Comparison of the TOFscan and the TOF-Watch SX during pediatric neuromuscular function recovery: a prospective observational study

When using neuromuscular blocking agents (NMBA), quantitative neuromuscular monitoring is mandatory to optimize intubation time, monitor intraoperative muscle relaxation, determine adequate pharmacologic reversal agents, and reduce postoperative residual paralysis (Murphy et al., 2013; Naguib et al., 2017; Brull & Kopman, 2017; Fortier et al., 2015; Naguib et al., 2018). National guidelines, such as World Federation of Societies of Anaesthesiologists (WFSA) International Standards for Safe Practice of Anesthesia, and French Society of Anesthesia and Intensive Care (SFAR), recommend incorporating objective neuromuscular monitoring into daily practice (Plaud et al., 2020; Gelb et al., 2018; Nemes & Renew, 2020). Although many anesthesiologists rely on subjective evaluation, dependence on such clinical parameters has limited success (Naguib et al., 2018).

Clinical assessment of adequate reversal is challenging in children given difficulties with communication and non-compliance with instructions. Previous evidence indicates that 10–28% of children experience postoperative residual block (i.e., train-of-four (TOF) ratio <0.9), with 6.5% displaying severe block (i.e., TOF ratio <0.7) (Ledowski et al., 2015; de Souza et al., 2011). Complications due to residual block can be detrimental given that children have smaller oxygen reserves and are more vulnerable to airway collapse (Fortier et al., 2015; von Ungern-Sternberg et al., 2006; Hardman & Wills, 2006).

The current standard device for determining the depth of muscle relaxation is the TOF-Watch SX (Organon, Swords Co., Dublin, Ireland), which is a one-dimensional acceleromyograph requiring initial calibration (Colegrave et al., 2016). Without calibration, TOF-Watch SX transmits erroneous information regarding the degree of residual neuromuscular blockade (Martin-Flores et al., 2012). Acceleromyography-based devices frequently overestimate TOF ratio more than mechanomyography or electromyography (Suzuki et al., 2006). Therefore, when using acceleromyography, “normalization” of taking baseline TOF ratio into account by dividing the expected TOF ratio with baseline TOF ratio at each interval throughout the neuromuscular recovery is recommended (Claudius et al., 2009).

Recently, TOFscan (Drager Technologies, Canada), a new device also using accelerometry, has emerged on the market. TOFscan does not require onerous calibration and differs from the TOF-Watch as it measures three-dimensional acceleration. A previous evaluation of the agreement between the two devices during neuromuscular recovery in adults indicated good agreement when TOF-Watch SX was calibrated and normalized (Murphy et al., 2018).

This study aimed to compare the performance of TOF-Watch SX and TOFscan in children. We hypothesized that TOFscan measures would be comparable to TOF-Watch SX’s among children aged ≤12 years.

In total, 39 children were screened, of whom 35 were enrolled. Demographic characteristics are shown in Table 1. Four patients were excluded due to missing data, and 16 required additional NMBA maintenance. Although the supramaximal current was secured, the TOF-Watch SX ratio of 12 children ended before reaching 1.0 due to short surgical time. Therefore, 12 data pairs of TOF ratio 1.0 in both normalized and non-normalized groups were missing. However, all 35 data pairs were secured until a TOF ratio of 0.9.

Baseline TOF ratio was 0.029 higher (P=0.026) for TOF-Watch SX than TOFscan (1.026±0.073 and 0.997±0.009, respectively). The time to reach TOF count 0 was shorter in TOF-Watch SX (100.5±34.6 s) than in TOFscan (112.1±34.6 s, P=0.018). Average calibration time for TOF-Watch SX was 36.88 s. At a TOF ratio of 0.7 in TOF-Watch SX, the TOF ratio of TOFscan was similar (0.712±0.116, P=0.545; Table 2).

In non-normalized TOF ratios, the bias was 0.010 (95% CI, −0.003 to 0.0236) with SD of 0.127. The 95% limits of agreement were −0.239 to 0.259 for non-normalized ratios. The CI for lower limit of agreement was −0.263 to −0.216, and CI for the upper limit of agreement was 0.235 to 0.282. Normalized TOF-Watch SX and TOFscan data showed bias of −0.002 (95% CI, −0.013 to 0.010) with SD of 0.111. The 95% limits of agreement were −0.219 to 0.216 for normalized ratios. The CI for lower limit of agreement was −0.239 to −0.119, and CI for the upper limit of agreement was 0.196 to 0.236 (Table 3). A discrepancy was detected in seven datasets because of additional TOF ratio gained at TOF ratio of 1.0 in the normalization process (non-normalized, 338, vs normalized, 345). Each measurement is displayed in the Bland-Altman plot along with each 95% CI shown in green (upper 95%) and red (lower 95%) (Fig. 2, Supplement 1).

