Background
Patient–ventilator asynchrony (PVA) in mechanically ventilated adults is associated with prolonged duration of mechanical ventilation (MV), increased use of sedatives and longer intensive care unit (ICU) and hospital stay [
1‐
3]. Although the occurrence of PVA in mechanically ventilated children is common as we and others have shown, the relationship between PVA and clinical outcome is unclear for this group of patients [
4‐
6].
Previously, we have shown in a heterogeneous group of mechanically ventilated children that one out of every three breaths was out of sync when the airway pressure and flow waveforms were visually inspected [
4]. However, such inspection is cumbersome and may not reflect the true prevalence of PVA as the neural breathing drive is not taken into consideration. Alternatively, electrical activity of the diaphragm measured with a specific nasogastric catheter (EAdi) or the esophageal pressure signal can be used and is in fact more accurate signals for identifying PVA [
6‐
11].
So far, use of these methods has been restricted to research purposes mainly because of the lack of ability to provide the clinician with real-time feedback of the level of PVA. In order to truly understand the clinical relevance of PVA in mechanically ventilated children, there needs to be a system that provides such feedback on both the occurrence and type of PVA. Recent advances have been made in the development of tools to automatically identify PVA [
2,
12‐
15]. Such real-time automatic analyses are needed for clinical trials investigating the efficacy of interventions targeted at reducing PVA and on patient outcome. Sinderby et al. [
16] developed an automated, objective and standardized neural index to quantify patient–ventilator interaction (NeuroSync) based on the measurements of EAdi and ventilator pressure waveforms. Determining patient–ventilator interaction by this method had a higher inter-rater reliability and proved to be more sensitive than manual analysis.
However, this new approach is only limited to ventilators capable of measuring EAdi. Furthermore, it mandates the insertion of an esophageal catheter which may be a disadvantage especially in the pediatric context. Transcutaneous recording of the electromyographic signals of the diaphragm (dEMG) may be considered as a suitable alternative [
17‐
19]. Although at this moment no correlation studies between dEMG and EADi have been performed, this non-invasive, easy-to-perform technique provides reproducible electromyographic signals of the diaphragm [
17]. We therefore tested the hypothesis that it would be feasible to automatically detect, quantify and display patient–ventilator interactions using a modified NeuroSync index (dEMG-phase scale) in mechanically ventilated children when analyzing dEMG together with ventilator pressure and flow versus time waveform.
Discussion
To our knowledge, this is the first study reporting that the interaction between infants and children and the mechanical ventilator can be quantified in a real-time non-invasive manner using transcutaneous electromyographic respiratory muscle recordings. Quantification of patient–ventilator interaction using a modification of a previously described method (dEMG-phase scale) proved to be a feasible and reliable method, reflected by high ICCs for both trigger and cycle-off errors and the dEMG-phase scale. This method may have important implications for both clinical use and research purposes, as it is not restricted to one type of ventilator and it is a non-invasive tool implying that it can be easily implemented in the pediatric population.
To date, measuring the electrical activity of the diaphragm was only feasible using a specifically developed esophageal catheter linked to a specific brand of ventilator. Alternatively, we used surface electrodes with their own limitations. First, when measuring respiratory electrical activity by surface electrodes, also other muscle activity could be measured, a phenomenon known as cross-talk [
25]. Reassuring, however, is that we noticed only minimal cross-talk in our study comparable with other studies [
17,
26]. Second, electrical interference by machines commonly used in the intensive care unit may interfere with the measured electrical activity [
26‐
29]. We therefore applied a 50 Hz notch filter and were subsequently able to use all data registrations. Third, the use of template subtraction and gating to remove heart activity from the dEMG signal could theoretically interfere with the exact determination of the onset and termination of the neural inspiration. Yet, Hutten et al. used a dEMG signal in which such a filter removed the ECG signal. They found that this filtered signal correlated well with tidal airflow and was fairly robust against time delays [
26].
