Early prediction of hospital outcomes in patients tracheostomized for complex mechanical ventilation weaning

Background: Tracheostomy is often performed in the intensive care unit (ICU) when mechanical ventilation (MV) weaning is prolonged to facilitate daily care. Tracheostomized patients require important healthcare resources and have poor long-term prognosis after the ICU. However, data lacks regarding prediction of outcomes at hospital discharge. We looked for patients' characteristics, ventilation parameters, sedation and analgesia use (pre-tracheostomy) that are associated with favorable and poor outcomes (post-tracheostomy) using univariate and multivariate logistic regressions.

Results: Eighty tracheostomized patients were included (28.8% women, 60 [52-71] years). Twenty-three (28.8%) patients were intubated for neurological reasons. Time from intubation to tracheostomy was 14.7 [10-20] days. Thirty patients (37.5%) had poor outcome (19 patients deceased and 11 still tracheostomized at hospital discharge). All patients discharged with tracheostomy (n = 11) were initially intubated for a neurological reason. In univariate logistic regressions, older age and higher body-mass index (BMI) were associated with poor outcome (OR 1.18 [1.07-1.32] and 1.04 [1.01-1.08], p < 0.001 and p = 0.025). No MV parameters were associated with poor outcome. In the multiple logistic regression model higher BMI and older age were also associated with poor outcome (OR 1.21 [1.09-1.36] and 1.04 [1.00-1.09], p < 0.001 and p = 0.046).

Conclusions: Hospital mortality of patients tracheostomized because of complex MV weaning was high. Patients intubated for neurological reasons were frequently discharged from the acute care hospital with tracheostomy in place. Both in univariate and multivariate logistic regressions, only BMI and older age were associated with poor outcome after tracheostomy for patients undergoing prolonged MV weaning.