Does delaying discharge from intensive care until after tracheostomy removal affect 30-day mortality? Propensity score matched cohort study

Objective To investigate the short-term mortality effect of discharge from an intensive care unit (ICU) with a tracheostomy in place in comparison to delaying discharge until after tracheostomy removal. Design A propensity score matched cohort study using data from the TracMan study. Setting Seventy-two UK ICUs taking part in the TracMan study, a randomised controlled trial comparing early tracheostomy (within 4 days of critical care admission) with deferred tracheostomy (after 10 days if still indicated). Participants 622 patients who underwent a tracheostomy while in the TracMan study between November 2004 and November 2008. 144 patients left ICU with a tracheostomy. 999 days of observation from 294 patients were included in the control pool. Interventions We matched patients discharged with a tracheostomy in place 1:1 with patients who remained in an ICU until either their tracheostomy was removed or they died with the tracheostomy in place. Propensity models were developed according to discharge destination, accounting for likely confounding factors. Primary outcome measure The primary outcome was 30-day mortality from the matching day. For the ‘discharged with a tracheostomy’ group, this was death within 30 days after the discharge day. For the ‘remained in ICU’ group, this was death within 30 days after the matched day. Results 22 (15.3%) patients who left ICU with a tracheostomy died within 30 days compared with 26 (18.1%) who remained in ICU (relative risk 0.98, 95% CI 0.43 to 2.23). Conclusion Keeping patients on an ICU to provide tracheostomy care was not found to affect mortality. Tracheostomy presence may indicate a higher risk of mortality due to underlying diseases and conditions rather than posing a risk in itself. The TracMan trial was registered on the ISRCTN database ( ISRCTN28588190 ).