An audit of timing of tracheostomy formation in neurosurgical patients: how long should we wait?

Introduction

In neurosurgical patients requiring ventilation on the ICU, a tracheostomy is frequently formed to facilitate airway protection and weaning from the mechanical ventilator. However, the issue of when to form a tracheostomy remains contentious. In order to better inform our decision-making processes we audited practice within our own institution.

Methods

A retrospective study was conducted in which the ICU charts of all neurosurgical patients admitted to a tertiary referral ICU during the calendar year 2007 were reviewed. Patients who did not require mechanical ventilatory support or who died within 7 days of admission were excluded. Demographic data, diagnosis, duration of mechanical ventilation, ICU day of stay on which tracheostomy was formed, and ICU length of stay were recorded and the data analysed accordingly.

Results

A total of 106 patients were included, 65 male and 41 female. The mean age was 49 years. Sixty-three patients were able to be separated from the mechanical ventilator within 7 days of commencement of ventilation via a cuff ed oral endotracheal tube. Of the remaining 43 patients, 34 (79%) went on to undergo tracheostomy formation as determined by the attending intensivist. In this group the median time of tracheostomy formation was 13 days. The median time from tracheostomy formation to separation from the mechanical ventilator was 3 days. Of those patients who could not be extubated within 7 days of ventilatory support, nine (14%) were successfully separated from the mechanical ventilator without the need for a tracheostomy. There were no significant differences in age or diagnosis between the two groups.

Conclusions

Our data suggest that the failure of a neurosurgical patient to separate from the mechanical ventilator within 7 days is predictive of the eventual requirement for tracheostomy formation. In light of this we intend to expand our sample size over a 5-year period and subject the data to multivariate regression.