Many studies have investigated the benefits of performing early tracheostomy (ET), but these studies contain a considerable variation in the definition of time to ET and in design quality.
We performed a random effects meta-analysis to assess whether ET in intensive care unit (ICU) patients is superior to late tracheostomy. We performed subgroup analyses for trials defining ET as within five days and for ET with only high quality trials. Medline, EMBASE, and Cochrane Controlled Trials Register from 1966 through December 2008 were searched for randomized trials. Two reviewers independently rated study quality using the Jadad index and extracted outcome data.
Seven studies were identified with 641 patients (311 patients in the ET arm and 310 patients in the LT arm). ET did not significantly affect mortality (Odds ratio (OR) 0.79, 95% confidence interval (CI) 0.3 to 1.45), risk of pneumonia (OR 0.67, CI 0.36 to 1.23), or duration of mechanical ventilation (MV) (−6.46 days, CI −12.91 to −0) and these outcomes were not affected by ET performed specifically within five days. Including only trials of acceptable design, ET did not affect risk of pneumonia or duration of MV. Within the analysis of quality studies, ET significantly reduced mortality (OR 0.49, CI 0.25 to 0.97), but this included only three trials and was heavily influenced by one study. ICU length of stay (LOS) was significantly reduced (−10.96 days, CI −17.42 to −4.38), although this analysis included only three trials with two using quasi-randomization.
ET and specifically ET within five days do not affect significant patient outcomes. Analysis limited to high quality trials does not reveal benefit. The finding of reduced mortality within high quality trials should be interpreted with caution. ET may shorten ICU LOS but data is limited and biased.
Eliminating heterogeneity in time to ET and study quality within the literature did not reveal significant benefit from ET. Further high quality studies are required to elucidate a role for ET.
Michael Perkins, No Financial Disclosure Information; No Product/Research Disclosure Information