High-flow oxygen therapy in tracheostomized patients at high risk of weaning failure

High-flow nasal cannula oxygen therapy is the administration of a warmed and humidified air/oxygen mixture at a flow rate between 20 and 60 L/min [1] through nasal cannulae (HFNC). HFNC is increasingly used in several clinical contexts, particularly in de novo hypoxemic respiratory failure [2‐5]; post-extubation [6, 7] and in post-cardiothoracic surgery [8]. Compared with conventional oxygen therapy (conventional O₂), HFNC produces several physiological effects [9], which include: a better matching between the delivered gas flow mixture and patient’s spontaneous inspiratory flow; a positive end-expiratory pressure (PEEP) effect (generally between 2 and 8 cm H₂O [10, 11]) and a “CO₂ wash out” effect from the upper airways [12, 13]. Further, the humidified and warmed gas mixture favors the mucociliary function and reduces upper airway resistance [2, 14]. The synergistic combination of these mechanisms leads to improved oxygenation [2], decrease in neuro-ventilatory drive and work of breathing [15‐17].

High-flow oxygen therapy can also be applied to the tracheostomy opening (T-HF) in tracheostomized patients. Since HFNC decreases the risk of re-intubation in patients at risk of extubation failure [6, 7], it is reasonable to hypothesize that T-HF could aid the separation from mechanical ventilation in mechanically ventilated, tracheostomized patients at high risk of weaning failure. However, T-HF is a novel therapy and it is not known whether it replicates the physiological advantages of HFNC.

Our study hypothesis was that T-HF, compared to conventional O₂, would decrease the neuro-ventilatory drive, work of breathing, respiratory rate and improve gas exchange, in a mixed population of tracheostomized patients at high risk of weaning failure [18].

The CONSORT study diagram (Fig. 3) shows that in the study period, 22 out of 38 tracheostomized patients with prolonged weaning from mechanical ventilation at high risk of weaning failure were eligible for the study and 14 were finally enrolled and completed the study protocol. Patients’ demographics and clinical characteristics are shown in Table 1.

In this study, we tested the effects of high-flow oxygen therapy through a dedicated tracheostomy interface on gas exchange, neuro-ventilatory drive, respiratory rate and work of breathing in tracheostomized patients at high risk of weaning failure. Our data suggest that T-HF does not produce the same physiological effects as HFNC on these outcome parameters. Although our study was not designed for a detailed examination of the physiological mechanisms generated by high flow at the tracheal level, we discuss the possible physiological mechanisms that may explain our results.

Recent studies have shown that during HFNC, patients decrease their minute inspiratory volume maintaining stable the PaCO₂ [15‐17]. These findings indirectly support the hypothesis that HFNC decreases the neuro-ventilatory drive and work of breathing mainly through its “CO₂ wash out” effect [11, 13, 15‐17]. Since we found that T-HF does not impact on the neuro-ventilatory drive and work of breathing, we speculate that these negative results are explained by the fact that T-HF has negligible “CO₂ wash out effect” compared with HFNC. This can be explained by the fact that HFNC washes out the CO₂ mainly from the nasopharyngeal anatomical dead space [12, 13], whereas the tracheostomy tube bypasses the upper airways. The effect of tracheostomy on dead space has been confirmed by Chadda et al. [30] who show that physiological dead space increases from 156 ± 67 mL before decannulation to 230 ± 82 ml after decannulation.

We were not able to show any impact of T-HF on the PaO₂/FiO₂ ratio, compared with conventional O₂ (Table 2). We have no data on tracheal pressures and lung volumes, but—given that conventional O2 and T-HF achieved similar results on a multitude of parameters—it is unlikely that a “PEEP effect” occurred in our patients. Our results are in contrast with those reported in a research letter by Corley et al. [31], showing a slight, but statistically significant improvement in oxygenation with T-HF compared to conventional O₂ in prolonged wean patients. Corley et al. applied T-HF for 15 min, whereas our study periods lasted 1 h and the gas delivery flow rate during conventional O₂ was 15 L/min versus 10 L/min in our study. However, it seems unlikely that these methodological differences could explain the discrepancy between Corley’s results and ours.

