Benefit with preventive noninvasive ventilation in subgroups of patients at high-risk for reintubation: a post hoc analysis

The most important finding of the present study is that traditional definition of high risk for reintubation result in heterogeneous populations, but stratifying this population reveals that outcomes for the two preventive treatments differ according to the number of risk factors and possibly in patients with overweight. Thus, our results reinforce those previously reported in randomized trials supporting the use of NIV over HFNC in some subgroups of patients like those with chronic pulmonary disease, mainly those who develop hypercapnia at the end of the spontaneous breathing trial [5, 16], and patients with chronic heart diseases [16].

Not all risk factors associate the same reintubation rate, making this topic even more complex. Our analysis presented in Additional file 1: e-Figure 1 revealed that prolonged MV, APACHE II at extubation day, not-simple weaning, airway patency problems and secretions management had a stronger association with reintubation. However, a more complex model including these differences could limit its applicability at the bed side.

Considering the number of risk factors for reintubation in analyzing the response to preventive therapies showed that different patients are more likely to benefit from one treatment or the other depending on their level of risk. We found that patients with ≤ 3 risk factors (accounting for 65% of those considered at high-risk under the traditional definition) are likely to have a non-inferior response to preventive HFNC (reintubation rate 12.2%) than to NIV (reintubation rate 16.5%), whereas those with ≥ 4 risk factors are likely to have a better response to preventive NIV (reintubation rate 23.9%) than to HFNC (reintubation rate 45.3%). This result is in accordance to that recently reported by Casey et al. [15] showing an additive effect of simultaneous risk factors and possibly leading to underuse of NIV.

Considering the presence of overweight in analyzing the response to preventive therapies showed that some specific physiological effects of HFNC might be detrimental in this population. The exclusion in the original study of hypercapnic patients reveals that obesity hypoventilation syndrome and the risk for postextubation hypercapnic respiratory failure can be an important mechanism explaining the benefit obtained in a previous study with preventive NIV [17]. Other mechanisms leading to hypoxemia described in overweight patients can be responsible for the worse results observed in our study (e.g., atelectasis after extubation secondary to increased intraabdominal hypertension). Thus, excluding hypercapnic patients our study was underpowered to obtain a better result with NIV. In addition, other differences between the protocol by Thille et al. and ours (e.g., more prolonged preventive therapy according to clinical condition instead of fixed period of 24 h, respectively, different definition of reintubation 7 vs 3 days, respectively, and our higher median number of risk factors), could have led to a better reintubation rate in the HFNC group in the study by Thille et al. (7% vs 33%, respectively). Furthermore, combining HFNC and NIV in the study by Thille et al. led to a 24 h a day prevention protocol while our only NIV led to a 14 h a day in that group, probably limiting efficacy of NIV therapy.

This is a complex topic: obesity has not been associated to an absolute increase in the reintubation rate in previous studies [22], but seems to benefit with preventive NIV [2, 17]. This obesity paradox in weaning could be partially explained with the recent results by Torrini et al. [21] showing that obesity can be a protective factor for reintubation. However, which conditions are necessary to fully obtain that protective effect remains to be elucidated.

Study limitations

The present study has some limitations. First, it is a post hoc analysis of a non-inferiority trial that can yield false-positive results due to fixed boundaries for pre-planned analyses that include groups with different margins of benefit. The original study defined non-inferiority with a between-group difference in treatment failure < 10%, showing a reintubation rate in NIV patients of 19.1% vs 22.8% in HFNC patients. These findings are currently being tested in a prospective randomized trial (Clinicaltrials.gov ID: NCT04125342). Second, in the absence of a validated model to predict extubation failure, the original study used the ten risk factors for which the most evidence was available. The factors that most increased the reintubation rate in the original trial were prolonged mechanical ventilation, APACHE II score > 12 on the day of extubation, not-simple weaning, and inability to manage respiratory secretions. However, these results are highly dependent on the specific definition of risk factors and are difficult to benchmark. Third, it should be also noted that some specific subgroups of patients were excluded in the original study (e.g., patients who were hypercapnic at the end of the spontaneous breathing trial before extubation). In addition, the retrospective design of this secondary analysis precludes definitive conclusions about the association between any given risk factor and the reintubation rate, as there are no predefined control group for comparison. Fourth, it is likely that the sample size was insufficient to ensure adequate statistical power if all risk factors were included, the adjusted model for the number of risk factors and overweight included only factors associated with reintubation at P < 0.10 (APACHE II on extubation day, difficult or prolonged weaning, inability to deal with respiratory secretions, and vascular disease as a comorbidity). Thus, effect modifications and interactions in addition to those related to the number of risk factors and BMI ≥ 25 cannot be ruled out.

留言 (0)

沒有登入
gif