Long-Term Outcome Following Tracheostomy in Critical Care: A Systematic Review

The prevalence and impact of longer-term outcomes following percutaneous tracheostomy, particularly tracheal stenosis, are unclear. Previous meta-analyses addressing this problem have been confounded by the low prevalence of tracheal stenosis and a limited number of studies.

Embase, PubMed-Medline, and the Cochrane Central Register of Clinical Trials were searched to identify all prospective studies of tracheostomy insertion in the critically ill. To reflect contemporary practice, the search was limited to studies published from 2000 onward. We scrutinized the bibliographies of returned studies for additional articles. Meta-analyses were undertaken to estimate the pooled risk difference of tracheal stenosis, bleeding, and wound infection comparing different techniques.

We identified a total of 463 studies, 29 (5,473 patients) of which met the inclusion criteria. Nine were randomized controlled trials, six were nonrandomized comparative studies, and 14 were single-arm cohort studies. Risk of wound infection was greater for the surgical tracheostomy than for the Ciaglia multiple dilator technique, pooled risk difference 0.12 (95% CI, 0.02-0.23). We did not identify significant risk differences in other meta-analyses. Pooling across all studies according to the random-effects proportion meta-analysis suggests a higher prevalence of tracheal stenosis, wound infection, and major bleeding for surgical tracheostomies.

Considering comparative data, there was no significant difference in the prevalence of tracheal stenosis or major bleeding between percutaneous and surgical tracheostomy. In relation to wound infection, we have found a reduction associated with the original Ciaglia technique when compared with that with the surgical tracheostomy. Considering all published data reporting long-term outcomes pooled proportion meta-analysis indicates a trend toward a higher rate of tracheal stenosis and an increased risk of major bleeding and wound infection for surgical tracheostomies. This finding may be biased as a result of targeted patient selection, and further, high-quality long-term comparative data are needed to confirm these findings.