Duration of diaphragmatic inactivity after endotracheal intubation of critically ill patients

In critically ill adults, diaphragm weakness is exceedingly common and has been described in more than 60% of patients [1]. There is mounting evidence that it may be associated with poor outcomes in mechanically ventilated patients, including difficult weaning, prolonged intensive care unit (ICU) stay, a higher risk of complications, and an increased mortality rate [2‐5]. The most well-established form of mechanical ventilation related diaphragm injury (also known as myotrauma) is disuse atrophy, resulting from excessive ventilatory support, deep sedation and/or paralysis and has been referred to as over-assistance myotrauma [6, 7]. Daily examination of the diaphragm with ultrasound in critically ill ventilated patients has shown a progressive reduction of the thickness of the diaphragm occurring in the first three days in approximately 40% of patients [2, 3]. Studies in specific populations such as brain-dead organ donors have demonstrated muscle fiber atrophy and contractile dysfunction occurring within a few hours or days of controlled mechanical ventilation [8]. Diaphragm biopsies from ICU patients also demonstrated a reduction in active and passive diaphragm myofibrillar force and reduced cross-sectional area of slow-twitch and fast-twitch fibers as compared with control values [9‐11]. Importantly, maintaining diaphragm activity under mechanical ventilation has been shown to be protective [12, 13]. Changes in diaphragm thickness were correlated with the amount of inactivity and are associated with complications of mechanical ventilation and a poor outcome [12, 13].

While the detrimental effects of diaphragm inactivity are widely recognized, no data yet exists regarding the duration of diaphragm inactivity after intubation in critically ill patients, and the factors associated with a delay in the resumption of minimal or normal diaphragm activity have not been explored. Such a delay may be primarily related to the severity of illness, the lung severity or the oxygenation defect, impacting the sedation delivered. Therefore, before considering implementing new strategies to mitigate the risks of diaphragm dysfunction in critically ill patients, the effects of current practices should be better understood. In this study, we aimed to assess the duration of absent or excessively low diaphragm activity after intubation by continuously measuring the electrical activity of the diaphragm (EAdi) soon after intubation in critically ill patients and also assessed the relationship between EAdi and diaphragm activity on ultrasound. We hypothesized that resumption of activity was influenced by disease severity and sedation management.

Full methods can be found in the Additional file 1.

In this short-term observational physiological study, we evaluated the temporal pattern of EAdi resumption in critically ill adult patients requiring endotracheal intubation. The prevalence of excessively low diaphragm activity in the first hours after intubation was high and the median time to EAdi resumption was 22 h. Younger patients and those on higher doses of sedative infusions in the first 24 h of mechanical ventilation experienced delayed resumption of EAdi. Delayed EAdi resumption was not associated with the reason for intubation, the severity of illness, renal or hepatic dysfunction, the level of intensity of ventilation or oxygenation defect. Around the time of EAdi resumption, at least 20% of patients exhibited reverse triggering and this was the case for almost 40% of those still on controlled ventilation. EAdi amplitude was well correlated with TFdi as measured with ultrasound.

Our results show that an absent or abnormally low diaphragm activity is present in almost half of the patients (49%) for more than 24 h after intubation. Decreases in diaphragm strength and fiber cross-sectional area that have been demonstrated in both animal and human studies [2, 17‐19] might thus be a consequence of frequently suppressed activity early during invasive mechanical ventilation.

When clinically stable, transitioning a patient to spontaneous breathing using assisted modes of ventilation with reduced or interrupted sedation might preserve diaphragm activity and prevent the development of disuse atrophy. This approach, however, must be weighed against the need for lung protection [7], especially in the presence of excessive respiratory drive. We confirmed that a transition to assisted modes of ventilation, however, does not imply direct resumption of normal EAdi levels. In fact, the relatively low EAdi values in our patients might be the result of ventilator over-assistance, which is common and often unnoticed in modes such as pressure support ventilation [2, 20, 21]. This underscores the potential importance of monitoring respiratory effort during mechanical ventilation in order to target diaphragm activity within a safe physiological range [3, 7].

Our analysis shows that the presence of a continuous sedative infusion after intubation is the main risk factor for delayed EAdi resumption, while severity of illness, including intensity of ventilation and oxygenation, renal and hepatic function, reason for intubation and other clinical factors were not statistically associated. Our findings support a previous single center randomized controlled trial of a “no sedation” approach in mechanically ventilated patients who had significantly reduced time on the ventilator [22, 23]. Our findings confirm that the prescription of continuous sedation has a major impact on the resumption of diaphragm activity and this seems independent from a patient’s clinical severity. Although the effect of sedation was expected, the lack of relationship with severity suggests that the administration of sedation may not be dictated by the needs of the patients but more by predetermined behavior, habits and/or protocols. Also, these findings suggest that the administration of sedation may be amenable to improvement. Somewhat unexpectedly, we found that older patients had an earlier resumption of EAdi compared to younger ones. This finding is explained by the fact that older patients also received significantly less propofol, suggesting that clinicians limit continuous sedation in older patients given the potential for exacerbating delirium or risk of deep sedation. This also confirms indirectly that sedation might be administered based on clinicians’ decisions. Although our study did not capture in more granular details the characteristics of the patient at time of early ventilation, this result suggests that sedation administration for maintaining respiratory muscle activity could be better controlled.

