Tracheostomy risk factors and outcomes after severe traumatic brain injury

Abstract

Objective: To determine risk factors associated with tracheostomy placement after severe traumatic brain injury (TBI) and subsequent outcomes among those who did and did not receive a tracheostomy.

Methods: This retrospective cohort study compared adult trauma patients with severe TBI (n = 583) who did and did not receive tracheostomy. A multivariable logistic regression model assessed the associations between age, sex, race, insurance status, admission GCS, AIS (Head, Face, Chest) and tracheostomy placement. Ordinal logistic regression models assessed tracheostomy’s influence on ventilator days and ICU LOS. To limit immortal time bias, Cox proportional hazards models assessed mortality at 1, 3 and 12-months.

Results: In this multivariable model, younger age and private insurance were associated with increased probability of tracheostomy. AIS, ISS, GCS, race and sex were not risk factors for tracheostomy placement. Age showed a non-linear relationship with tracheostomy placement; likelihood peaked in the fourth decade and declined with age. Compared to uninsured patients, privately insured patients had an increased probability of receiving a tracheostomy (OR = 1.89 [95% CI = 1.09–3.23]). Mortality was higher in those without tracheostomy placement (HR = 4.92 [95% CI = 3.49–6.93]). Abbreviated injury scale-Head was an independent factor for time to death (HR = 2.53 [95% CI = 2.00–3.19]), but age, gender and insurance were not.

Conclusions: Age and insurance status are independently associated with tracheostomy placement, but not with mortality after severe TBI. Tracheostomy placement is associated with increased survival after severe TBI.

Keywords:

• Traumatic brain injury
• outcome
• trauma
• tracheostomy
• insurance

Acknowledgements

This was an oral presentation at the American College of Surgeons’ 99th Clinical Congress, Surgical Forum, Neuro-surgery, 7 October 2013, Washington, DC.

Declaration of interest

The authors report no conflicts of interest. The authors alone are responsible for the content and writing of this paper. The project utilized REDCap, which is supported by Clinical and Translational Science Award (CTSA) Grant (UL1 TR000448 from NCATS/NIH). Author MBP was supported by the Vanderbilt Faculty Scholars Research Program, AHRQ Health Services (5T32HS013833-08), NIH R01HL111111, and the Eastern Association for the Surgery of Trauma Foundation Research Scholarship. Author PPP is supported by NIH grants (R01AG027472, R01HL111111, R01AG035117).