Web site designed and maintained by Chris Cairns  © SICS EBM Group 2004                                  

Early tracheostomy 4

Citation: Rodriguez JL, Steinberg SM, Luchetti FA, Gibbons KJ, Taheri PA, Flint LM. Early tracheostomy for primary airway management in the surgical critical care setting. Surgery. 1990; 108: 655-659.

Lead author: Jorge L Rodriguez, Dept Surgery, University of Michigan.

Three-part Clinical Question: Does early compared to delayed tracheostomy in multiply injured patients alter the outcomes of death, pneumonia, ventilator days, intensive care stay and hospital stay.

Search Terms: This is one of the five studies that contributed to the meta-analysis of Griffiths et al.

The Study: Non-blinded, pseudo-randomised controlled trial, apparently without intention-to-treat.

The Study Patients: These were multiply injured patients who required ventilation. Recruitment took place over a 12-month period. Patients were assessed on day two and those that were not being actively weaned were randomised into either early or late tracheostomy groups. Entry into the study appears to be subjective. Numerous demographic and comparative data were presented. The two groups were comparable on entry to the study. The technique of obtaining a tracheostomy was not given. Ventilation and weaning protocols were not discussed. Sedation was not discussed. Pneumonia was defined and this definition appears to be appropriate. The diagnosis of pneumonia does not appear to be independent of the study, and may therefore be prone to detection bias. Other outcomes are perhaps more robust.
 

Control group (N = 55; 55 analysed): 55 patients underwent tracheostomy on or after day 8: 45 between days 8-15 and 10 at days 15-21.

Experimental group (N = 51; 51 analysed): 51 patients underwent tracheostomy within 7 days: 12 within 2 days and 39 between days 3-7.

The Evidence: (Control is the late tracheostomy group)

Comments and EBM questions:

1)      Do the methods allow accurate testing of the hypothesis?

Not really. Patients were excluded from analysis if, on day 1, they died, had been weaned successfully or were actively being weaned. Of 120 live patients who were excluded from the trial, 65 were weaned from day 3 onward- so perhaps should have been entered into the trial. This problem of how to predict who needs a tracheostomy is of course not restricted to this study and may be a source of selection bias.

However, it is unclear from the text if the patients were randomised before or after exclusion of those undergoing active weaning. Patients who were weaned without tracheostomy appear to have been excluded. These patients should have been included in an intention-to-treat analysis. Patients allocated to the late tracheostomy group would logically have a greater chance of being weaned without tracheostomy, yet appear to have been excluded from the study. Conversely, some patients in the early tracheostomy group may have had a greater chance of successful weaning, despite their tracheostomy, but were included in the study.

Alternatively, all patients randomised to the late tracheostomy group may have proceeded to it: this seems unlikely. In all, the results are difficult to interpret.  There is a potential source of selection bias since 65 patients have been excluded: more than in either arm of the study.

2)  Do the statistical tests correctly test the results to allow differentiation of statistically significant results?

 No. There was pseudo-randomisation based on date of admission. There does not appear to have been a power calculation performed to account for the patient numbers. Analysis was probably not on an intention to treat basis. Student’s t-test is probably inappropriate for time-related outcomes. It is unlikely that ventilator days and lengths of stay would be normally distributed in the two groups.

3)      Are conclusions valid in light of the results?

Probably not. Given that 65 patients were excluded from the study, the outcomes could have been the result of selection bias. They concluded that early tracheostomy reduced the incidence of pneumonia; number of days ventilated; in the ICU and in hospital. The complications of tracheostomy were recorded and these do not appear to differ between the early and late groups. A subgroup analysis suggested that the incidence of pneumonia in those patients who underwent tracheostomy at day 2 or earlier had the lowest incidence of pneumonia. Comparison between intermediate (day 3-7) and late timed tracheostomy did not reveal a difference. On the results presented, the conclusions may only be appropriate if we bear in mind that a large amount of information on the management of these patients was not presented and there are potential sources of bias as discussed.

4) Did results get omitted, and why? No.

5) Did they suggest areas of further research? No. The short discussion section re-iterated the conclusions.

6) Did they make any recommendations based on the results and were they appropriate?

Yes and No. They recommended that early tracheostomy is of benefit for those patients who require more than 7 days of ventilation; the seven day suggestion cannot really be supported on the data presented. If anything, their data supports very early tracheostomy.

7) Is the study relevant to my clinical practice?

Yes. The study suggests a shorter ventilation time in multiply injured trauma patients who proceed to early tracheostomy.

8) What level of evidence does this study represent? 1^- (RCT with high risk of bias)

9) What grade of recommendation can I make on this result alone? C

10) What grade of recommendation can I make when this study is considered along with other available evidence?../jargon.htm#Grade of recommendation

This paper and that of Dunham & LaMonica (1984) looked at early tracheostomy in trauma patients although the results do not agree. There are mixed results from other patient groups and these have been subject to meta-analysis.

11) Should I change my practice because of these results?

Perhaps. This paper tends to favour early tracheostomy in patients with severe multiple injury, with the caveats discussed. The potential utility of prolonged trans-laryngeal intubation was not addressed.

12) Should I audit my current practice because of these results?

Yes. It may be reasonable to audit the timing and outcome of patients with severe injury who proceed to tracheostomy.

Appraised by: Richard Price; SpR, Dept. Anaesthetics, Gartnavel Hospital, Dumbarton Road, Glasgow, G12 0YN.  Monday, 27 March 2006
Email: rjp@doctors.org.uk

Kill or Update By: March 2011

Printer friendly view

©SICS EBM 2006