Electrical activity of the diaphragm (EAdi) as a monitoring parameter in difficult weaning from respirator: a pilot study

Successful weaning from respiratory support is a matter of major importance in critically ill patients recovering from acute respiratory failure. The assessment of lung function before discontinuation of ventilator support is essential to predict the patient’s ability to breathe spontaneously without mechanical assistance. Weaning readiness has been the subject of research over the past years, and numerous parameters have been proposed as predictors of weaning success, indicating the lack of a reliable solution. An easy-to-obtain and therefore widely accepted parameter is the rapid shallow breathing index (RSBI), which is the ratio of respiratory rate and tidal volume (RR/VT) [1, 2].

Electromyography of the diaphragm has been used to describe the patients’ functional status and neuromuscular coupling in numerous studies [3‐5] and has been shown to provide a real-time, breath-by-breath measure of neural respiratory drive [6]. With the clinical introduction of neurally adjusted ventilatory assist (NAVA) as a ventilator mode, the electrical activity of the diaphragm (EAdi) gives the clinician a monitoring parameter of diaphragmatic function. The electromyogram of the crural diaphragm is recorded by a set of electrodes incorporated into a special nasogastric tube (EAdi catheter). The filtered and amplified signal is displayed on the ventilator’s control panel [7]. Experimental studies have shown its value in measuring diaphragmatic function [8, 9], as the amplitude of the signal correlates with the patients’ respiratory drive [3, 10]. Clinical studies and results from animal experiments confirmed the physiological concepts [11, 12]. Although primarily intended to control the ventilator, the EAdi allows monitoring of the respiratory drive in relation to tidal volume. The tidal volume (VT) during unassisted spontaneous breathing is proportional to the electrical activity of the diaphragm and reflects the neural output of the central respiratory regulation in healthy dogs [8].

The diaphragmatic electromyogram has been used to describe respiratory effort in patients [13]. In patients with poor neuromuscular coupling, tidal-volume generation might not follow the patients’ central respiratory drive. As shown by Sinderby and colleagues [14], severe chronic obstructive pulmonary disease and restrictive pulmonary disease induce increased diaphragm activation at rest. During weaning, respiratory drive and thus EAdi are expected to increase earlier than conventional assessment parameters of muscular function (for example, tidal volume, respiratory rate, gas exchange, and clinical signs of exhaustion). The EAdi signal might therefore be a better monitoring parameter during weaning. Recently it was shown that EAdi-derived indices might provide reliable, early predictors of weaning outcome [15], and EAdi has also been used to monitor weaning from the ventilator in pediatric patients [16]

In this pilot study, we compared the EAdi with conventional weaning parameters in difficult-to-wean patients scheduled for an SBT. Our primary goal was to observe the expected increase in the EAdi. A secondary goal was to compare the conventional weaning parameters with those derived from the EAdi.

The study was performed in the anesthesiologic intensive care unit (ICU) at the University Hospital in Göttingen, Germany, after approval by the local ethics committee (Ethic Committee of the Medical Faculty, Georg-August-University, Göttingen). Written informed consent was waived because of the observational design of the study. All patients were recovering from respiratory failure after severe illness and were considered difficult to wean. These patients had either already failed an initial weaning attempt or had been on the ventilator for a long period and were expected to require more than 7 days for successful weaning. Patients who proceeded from initiation of weaning to extubation on the first attempt were excluded [17].

We identified patients as ready to wean by using criteria adapted from Macintyre et al. [18]:

All patients had a stable respiratory pattern at least 1 hour before starting the SBT. The patients were disconnected from the ventilator, and supplementary oxygen was supplied through a T-piece to keep peripheral oxygen saturation >90%.

The SBT was discontinued after 30 minutes or when one of the following events occurred [18]:

The correct position of the EAdi catheter was verified before the trial as described elsewhere [20]. The VT/EAdi ratio was calculated according to Wolf [16] by using VT (ml) and delta EAdi (EAdi_max - EAdi_min) to quantify the patients’ electrical effort with tidal volume. We used respiratory-inductive plethysmography (RIP) with a thoracic band at mammillary level and an abdominal band at the level of the naval. Baseline measurements were used to calibrate tidal volume measured by RIP with flow-measured values from the ventilator by using a simplified qualitative diagnostic calibration (QDC) [21], assuming an equal contribution of thoracic and abdominal expansion to VT. RIP data were then used to estimate tidal volume during the disconnection period in the first 11 patients (VT_est). For patients 12 to 18, we used an additional flow sensor (Bicore II; CareFusion, San Diego, CA, USA) to measure tidal volume. Relative changes in lung volume were derived for each patient by using RIP [22, 23].