In our previous study, we found that PVA was extremely common in mechanically ventilated children and the predominant type was ineffective triggering [
4]. There was some type of asynchrony in one out of every three breaths. Unlike the present study, we had detected PVA by analyzing the ventilator flow and pressure waveforms. Such a method is prone to underreporting the true prevalence of PVA [
10]. This is confirmed by the results from the present study, in which we found that only 12.2% (1.9–33.8) of breaths was synchronous. Thus, incorporating dEMG measurements and analyzing the waveforms automatically using the dEMG-phase scale is superior to manual analysis of ventilator waveforms alone. By incorporating the dEMG-phase scale, we were able to improve our definition of PVA [
4]. For instance, breaths with relative timing differences > 33% were now classified as dyssynchronous instead of asynchronous, which may explain the difference in occurrence of PVA between this and previous studies. Although the error limits were adopted from Sinderby et al., these limits were arbitrarily chosen and may not be appropriate for defining synchrony, dyssynchrony, and asynchrony in mechanically ventilated children [
16]. To determine more accurate inspiration times studies comparing dEMG with the esophageal pressure versus time tracings have to be performed. In addition, in the present study we have used a different brand of ventilator (AVEA, CareFusion, Yorba Linda, CA, USA) than in our previous study (EvitaXL Draeger Medical, Lubeck, Germany). Since a poor patient–ventilator interaction is not only caused by patient but also by ventilator-related factors, it may be surmised that differences in ventilator performance may influence the observed level of asynchrony [
30].
Implementing this method to quantify patient–ventilator interaction in the daily evaluation of mechanically ventilated children may be a very promising approach in individually setting the ventilator. For instance, the intra-breath patient–ventilator interaction diagram could be used to adjust the trigger sensitivity and for optimizing cycling criterion. It may be postulated that such guided individual titration may improve patient–ventilator interaction and decrease patient effort, although obviously, this assumption needs to be confirmed in clinical studies. To date, only in observational adults studies a significant association between the level of asynchrony and prolonged duration of mechanical ventilation and mortality has been shown [
1,
2]. Pediatric data are lacking. However, a better understanding of patient–ventilator interaction by means of dEMG monitoring may aid in understanding the effects of dys- and asynchrony on patient outcome in ventilated children.
Some limitations of our study need to be discussed. First we used the surface EMG of the diaphragm in the same manner as the EADi signal. To our best knowledge, no correlation studies between the surface EMG and EADi have been performed. Sinderby et al. have shown in a small study population that peak EAdi signals obtained from esophageal catheter were comparable with peak costal surface EMG signal [
31]. This manuscript shows that automatic algorithms for transcutaneous electromyographic respiratory muscle recordings to quantify patient–ventilator interaction in mechanically ventilated children can be developed. However, this does not mean surface EMG is equivalent to EADi measurements. More validation studies need to be performed. Secondly, we included patients in the 24 h prior to extubation. The rationale for this was the expectation that patients in the weaning phase are likely to have more interaction with the ventilator. In fact, Emeriaud et al. indeed showed a significant lower diaphragm activity during the acute phase of illness [
32]. Last, it should be noted that currently to estimate a patient’s respiratory center output, only dEMG was analyzed. Analyzing both dEMG and EMG of intercostal muscles simultaneously may have an added value in patients characterized by an early trigger error, because the ventilator might be triggered by inspiratory flow generated by intercostal muscle activity. Moreover, it is shown that external intercostal muscles are normally stimulated before the diaphragm as an initial stabilization of the chest wall to make diaphragmatic contraction more efficient [
33].
Conclusions
The transcutaneously measured electrical activation of the diaphragm is a useful signal for evaluating and monitoring patient–ventilator interaction. The dEMG-phase scale was demonstrated to be reproducible and to be an accurate scale to quantify patient–ventilator interaction of mechanically ventilated children. This method may have important implications for both clinical use and research purposes, as it is not restricted to a type of ventilator and it is a non-invasive tool implying that it can be easily implemented in the pediatric population.
The described method could be the first step to determine the effects of patient–ventilator synchrony, dyssynchrony and asynchrony in mechanically ventilated children. Further research is needed to validate cut-off points used in this study. Finally, validation studies are needed to explore the correlation between electrical signals from the diaphragm measured transcutaneously and EADi signals obtained by an esophageal catheter.
Authors’ contributions
AAK and RGTB provided equally to the manuscript and share first authorship. AAK and RGTB collected and analyzed the data. RGTB drafted the manuscript. JB contributed to the statistical analysis and provided intellectual content to the manuscript. LvE and FdJ advised on EMG analysis and provided intellectual content to the manuscript. MK supervised the study and is responsible for the final version of the manuscript. All authors read and approved the final manuscript.