Patrick Brass, Martin Hellmich, Angelika Ladra, Jürgen Ladra, Anna Wrzosek

Tracheostomy formation is one of the most commonly performed surgical procedures in critically ill intensive care participants requiring long-term mechanical ventilation. Both surgical tracheostomies (STs) and percutaneous tracheostomies (PTs) are used in current surgical practice; but until now, the optimal method of performing tracheostomies in critically ill participants remains unclear. We evaluated the effectiveness and safety of percutaneous techniques compared to surgical techniques commonly used for elective tracheostomy in critically ill participants (adults and children) to assess whether there was a difference in complication rates between the procedures. We also assessed whether the effect varied between different groups of participants or settings (intensive care unit (ICU), operating room), different levels of operator experience, different percutaneous techniques, or whether the percutaneous techniques were carried out with or without bronchoscopic guidance. We searched the following electronic databases: CENTRAL, MEDLINE, EMBASE, and CINAHL to 28 May 2015. We also searched reference lists of articles, 'grey literature', and dissertations. We handsearched intensive care and anaesthesia journals, abstracts, and proceedings of scientific meetings. We attempted to identify unpublished or ongoing studies by contacting manufacturers and experts in the field, and searching in trial registers. We included randomized and quasi-randomized controlled trials (quasi-RCTs) comparing percutaneous techniques (experimental intervention) with surgical techniques (control intervention) used for elective tracheostomy in critically ill participants (adults and children). Three authors independently checked eligibility and extracted data on methodological quality, participant characteristics, intervention details, settings, and outcomes of interest using a standardized form. We then entered data into Review Manager 5, with a double-entry procedure. Of 785 identified citations, 20 trials from 1990 to 2011 enrolling 1652 participants fulfilled the inclusion criteria. We judged most of the trials to be at low or unclear risk of bias across the six domains, and we judged four studies to have elements of high risk of bias; we did not classify any studies at overall low risk of bias. The quality of evidence was low for five of the seven outcomes (very low N = 1, moderate N = 1) and there was heterogeneity among the studies. There was a variety of adult participants and the procedures were performed by a wide range of differently experienced operators in different situations.There was no evidence of a difference in the rate of the primary outcomes: mortality directly related to the procedure (Peto odds ratio (POR) 0.52, 95% confidence interval (CI) 0.10 to 2.60, I² = 44%, P = 0.42, 4 studies, 257 participants, low quality evidence); and serious, life-threatening adverse events - intraoperatively: risk ratio (RR) 0.93, 95% CI 0.57 to 1.53, I² = 27%, P = 0.78, 12 studies, 1211 participants, low quality evidence,and direct postoperatively: RR 0.72, 95% CI 0.41 to 1.25, I² = 24%, P = 0.24, 10 studies, 984 participants, low quality evidence.PTs significantly reduce the rate of the secondary outcome, wound infection/stomatitis by 76% (RR 0.24, 95% CI 0.15 to 0.37, I² = 0%, P < 0.00001, 12 studies, 936 participants, moderate quality evidence) and the rate of unfavourable scarring by 75% (RR 0.25, 95% CI 0.07 to 0.91, I² = 86%, P = 0.04, 6 studies, 789 participants, low quality evidence). There was no evidence of a difference in the rate of the secondary outcomes, major bleeding (RR 0.70, 95% CI 0.45 to 1.09, I² = 47%, P = 0.12, 10 studies, 984 participants, very low quality evidence) and tracheostomy tube occlusion/obstruction, accidental decannulation, difficult tube change (RR 1.36, 95% CI 0.65 to 2.82, I² = 22%, P = 0.42, 6 studies, 538 participants, low quality evidence). When compared to STs, PTs significantly reduce the rate of wound infection/stomatitis (moderate quality evidence) and the rate of unfavourable scarring (low quality evidence due to imprecision and heterogeneity). In terms of mortality and the rate of serious adverse events, there was low quality evidence that non-significant positive effects exist for PTs. In terms of the rate of major bleeding, there was very low quality evidence that non-significant positive effects exist for PTs.However, because several groups of participants were excluded from the included studies, the number of participants in the included studies was limited, long-term outcomes were not evaluated, and data on participant-relevant outcomes were either sparse or not available for each study, the results of this meta-analysis are limited and cannot be applied to all critically ill adults.