We reviewed ventilation settings, sedation–analgesia and outcomes of patients ventilated for more than 72 h and tracheostomized during the ICU stay, both for neurological and non-neurological reasons. For the global patient group, in univariate logistic regressions, only older age and higher BMI were associated with poor outcome, defined as in-hospital death or hospital discharge without decannulation. This remained true in the multivariate logistic regression analysis. For the subgroup of patients intubated for non-neurological reasons, the multivariate analysis led to similar conclusions. In this study, we also confirmed high ICU-admission severity scores, high hospital mortality and long ICU and hospital length of stay in tracheostomized patients [1].
Patients intubated and ventilated for all causes and tracheostomized for difficult weaning have high mortality [14, 23]. For example, in the population of patients ventilated for more than 10 days and tracheostomized following acute respiratory distress syndrome, high 28-day and 90-day mortality was reported (30.8% and 45.2%, respectively [12]). Our study population has the characteristics of a general ICU population, including both medical and surgical patients and patients intubated both for neurological and non-neurological reasons. We found relatively low ICU mortality for tracheostomized patients initially intubated for neurological reasons compared to other studies [24, 25] but a high hospital mortality in line with the literature for the global group of patients [1]. The high hospital mortality observed in our population was expected, considering the high severity scores at admission. Death after withdrawal of life-sustaining therapy concerned 63.2% of patients, suggesting frequent poor evolution after tracheostomy, underlining the difficulty of predicting global evolution at the time of tracheostomy.
In the literature, ICU and hospital length of stay differ in tracheostomized patients depending on the series of patients. Hospital length of stay in our population was higher compared to most available data. This could be related to differences in health care policies. Indeed, long-term weaning facilities are not available in Switzerland. In addition, most long-term care facilities do not manage mechanical ventilation in tracheostomized patients and home discharges with home ventilation on tracheostomy is unusual in Switzerland. Those factors could explain the prolonged length of stay in acute-settings hospital. Local practices regarding late or early withdrawal of life-sustaining treatment could also impact hospital length of stay and this can differ between countries, hospitals and even health-care practitioners [26]. Our high percentage of WLST combined with high length of hospital stay, when compared to other countries in Europe and in the world [27] underline the fact that WLST only takes place relatively late in our hospital in tracheostomized patients. This emphasizes the fact that predicting outcomes of tracheostomized patients takes time and requires, at least in Switzerland, a multi-disciplinary consensus, which is sometimes difficult to reach. Such a consensus can sometimes even be difficult to find within the team in charge of the patient because of different individual perceptions of the clinical situation. This is true not only at time of tracheostomy, but also after ICU discharge.
In addition to common ICU parameters to predict outcomes, our study specifically assessed the relationship between outcome and ventilation, sedation, opioid and NMBA use before tracheostomy. No ventilation parameters showed any association with poor outcome. To note, our data also showed good compliance with international ventilation guidelines [28]. Sedation, opioids and NMBA use before tracheostomy did not show any association with outcome. This remained true for tracheostomized patients who intubated for neurological and non-neurological reasons.
This study adds a new perspective on the prediction of unfavorable outcome in tracheostomized patients by demonstrating that ventilation data prior to the tracheostomy did not help predict outcome. Indeed, only higher BMI and older age were associated with poor outcome. This highlights yet the fact that decision to undergo tracheostomy can only be based on general clinical judgment and that more or less severe respiratory status and worse ventilation parameters cannot be used to select patients who could benefit from tracheostomy.
Separation attempts were performed in 73.4% of patients before tracheostomy, which is similar to numbers reported in the recent WIND study collective [1]. The number of separation attempts before tracheostomy was not associated with better or worse outcome. We initially hypothesized that the relationship between mortality and the number of SBT takes the form of a U-shaped curve, with worse outcomes in patients with no SBT (i.e., because of persistent organ failure) and in patients with many SBT (i.e., very prolonged weaning). Therefore, we looked in our data but found no pattern to corroborate this hypothesis. To note, this study did not address the subject of MV weaning strategies after tracheostomy.
As limitations for this study, we must mention that, in this retrospective study, mechanical ventilation data were collected at arbitrary time-points and do not always accurately represent 24-h data. However, as data were collected daily, the ventilation parameters reflect the whole duration of mechanical ventilation before tracheostomy. Second, regarding patients sub-groups, it can be argued that patients intubated because of cardiac arrest could have been classified as patients with neurological impairment. However, eight of them were tracheostomized because of difficult weaning and one because of difficult secretion management. Only one patient was tracheostomized because of persistent neurological impairment. Thirdly, no comparison with a control group without tracheostomy was performed. Even if this comparison would be interesting to assess the impact of tracheostomy on patients’ outcome, we could not do it. Tracheostomy is part of the protocolized management of prolonged weaning in our ICU in the absence of poor prognosis regarding recovery potential. Patients with prolonged weaning who are not tracheostomized have thus different characteristics compared to the tracheostomized patients and cannot be used as a control group. Fourthly, because of the monocentric nature of this study and very different practices regarding tracheostomy between different centers due to the lack of unifying guidelines for tracheostomy indications, our conclusions can probably not be generalized to all other ICUs. Health-care policies and organizational differences concerning the transfer from the ICU to a step-down unit (or other health-care facilities) can also limit the applicability of our results to other hospitals. Finally, in the absence of sample size calculation, our study could potentially have been underpowered to evaluate the association between some factors and outcome. However, tracheostomies for prolonged weaning in the ICU is relatively rare, and monocentric studies rarely have much larger collectives.