Is tracheostomy associated with better outcomes for patients requiring long-term mechanical ventilation?

Objective: To evaluate the effect of tracheostomy on intensive care unit (ICU) and in-hospital mortality for patients requiring prolonged (>3 days) mechanical ventilation (MY). Design, Setting, and Patients. We retrospectively reviewed the charts of all consecutive patients admitted to our 18-bed tertiary care ICU over 3 yrs (2002-2004) and who received prolonged MY. Outcomes of tracheostomized and nontracheostomized patients were evaluated using univariable and multivariable logistic-regression analyses and by constructing a case-control cohort using a propensity score for performing tracheostomy. MY duration for controls was at least equal to the time from MY onset to tracheostomy for the matched case. Measurements and Main Results: Of the 506 patients requiring prolonged MY, 166 were tracheostomized after a median of 12 days of MY. Nontracheostomized patients had higher ICU (42% vs. 33%, p =.06) and in-hospital (48% vs. 37%, p =.03) mortality rates and shorter MY durations and ICU lengths of stay. Performing a tracheastomy (odds ratio, 0.58; 95% Cl, 0.37-0.90) was independently associated with a lower probability of ICU death, even after adjusting for other important prognostic factors. No significant differences were detected between the 120 cases and their matched controls regarding ICU admission and day-3 clinical characteristics. After conditional logistic-regression analysis, tracheostomy was associated with lower risk of ICU (odds ratio, 0.47; 95% Cl, 0.24-0.89) and in-hospital (odds ratio, 0.48; 95% Cl, 0.25-0.90) death. Conclusions. Tracheostomy performed in our ICU for long-term MY patients was associated with lower ICU and in-hospital mortality rates, even after carefully controlling for ICU admission and day-3 clinical and physiologic differences between groups. Whether these results reflect that physicians were able to adequately select for tracheastomy patients who, despite having similar physiologic and demographic variables, had the highest probabilities of survival or that the procedure itself really affected the outcomes of these patients will remain speculative.