Table 1 depicts basic patient data of the 18 patients studied. Twelve patients successfully completed the SBT (success patients), and six failed (failure patients). Based on the failure criteria defined in the Methods section [18], in 16 instances, failure criteria were met. A single patient could have more than one failed criterion. Reasons for failure were the following: an oxygen saturation >90% in two patients, an HR >120 bpm or a >20% change in HR in two patients, systolic BP >180 mm Hg or a >20% change in BP in four patients, a RR >35 breath/minute or a >50% change in RR in three patients (both criteria occurred in two patients), or clinical signs of distress in three patients.

With regard to the outcome after 72 hours, none of the six patients who failed the SBT was able to be weaned from mechanical ventilation, whereas eight of the 12 patients who passed the SBT were successfully weaned. Of the remaining four patients, three were already given a tracheotomy before the SBT and were classified as “prolonged weaning” [17]. The remaining patient successfully passing the SBT, who could not be weaned from the ventilator, received a tracheostomy later. At study inclusion, 10 patients had a tracheostomy; three of these failed the SBT. In the failure group, three of the six patients died in the ICU. This is consistent with the high SAPS II scores of our patients. No significant difference was found in gender, BMI, SAPS II, duration of ICU stay, duration of ventilator support, PEEP, and pressure support before study inclusion between failure and success patients. Disconnection from the ventilator led to changes in the weaning indices, EAdi_max (Figure 1), heart rate, and blood pressure in all patients (Table 2).

Respiratory rate during the SBT differed significantly between the failure and success groups (P < 0.05). Large differences occurred in other parameters (VT, VT/EAdi, RR/VT, BP, EAdi_max), but these did not reach a statistically significant level (Table 2). During the SBT, significant changes were noted in VT/EAdi, HF, BP, MAAS, and EAdi_max in the group of failure patients, which were not observed in the success group (Table 2). These results are illustrated by the tracings of two patients in Figure 2. The EAdi_max tracing of the success patient shows a stable course, increasing to a moderate degree only late during the disconnection, whereas the EAdi of the failure patient increased immediately after disconnection from the ventilator. The changes during the SBT in relation to the starting point (pre D) are summarized in Figure 3.

The analysis of the RIP data consisted of the thoracic (RIP_TX) and abdominal band (RIP_AB) and the sum of the two (RIP_SUM) (Figure 4). Adding the changes of the thoracic and the abdominal band can be used to illustrate overall changes in functional residual capacity (FRC). This summation value decreases after disconnecting the patients from the ventilator (D(f)), reflecting a decrease in thoracic circumference. In the failure patients, an increase in the summation of both bands toward the end of the SBT (D(l)) is visible. After reconnection to the ventilator, this value reflects an increase compared with the pre-disconnection level in all patients, which is more pronounced in the failure patients.

Our observations show that EAdi can be used as additional monitoring when weaning difficult-to-wean patients from the ventilator.

During the SBT, we assessed the patients by using a standard protocol [18]. The patient was thereby completely disconnected from the ventilator to see inspiratory muscle activity during the whole inspiratory period and to avoid effects of triggering pressure support by only brief diaphragmatic contractions [24]. Conventional criteria were used to decide when to terminate the SBT. Six of the patients failed the SBT, and 12 passed. The EAdi differed between the two groups, with an early increase in the failure patients that was more pronounced than that in those who passed the SBT. Our findings confirm those of Dres and colleagues [15], who described an early increase of EAdi-derived indices (such as EAdi_max/VT) in patients who failed a weaning trial. The result is reflected by the decreasing VT/EAdi ratio in the failure group. In an animal model, Grasselli and co-workers [5] demonstrated that the (VT/EAdi, no-assist)/(VT/EAdi, assist) ratio decreases with increasing ventilatory support, as does the work of breathing. Although evaluation of EAdi confirms physiological concepts that an increase of EAdi can be expected in a failing diaphragm, it does not outperform respiratory rate-derived indices in our study. The potential benefits of EAdi might be visible when the parameter is assessed continuously, helping to detect pathophysiologic changes early. Figure 1 shows the anticipated increase of EAdi in a failure patient.

