Ann Thorac Surg 2007;84:1991-1992. doi:10.1016/j.athoracsur.2007.08.066
© 2007 The Society of Thoracic Surgeons
Original Articles: Cardiovascular
Invited commentary
Luis D. Berrizbeitia, MD
Section of Thoracic Surgery, University Medical Center at Princeton, Suite F, 253 Witherspoon St, Princeton, NJ 08540
(Email: ldberriz{at}ctsx.net).
The proposition that a tracheostomy after a sternotomy increases the probability of deep sternal wound infection (DSWI) has intuitive appeal, but it is not uniformly supported by clinical data [1, 2]. Many surgeons follow a conservative approach and avoid doing a tracheostomy after a sternotomy hoping to decrease the probability of DSWI. Rahmanian and associates [3] challenge this approach by carrying a multivariate analysis of the determinants of DSWI and proposing that respiratory failure (RF), not tracheostomy, is the main predictor of DSWI.
The validity of this analysis relies on whether tracheostomy is a random independent factor with an equal chance of occurring in all patients in the data set. This means that the variable tracheostomy follows an independent stochastic distribution. This may be difficult to prove as there is a deterministic relationship between RF and tracheostomy (ie, no patients without RF undergo a tracheostomy, whereas all patients with RF should undergo a tracheostomy at a fixed interval after intubation). This is a classic example of confounding variables by which the relationship between two variables (ie, DSWI and tracheostomy) is affected by a third (RF) [4]. The authors address this difficulty by showing that there are no univariate predictors of tracheostomy in the patients who had respiratory failure (see Table 2 in the accompanying article). The fact that the variables age and ejection fraction appear as significant predictors by virtue of p values less than 0.05 and that other interesting variables have p values around 0.10 is not mentioned. The question needs to be asked whether incorporating these variables in a multivariate model would have uncovered other factors that explain the use of tracheostomy in the study population and its role in the development of DSWI.
If the assumption is accepted that a tracheostomy is a random independent factor, the observation needs to be made that the final multivariate analysis is based on DSWI developing in only 38 patients. Whereas this speaks highly of the clinical skills of the authors, the relatively low number of observations raises concerns about the power of the study, based on the relatively low number of patients that had the outcome variable of interest (ie, DSWI).
Regardless of the intricacies of the numerical analysis, there is a clear message for the clinician: RF is a very serious postoperative complication that increases the probability of DSWI and postoperative death. Also, RF should be managed aggressively and a tracheostomy should not be withheld once the need for prolonged ventilatory support has been established.
 |
References
|
|---|
- Force SD, Miller DL, Petersen R, et al. Incidence of deep sternal wound infections after tracheostomy in cardiac surgery patients Ann Thorac Surg 2005;80:618-621discussion 621–2.[Abstract/Free Full Text]
- Stamenkovic SA, Morgan IS, Pontefract DR, Campanella C. Is early tracheostomy safe in cardiac patients with median sternotomy incisions? Ann Thorac Surg 2000;69:1152-1154.[Abstract/Free Full Text]
- Rahmanian PB, Adams DH, Castillo JG, Chikwe J, Filsoufi F. Tracheostomy is not a risk factor for deep sternal wound infection after cardiac surgery Ann Thorac Surg 2007;84:1984-1992.[Abstract/Free Full Text]
- Pearl J. Causality: models, reasoning, and inferenceCambridge, UK: Cambridge University Press; 2000. pp. 173-199.
Related Article
-
Tracheostomy is Not a Risk Factor for Deep Sternal Wound Infection After Cardiac Surgery
- Parwis B. Rahmanian, David H. Adams, Javier G. Castillo, Joanna Chikwe, and Farzan Filsoufi
Ann. Thorac. Surg. 2007 84: 1984-1991.
[Abstract]
[Full Text]
[PDF]
Top
References