients. Of six prospective, randomized, controlled trials,5,10,11,13–15 only one has demonstrated a reduction of mortality. In a study of 120 MICU patients who were projected to need ventilation beyond 14 days, patients who were randomized to early percutaneous tracheostomy within 48 hours had a lower mortality rate than those receiving delayed tracheostomy at days 14 –16 (31.7% vs. 61.7%, p < 0.05).10 Therefore, although early tracheostomy may reduce ventilator days and pneumonia rates, an effect on mortality remains to be seen. FUTURE INVESTIGATIONS Future investigations should be performed in a prospective, randomized manner with a sufficient number of patients to enable clinicians to draw valid, concrete conclusions as to the optimal methods of evaluating these patients. Prospective randomization will decrease the baseline differences between the groups and allow more concrete conclusions to be drawn. Quasirandomization methods (e.g., hospital record number and even-odd days) should be avoided. Consensus as to what constitutes “early” versus “late” tracheostomy should be established, so that various studies can be compared. Because blinding is unrealistic, systematic weaning protocols should be used to reduce the effect of different approaches toward weaning. It remains unclear as to which patients will need prolonged ventilation. Multi-institutional studies of sufficient sample sizes of specific patient populations, such as the head injured, should identify objective criteria to aid the individual physician in determining which subgroups of patients are likely to require prolonged ventilation and might subsequently benefit from an early tracheostomy. Given the current condition of shrinking resources, future studies should also routinely include cost-effectiveness analysis.

	Year	Citations

Abstract

References