A recent study by Wolf and co-workers [16] used the ratio VT/(EAdi_{max _} EAdi_min) to assess extubation readiness in a pediatric population. A lower ratio of tidal volume to deltaEAdi was found in the patients who passed the SBT, suggesting that patients with a higher deltaEAdi relative to tidal volume at the end of a 1-hour SBT may have better-preserved diaphragmatic function and therefore may be more easily weaned from ventilator support. Our observations showed that the patients who failed the SBT had an early reduction in the VT/EAdi ratio that was mainly because of a steep increase in EAdi. This could be explained by the patients’ efforts to maintain sufficient tidal ventilation.

Provided intact feedback exists to the patients’ central respiratory regulation, patients experience a decrease in minute ventilation (via increasing PaCO₂) and lung volumes (for example, FRC (via PaO₂ and stretch receptors)) when disconnected from the ventilator. Those who are unable to generate a sufficient tidal volume and maintain the FRC will increase inspiratory drive and thus neural output to the diaphragm. Insufficient increase in muscle effort will induce a further increase in EAdi, with ultimate failure of the SBT. The VT/EAdi ratio may be affected by increased resistive load (for example, upper airway obstruction), decreased pulmonary or chest-wall compliance (for example, pulmonary edema, obesity), mechanical alterations of diaphragmatic function (for example, ventilator-induced diaphragmatic dysfunction [25]) and reduced neuromuscular coupling (for example, critical illness neuromuscular abnormalities [17, 26, 27]). Under such circumstances, each patient has an individual threshold of respiratory efficacy, reflected by the VT/EAdi ratio, corresponding to the idea of a “fatigue zone” from the work of Tobin [28].

The study of Wolf et al.[16] differed from ours with regard to patient population and study design (1-hour SBT with PEEP reduced to 5 cm H₂O and pressure support reduced to overcome resistance of endotracheal tube), but we agree that the VT/EAdi ratio reflects the effectiveness of diaphragmatic activation and may be a useful parameter in predicting successful weaning, although the results must be interpreted with caution. The parallel evaluation of a pressure-derived weaning index (for example, P = 0.1) may elucidate underlying reasons for weaning failure.

We compared the EAdi and VT/EAdi with the RR/VT ratio, which is the most commonly used and best-established predictor of weaning [29]. The RR/VT ratio is influenced by PSV and PEEP [30, 31]. Therefore, we measured the RR/VT ratio during a T-piece trial, to maintain its typical cut-off value, although no difference appeared in the outcome of SBT either PSV or a T-piece [32, 33].

The failure and success groups differed greatly with regard to the two ratios, RR/VT and VT/EAdi, but the differences were not statistically significant. This could be attributable to the small population size. The failure group reached a median value of >105 bpm/L after disconnection form the ventilator, indicating extubation failure. However, because the RSBI was not available at the bedside during the trial, the decision to terminate the SBT was based on the clinical criteria described earlier. The decision to discontinue was reached relatively late (18 (8/22) minutes) compared with the early increase in EAdi at D (f) after 5 minutes.

The results of the RIP data add to the recent study by Dres and colleagues [23] on this matter and indicate that lung volume decreases the patient is disconnected from the ventilator (Figure 4, pre D-D(f)); mainly as a result of decreased thoracic circumference. This may be interpreted as because of loss of PEEP and pressure support during disconnection [23]. Lung volume increased toward the end of the SBT in the patients who failed the SBT (Figure 4, preD-D(l)). A possible explanation is the rapid, shallow breathing that these patients increasingly exhibit. Combined with expiratory-flow limitation, which has been demonstrated in patients with respiratory failure [34], this is the leading mechanism responsible for the development of intrinsic PEEP [35]. The additional effort is also indicated by the increase in EAdi. When the patient was reconnected to the ventilator, the thoracic and abdominal circumferences increased in the patients in both groups, probably because of the applied PEEP. The effect is more pronounced in the failure patients, which may indicate intrinsic PEEP.

Continuous EAdi monitoring provides additional information for the clinician caring for patients during weaning. We found that EAdi shows an early increase during the SBT in patients who failed the trial, and this increase can be detected before a patient finally fails according to clinical criteria. Thus, EAdi monitoring has the potential to guide respiratory weaning in difficult-to-wean patients and might help to prevent the muscular exhaustion during weaning trials, known to have detrimental effects on the patient. Further studies of the EAdi signal during liberation from mechanical ventilation and comparison with other weaning parameters is needed.