The EAdi is a “processed” diaphragmatic electromyography signal recorded through an array of electrode pairs mounted on the wall of a nasogastric feeding tube [32] and is proportional to the intensity of the electrical stimuli directed to the diaphragm [24, 33, 34]. In other words, the EAdi is a measure of the discharge rate of phrenic nerves that have been shown to reliably reflect the neuro-ventilatory drive and work of breathing [24, 32‐36]. Recently, Bellani and coworkers [23] demonstrated that EAdi can be used to estimate the instantaneous work of breathing, as we did in our study. Liu and coworkers found that an EAdi_PEAK lower than 15–20 μV during a spontaneous breathing trial was associated with extubation success [25]. Dres et al. [37] and Barwing et al. [38] obtained similar results. In our study, we found that the EAdiP_EAK was below 20 μV (Fig. 5) both during the T-HF and the conventional O₂ periods. We speculate that these findings could be explained by the fact that our patients were tracheostomized. Previous studies have shown that the neuro-ventilatory drive is up to 30% lower in tracheostomized patients [30], probably because the tracheostomy decreases the overall airway resistance [39] and the physiological dead space [30]. Furthermore, to our knowledge, data on the normal ranges of neuro-ventilatory drive in tracheostomized patients are scanty. In analogy with EAdi, the PTP_musc/min, a well-known index of work of breathing, was within its “acceptable” range (i.e., between 50 and 150 cm H₂O/s/min [40, 41]), in 72%, 65% and 72% of our patients in the T-HF1, conventional O₂ and T-HF2 periods, respectively (Fig. 5).

We must acknowledge some study limitations. First, we conducted a non-randomized ON–OFF–ON cross-over study that, for practical reasons, could not be blinded. Furthermore, we were not able to measure several respiratory parameters (VT, EELV, airways pressure) that could have provided us with a more complete interpretation of the treatment effect. However, our study was conducted in spontaneously breathing patients and we sought to avoid any modification in breathing pattern caused by the measurement apparatus. Second, we estimated Pmusc based on a method recently validated by Bellani and coworkers [22] and applied in different studies, but we must point out that this method assumes a diaphragm contribution to the overall Pmusc of 75%, as occurs in normal conditions [42]. On the other hand, if the contraction of the accessory inspiratory muscles is dominant (as occurs in case of paradoxical abdominal movements) the estimation of Pmusc from EAdi could be biased. However, we excluded from our study patients showing signs of paradoxical abdominal movement and use of accessory inspiratory muscles (see “Methods” section). In addition, the method described by Bellani et al. [23] assumes a linear relationship between EAdi and Pmusc at different lung volumes but other authors have shown a nonlinearity between EAdi and lung volumes [24]. However, this occurs only in case of intense diaphragmatic contractions [24]. Fourth, our physiological study was not designed to test the impact of T-HF on patients’ comfort. Over a prolonged period, however, T-HF could better preserve the mucociliary function compared with conventional O₂ [14], improving the secretion clearance and patient’s comfort. Further studies are needed to test this hypothesis. Fifth, we used a fixed high flow rate (50 L/min) but it is possible that, in analogy with HFNC [43], our results would have been different at different flow rates. Further studies are needed to construct a dose–response between T-HF flow rate and physiological effects. On the other hand, a different interface design, for example, with a different flow delivery angle (Fig. 2), could improve the CO₂ clearing from the tracheal space, compared with the interface tested in the present study.

In conclusion, in our cohort of tracheostomized patients at high risk of weaning failure, after disconnection from the ventilator T-HF did not affect neuro-ventilatory drive, work of breathing, respiratory rate and gas exchange, compared to conventional O₂. The present findings might suggest that physiological effects of high-flow therapy through tracheostomy substantially differ from nasal high flow.