we observed unevenly distributed periods of “Low” and/or “High” neuro-ventilatory drive in critically ill patients ventilated in the PSV mode. Our data suggest that the clinical approach to PSV setting may result in significant periods of over and/or under-assistance.
By amplifying the patient’s breathing effort, mechanical assistance should normalize the neuro-ventilatory drive when the respiratory muscles are challenged by an absolute or relative increase in workload [16, 37, 38]. However, the PSV algorithm leaves to the clinician the task of setting the level of assistance and, therefore, of estimating patient’s neuro-ventilatory drive [17, 39, 40]. Physiological observations during stepwise PSV titration suggest that excessive or insufficient assistance (over and under-assistance, respectively) are associated with peculiar breathing patterns. Briefly, low VTs (i.e. lower than 5 ml/Kg PBW) and high RRs (i.e. higher than 30 breaths/min) denote under-assistance, whereas high VTs (i.e. higher than 8 ml/Kg PBW) and low RRs (i.e. lower than 20 breaths/min) denote over-assistance [17, 41, 42]. In this study, according to these physiological observations and to a consolidated clinical protocol [6, 34, 43], we titrated PSV to a VT between 5 and 8 ml/PBW and a RR between 20 and 30 breaths/min. The fact that we found wide variations in neuro-ventilatory drive since the beginning of the study, challenges the “classical” approach to PSV setting (Fig. 3). Despite our study was physiologically oriented and conducted in a small cohort of patients, we believe that these findings could be of clinical interest.
As reviewed elsewhere [8, 11, 25], the neuro-ventilatory drive originates from the respiratory centers, a network of interconnected neurons in the pons and medulla, modulated by gas exchange, physical exercise, sleep, emotional and behavioral inputs, pain, discomfort, sedation and analgesia. In pathological conditions, air trapping, decreased lung and/or chest wall compliance, increased airway resistance and/or respiratory muscle weakness may alter the coupling between patient’s effort and diaphragmatic excursion (neuro-ventilatory coupling) increasing the neuro-ventilatory drive [16, 44–46]. In our patients, we observed wide variations in neuro-ventilatory drive despite the sedation level was kept constant (i.e. RASS score between 0 and – 1) [33] throughout the study period. To explain these findings, we hypothesize: a) that our patients underwent to subclinical episodes of discomfort, pain, excess of sedation or sleep (sleep rhythm and architecture are disrupted in critically ill patients [47–49]), or, b) that they underwent to sudden variations in mechanical workload or metabolic status. Regarding the latter hypothesis, it is worth remarking that during PSV the assistance level is fixed and it has been shown that the response to a sudden metabolic [50] or elastic load [16] is “not physiologically oriented” (i.e. rapid shallow breathing). At variance with PSV, during neurally adjusted ventilatory assist mode (NAVA) the assistance is proportional to the EAdi [19, 51] and during proportional assist ventilation plus mode (PAV) the assistance is proportional to the patient’s inspiratory effort [37, 38, 52]. Thus, an attracting hypothesis is that the “proportional” modes would stabilize the neuro-ventilatory drive more than PSV [28, 53]. Further studies are needed to test this hypothesis.
During PSV, a “Low” neuro-ventilatory drive puts the patients at risk of diaphragmatic atrophy and patient-ventilator asynchrony [10, 54, 55]. Interestingly, we found an asynchrony index of 20.6 ± 3.5% during the of “Low” neuro-ventilatory drive periods (Table 3), well above the 10% threshold, that predicts prolonged weaning and ICU length of stay [56, 57]. On the other hand, an “High” neuro-ventilatory drive may induce diaphragmatic disfunction and patient self-inflicted lung injury [13, 58–60]. Previous data from our group suggest that a prolonged PSV period (48 hours) does not improve diaphragmatic efficiency [7]. Based on the present data, we speculate that the concurrence of “Low” and “High” neuro-ventilatory drive periods could explain our previous findings.
Overall, our physiological data strongly support the idea that the neuro-ventilatory drive should be continuously monitored during PSV. According to expert’s opinion, the EAdi is a “close” peripheral surrogate of the neuro-ventilatory [8, 13]. Monitoring the breathing effort could be an alternative [13, 25, 26] but EAdi amplitude and breathing effort may convey different information [24]. In case of diaphragmatic atrophy, the EAdi signal may be detectable in the absence of detectable pneumatic breathing efforts [61] and the relation between EAdi amplitude and breathing effort is influenced by the neuro-ventilatory coupling [37].
We must acknowledge the following study limitations. First: the EAdiPEAK thresholds applied in the present study to classify the neuro-ventilatory drive are empirical, although based on previous studies [7, 24, 28, 29] and on the manufacturer’s instructions (Maquet Critical Care AB, NAVA flow chart MX-6462 Rev 02/2015). Indeed, the EAdi signal is burdened by interindividual variability [62]. Nevertheless, we identified a rather broad “Normal” EAdiPEAK range and recently Piquilloud and coworkers observed an EAdiPEAK between 5 and 15 µV in healthy volunteers supported with different level of PSV and Liu and coworkers documented similar values in patients [63]. Furthermore, in our patients the PTPDI/min (a parameter of work of breathing) was largely within its normal range (i.e. between 50 and 150 cmH2O/s/min) during the periods of “Normal” neuro-ventilatory drive (Table 3). Second: an important issue to interpret our results is how the PSV level was set. As discussed above, our intent was to reproduce the “real life” clinical scenario [64–66]. However, we cannot exclude that different approaches to PSV setting would have produced a different impact on the neuro-ventilatory drive [28]. Third: ours was a physiologically oriented study and thus we have no data on the impact of the neuro-ventilatory drive patterns on diaphragmatic and pulmonary functions. Fourth, for technical reasons, we were not able to read reliable EAdi signals in 6 out of 26 patients (23%), despite our groups are experienced in clinical EAdi monitoring. This suggests caution when applying the EAdi monitoring in the clinical context.