Postextubation Noninvasive Respiratory Support in Patients with Acute Brain Injury: Not a Panacea, but Don’t Throw the Baby Out with the Bathwater!
Romain Pirracchio and Alejandro A. Rabinstein
Am J Respir Crit Care Med 2023; 208: 225-226
Extubation failure is a common problem in patients with acute brain injury, with rates ranging from 15% to 25% across studies (1–3). Older age, longer duration of mechanical ventilation, and impaired airway protective reflexes are the factors more consistently associated with increased risk of failed extubation in these patients (2). High-flow nasal oxygen alternating with noninvasive positive-pressure ventilation (NIPPV) may be beneficial for general ICU patients deemed to be at high risk of extubation failure or who have early signs of failure after extubation (4). Whether that benefit extends to patients with acute brain injury remains unclear (5).
In this issue of the Journal, Taran and colleagues (pp. 270–279) report the results of a secondary analysis of the data from the “Extubation strategies in Neuro-Intensive care unit patients, and associations with Outcomes” study—or the “ENIO” study (1)—in which they evaluated the association between the use of noninvasive respiratory support (NIPPV or high-flow nasal cannula [HFNC]) after extubation and the risk of reintubation in patients with acute brain injury (6). On the basis of a dataset of 1,115 patients, the authors report that, compared with standard oxygen therapy, neither prophylactic NIPPV nor prophylactic HFNC was associated with a reduction in the risk of reintubation.
The data used in this study come from a large multinational observational cohort of brain-injured patients undergoing liberation from mechanical ventilation. With 62 institutions from 19 different countries contributing data, the conclusions drawn from this analysis are certainly generalizable to a large proportion of neurocritical patients worldwide. The internal validity of the results is strengthened by the consistent results obtained in a series of sensitivity analyses, including Bayesian statistics and analyses using causal inference estimators. To further account for uncertainty, Bayesian analyses even included several priors; for example, a skeptical prior, where the odds ratio (OR) distribution was modeled as a normal distribution centered at an OR of 1, reflecting high uncertainty about which strategy will lead to less reintubation; and a data-driven prior, where the OR distribution was modeled as a normal distribution centered at the point estimated identified on the basis of a systematic review of the literature. Consistent results were also found with several secondary outcomes, including ICU length of stay, mortality in ICU, and mortality in hospital.
In a broader ICU population, previous studies have reported the differential impact of noninvasive respiratory support when used preventatively or as rescue after extubation (7). The primary goal of this study was to study prophylactic practices. However, rescue uses of NIPPV or HFNC were analyzed separately, and they were not associated with reduced reintubation rate compared with conventional oxygen therapy. Finally, HFNC and NIPPV were analyzed separately, and the results with each technique were similar.
Albeit large and multinational, ENIO is an observational study, exposing the results to a risk of bias caused by confounding by indication and unmeasured confounding. Residual confounding by indication from both measured and unmeasured confounding variables may have affected the results, given the degree of baseline imbalances. For instance, the decision to attempt extubation is likely affected by the clinician perception regarding the risk of reintubation, as is the decision to initiate a noninvasive respiratory support after extubation. This is well exemplified by a longer time to first extubation attempt in patients receiving noninvasive respiratory support who also had greater comorbidities and higher body mass index than patients who were treated only with conventional oxygen therapy.
The lack of granular information on physiological variables and clinical course after extubation should be taken into consideration when interpreting the results of this study. Although noninvasive ventilatory support was not associated with less reintubation, on average, in the case of heterogeneous response to treatment, it is quite plausible that some patients may have benefited from NIPPV or HFNC. Even if preextubation variables did not allow the identification of a subgroup of patients having such benefit, it is possible that postextubation assessments of the response to prophylactic (or even rescue) NIPPV and/or HFNC may discriminate between patients who can be helped by the noninvasive support and those who cannot. Also, some meaningful specific subgroups defined on the basis of preextubation variables were too small to look for differential response to treatment. This was the case, for instance, for patients with chronic lung disease or hypercarbia. Finally, because of the observational nature of this study, neither preventive nor rescue use of NIPPV and HFNC was standardized; a standardized approach, perhaps combining both modalities as proposed by Thille and colleagues (4), might still be useful.
Therefore, this study indicates that nonstandardized postextubation use of NIPPV or HFNC at the discretion of the treating clinician is not effective in reducing the risk of reintubation in patients with acute brain injury. This does not mean that postextubation NIPPV and/or HFNC use should not be further considered or, particularly, investigated. In fact, we agree with the authors that there is a need for a carefully conducted, prospective, controlled trial evaluating a protocolized implementation of postextubation prophylactic noninvasive ventilatory support in patients with acute brain injury, including an analysis of heterogeneity in the response to treatment. The risk of extubation failure in these patients is substantial, and our current protocols are not optimal. More high-quality research is needed to improve our care for these patients.
