High-flow nasal cannula (HFNC) is increasingly used in critically ill patients, because it has a number of advantages over conventional oxygen therapy (COT) or non-invasive ventilation (NIV) including favorable physiologic parameters, ease of use, and patient comfort [1,2]. It is mainly indicated for patients with acute hypoxemic respiratory failure [3,4], but can also be applied to various conditions including post-extubation [5-7].

HFNC has been demonstrated to be superior to COT in terms of reducing reintubation rate and post-extubation respiratory failure in patients with low risk for reintubation [6]. However, only a few patients meet the low risk criteria because intensive care unit (ICU) patients are getting older and often have multiple comorbidities [8]. Recently, two randomized trials compared the efficacy of HFNC on reintubation with NIV or NIV plus HFNC in high risk patients. HFNC was non-inferior to NIV but inferior to NIV plus HFNC as preventive measures in post-extubation period [9,10]. Based on these results, applying NIV during post-extubation period may be advisable in high risk patients. Indeed, current guidelines recommend preventative NIV after extubation in high risk patients though the strength of recommendation or the degree of evidence differs [11,12]. However, given that most of medical ICU patients belong to high risk group and NIV has several drawbacks such as poor tolerance, cost, and need for trained staff [13], it is difficult to apply NIV to every high risk patients in practice. Moreover, high risk group was defined differently in each study, and it was composed of heterogeneous patients. Therefore, it seems reasonable to refine and narrow down the high risk group.

Past studies found that cardiac dysfunction, either as a cause of respiratory failure or as an underlying disease, harbors increased risk of reintubation [14,15]. However, in those studies, COT was applied after extubation, so the situation is different from the current increasing use of post-extubation HFNC [16]. We planned this study to investigate association between cardiac dysfunction and reintubation rate within 72 hours in situation where HFNC was widely used during post-extubation period.

During the study period, 880 patients were treated with mechanical ventilation in the medical ICU. Of these patients, 222 were excluded because they failed to proceed to planned extubation; 52 patients were excluded because of insufficient EMRs; 288 were excluded because they were treated with COT after extubation; and 42 were excluded because they were treated with NIV after extubation. Therefore, 276 patients were included in the study. As EF data were not available for six patients, 270 were included in the final analysis (Figure 1). The median follow-up period was 22 days.

In this study, we categorized patients into normal and cardiac dysfunction groups based on EF, and compared reintubation rate under condition of HFNC following planned extubation. Interestingly, it was not higher in patients with cardiac dysfunction, which is established risk factor of reintubation. In addition, clinical outcomes and physiologic responses after extubation were similar between two groups.

For patients with cardiac dysfunction, positive pressure ventilation provides advantages of lowering left ventricular preload [19]. Because extubation eliminates these benefits, it could lead to pulmonary edema and respiratory failure. However, HFNC can supply inspiratory pressure based on flow rate. Roca et al. [20] reported that HFNC reduced inspiratory collapse of inferior vena cava (IVC) in patients with heart failure, indicating a decrease in preload. In addition, when the flow rate was increased from 20 L/min to 40 L/min, the inspiratory collapse of IVC was significantly reduced from 28% to 21% [20]. In our study, the median flow rate was about 45 L/min, which might be effective to prevent the increase in preload following extubation. Of note, all reintubation events occurred within first 24 hours in the cardiac dysfunction group. These findings might reflect rapid response to elimination of positive pressure in those patients who were greatly affected by positive pressure. However, there was no statistical significance, and few studies examined the relation of cause of extubation failure, its onset time, and patient characteristics.

In previous studies, patients at high risk of reintubation were heterogeneous, so the effectiveness of NIV may vary depending on patients’ characteristics. Ferrer et al. [21] reported that NIV was superior to COT in terms of respiratory failure and mortality, but there was no difference in reintubation rate, and a mortality benefit was found in patients with hypercapnia. In another study demonstrating the efficacy of NIV in high risk patients, about one third of patients had persistent hypercapnia [22]. A recent trial showed that NIV tended to work better in hypercapnic patients [10]. In contrast, Chang et al. showed that the reintubation rate was not different between heart failure patients treated with NIV or HFNC. Notably, the number of chronic obstructive pulmonary disease patients was small in that study [23]. In our study, carbon dioxide concentration of the two groups was in the normal range. Based on these findings, post-extubation NIV is helpful for hypercapnic patients. However, it does not seem essential for patients with cardiac dysfunction, who are normocapnic and supported with HFNC.

Although this study suggested the utility of post-extubation HFNC in patients with cardiac dysfunction, several limitations should be acknowledged. First, cardiac dysfunction was the accompanying characteristics, neither the primary diagnosis nor the cause of respiratory failure. In several studies, patients were regarded as high risk group when cardiac failure was the primary indication for mechanical ventilation [9,14,21]. Therefore, caution should be taken when applying our findings to such patients. Second, EF was not measured at the time of SBT or extubation. It would be better to understand the association between cardiac function and reintubation risk if EF was measured at that moment. However, unlike previous work that did not present EF measurement time [10,15,22], we provided EF near the time of SBT or extubation, reflecting cardiac function more precisely. Third, respiratory mechanics such as rapid shallow breathing index or maximal inspiratory pressure were not compared or included in the analysis though they were well-known weaning predictors. However, SBT was performed according to the institutional protocol in which tidal volume and respiratory are included [17], we believe that respiratory rate, tidal volume, and eventually rapid shallow breathing index would be in adequate range in both groups. In addition, this study was conducted retrospectively at a single institution in a small number of medical ICU patients. Therefore, one should be cautious to generalize our findings, and sample size would be not enough to detect statistical significance. As is well known, retrospective study is not free from bias. In patients with cardiac dysfunction, only those with a less severe condition may have been included in the study; severe cases may have been treated with NIV or tracheostomized. In fact, ICU LOS was shorter, and ICU mortality was lower in cardiac dysfunction group, although the differences were not statistically significant. Oxygenation indices were also better in cardiac dysfunction group. On the other hand, there were no differences in age, comorbidities, severity at admission, or degree of organ dysfunction on the day of extubation between the two groups. Renal failure was more prevalent in cardiac dysfunction group. In addition, the number of patients with NIV support following extubation was small during the study period. Patients in the cardiac dysfunction group were treated more frequently with dobutamine on the day of extubation. Use of dobutamine may have reduced the difference in EF between the two groups and helped to overcome hemodynamic burden following extubation. But, in Cox regression analysis, dobutamine use was not associated with increased reintubation rate (Table 3). Finally, we did not directly compare HFNC and NIV. However, as the reintubation rate did not differ between the normal function and cardiac dysfunction groups, we judged that cardiac dysfunction is not significant risk factor when HFNC is applied immediately after extubation.

In conclusion, cardiac dysfunction was not associated with increased reintubation rate within 72 hours when HFNC is immediately applied after planned extubation. Our findings suggest the possibility that risk factors of reintubation may differ depending on which modality are applied. Further research is required to identify and reassess risk factors of reintubation.