A median EAdi threshold of 7 µV for 24 h was chosen to represent a resumption of clinically relevant EAdi. In our initial protocol we were interested in the first time EAdi values were > 5 µV, but there was significant variability in the EAdi signals and a “one-time” value of 5 µV for a few minutes was often very transient, (Additional file 4: Fig. E3), with a high risk of noise [24]. This voltage threshold value was chosen arbitrarily as half of the minimal 10 µV value observed in different studies, and it was reasoned that this would represent a conservative value for resumption of minimal EAdi. Healthy subjects have median EAdi values during normal spontaneous breathing close to 17 µV and above 10 µV in 92% of subjects [25]. We found a significant correlation between EAdi and TFdi, which demonstrated that a minimal TFdi of 15% as previously defined [2, 3], would correspond to an EAdi > 7 µV. This physiological rationale justified using a threshold of 7 µV, while performing a sensitivity analysis using a threshold of 5 µV, with essentially identical results, as well as analyses performed at an onset time of 12-h and 24-h, again with very similar results.

The presence of patient-ventilator dysynchrony may be driven by complex interactions between sedatives and opioids [26]. We were particularly interested in reverse triggering [16], which was frequently observed at the time minimal EAdi was detected during controlled mechanical ventilation. The physiological mechanism and clinical consequences of this specific dysynchrony are subject of ongoing research. Reverse triggering can lead to breath-stacking and might be associated with lung and diaphragm injury. In contrast, reverse triggering may also be a protective reflex to prevent diaphragm atrophy and further investigation is required to better understand its consequences. Although our analysis was limited by having only a 1 h recording at a fixed time of the day, therefore underestimating the true incidence, we observed that reverse triggering may be a frequent way for diaphragm activity to restart. This is further supported by previous research which demonstrated an increased amount of dysynchrony with higher as compared to lower dose propofol for sedation [27].

To improve ventilatory management of the critically ill patient our data suggest that monitoring of respiratory drive and inspiratory effort could be helpful [28, 29]. Low respiratory drive and effort as a result of ventilator over-assistance and/or excessive sedation may lead to diaphragm atrophy and dysfunction, while strong inspiratory efforts might lead to load-induced diaphragm injury [3, 6, 28]. Clinicians frequently give sedation to treat or even prevent obvious patient-ventilator dysynchrony, but this is probably at the expense of excessive sedation. EAdi amplitudes vary between individuals and safe upper limits for diaphragm activity are currently unknown. A diaphragm-protective ventilation strategy might employ EAdi monitoring or other techniques to identify patients at risk for ventilator over-assistance myotrauma, for assessing changes in respiratory drive, and to adjust ventilator settings accordingly [25]. EAdi is not a direct measure of breathing effort but studies have shown reasonable correlation with functional parameters such as muscular pressure (Pmus) [30]. Here we demonstrate a reasonably strong correlation between EAdi and diaphragm thickening measured by ultrasound in critically ill mechanically ventilated patients. A tidal variation in TFdi in the range of 15 to 30% was associated with shorter duration of mechanical ventilation in a large cohort of critically ill patients [3]. Accordingly, TFdi below 15% might put the patient at risk for development of disuse atrophy [3, 31]. The current study shows that patients with TFdi values below 15% indeed had abnormally low EAdi amplitudes, which was significantly different from patients with normal TFdi and strengthens the rationale for titrating ventilator assistance to reach a minimal TFdi of 15%.

This study had a limited number of enrolled subjects and was conducted in a single ICU. Since we found sedation practice to be a significant contributor, it is possible that location and practice variation could influence the time and magnitude of EAdi resumption. However, sedation practices and therefore EAdi resumption were not associated with patient severity of illness or oxygenation deficit suggesting that this practice may be modified. Pain, agitation, delirium and sleep are complex pathophysiological processes in critically ill patients [32]. Clinically, each of these components should be addressed separately and assuming that all intravenous agents act uniformly on EAdi recovery is an oversimplification. As analgesics, opioids seem to have vastly different effects on respiratory function and muscle activity compared to other agents, manifested in reductions in ventilator asynchrony and less suppressed EAdi activity compared to non-opioid based strategies [26, 27]. Conversely, propofol based sedation, even in healthy volunteers, has a profound impact on diaphragm activity [33]. While we attempted to separately analyze individual agents, they were used most often in combination. Future research in this area should attempt to study the individual effects of agents more precisely.

Secondly, we enrolled 75 patients in a 2-year time frame, which is much less than the number of acutely intubated patients in our ICU. This was explained in part by the logistical difficulty to conduct such a study in which recordings had to start almost at time of intubation. Our findings have reasonable external validity as the baseline characteristics of our cohort were representative of a general medical ICU patient population, but the actual number of surgical patients was low limiting our ability to draw associations in that cohort. Third, we felt that we used a reasonable threshold to detect minimal EAdi activity, as confirmed with our ultrasound data, but did not measure directly the effort of the patient. Although it would have been less sensitive, this could also be assessed with transdiaphragmatic pressure measurements, but would have added to the complexity and time sensitive nature of the measurements. Moreover, we focussed only on diaphragm activity in this study, while the time to recovery of other inspiratory muscle activity may be interesting to characterize. Finally, there were episodes of intermittent catheter dislodgement which required adjustment of positioning and in 7 patients the inability to measure recordings for the complete duration of the study. These catheter disconnections occurred after 24 h and the patients were kept in the early vs late comparison.

Following endotracheal intubation, critically ill adults experience suppression of diaphragm activity and the time to resumption of a reasonably minimal EAdi takes almost one day. Continuous infusions of sedatives and use of controlled modes of mechanical ventilation contribute to delayed recovery while severity of illness scores did not. Our results suggest that sedation practices in particular contribute to prolonged disuse of the respiratory muscles with a risk of diaphragm atrophy independent of patient severity of illness. Better monitoring of respiratory drive and integration of breathing effort to sedation practice algorithm may be of benefit.