The Spearman rank correlation coefficients (ρ) were −0.237 and −0.236 for non-normalized and normalized biases, respectively. Similarly, Kendall rank correlation coefficients (τ) were −0.164 and −0.160 for non-normalized and normalized groups, respectively. This suggests a weak relationship in the bias over the range of measurements, regardless of normalization. The intraclass correlation coefficients were 0.921 and 0.929 for non-normalized and normalized biases, respectively, indicating the high reliability of both devices, regardless of normalization (Table 4).

This study presents lack of bias and good concordance between TOF-Watch SX and TOFscan in children. The normalized group displayed less absolute value of bias with smaller SD and narrower bidirectional (upper and lower) limits of agreement than the non-normalized group. Normalized values coincided with those of TOFscan more accurately than the non-normalized values. The Spearman rank correlation coefficient (ρ) and Kendall rank correlation coefficient (τ) of normalized and non-normalized measurements indicated a weak correlation over the course of neuromuscular recovery.

Three-dimensional acceleromyography devices, such as TOFscan, Stimpod NMS 450 (Xavant, South Africa), and Mindray neuromuscular transmission transducer (Shenzhen, China), use three perpendicular piezoelectric probes to thoroughly measure freely moving target muscles. To date, only two studies, both in adults, have compared the performances of TOF-Watch SX and TOFscan. One reported no significant differences between the two but suggested that TOF-Watch SX is more sensitive during deep neuromuscular blockade and that better agreement was observed along the recovery (Colegrave et al., 2016). Another study detected minimal bias between these devices (Murphy et al., 2018). Our findings also indicate minimal bias; the absolute value of bias was also smaller. Also, Spearman rank and Kendall rank correlation analysis revealed a weak negative relationship. It indicates that size of bias did not significantly change but that a decreasing trend over bias was observed during neuromuscular recovery. This result supports previous observations of a similar but slight negative trend in adults (Murphy et al., 2018) and observations that better agreement was achieved at more complete levels of recovery (Colegrave et al., 2016).

Assessing three-dimensional acceleration in children is challenging due to difficulty in isolating APB movement. The APB originates from two heads (oblique head from the second and third metacarpal bones, and transverse head from the third metacarpal bone) and inserts at the base of the thumb’s proximal phalanx. Contraction of the APB brings the thumb’s tip to the center of the palm. However, adjacent muscles, such as the opponens pollicis (OP) and flexor pollicis brevis (FPB) muscles, also contribute to similar movements. Given the shallow skin-to-nerve and nerve-to-nerve distance in pediatric patients, applying same the “3-cm” distance between the positive and negative electrodes can cause extra stimulation to the median nerve, which innervates OP and FPB (Alanazy, 2017). The distance between the two electrodes determines the penetration depth (Fuchs-Buder, 2011). With the distance between the electrodes relatively far apart in the relatively short pediatric forearm, nearby muscles may also react. Strapping prevents only the four fingers, not the thumb from moving. OP and FPB cannot be restricted by mere strapping. Moreover, the deep part of the OP and FPB in 20% of the population is often innervated by the deep branch of the ulnar nerve (C8, T1) (Gupta & Michelsen-Jost, 2012). The erroneous extra-apposition of the thumb due to the mixed effect of unwanted muscle movement is likely to occur in a pediatric patient with smaller hands and more proximal nerve-to-nerve distance. This anatomical inevitability may cause unwanted acceleration and induce a larger correlation coefficient than that demonstrated in Murphy et al.’s previous report.

Normalization of the TOF-Watch values resulted in lower bias, narrower limits of agreement, and higher intraclass correlation between devices. TOF ratio >1 has been previously demonstrated in TOF-Watch SX (Suzuki et al., 2006; Claudius & Viby-Mogensen, 2008; Bowdle et al., 2019). Without normalization, if the initial TOF exceeds 1, subsequent TOF ratios will be overestimated. Alternatively, if initial TOF <1, subsequent ratios may be underestimated. Individual differences in distribution volume, muscle mass, NMBA clearance, and age-dependent maturation of neuromuscular junction may contribute to differences in initial TOF ratio. As normalization adjusts the inherent TOF ratio, the results support the use of normalized TOF ratios to provide a more accurate TOF ratio along the recovery in children (Suzuki et al., 2006; Claudius et al., 2009).

The completion of TOF count 0 in TOFscan was longer by approximately 12 s. According to previous studies, the time to TOF count 0 showed inconsistent results between TOF devices (Colegrave et al., 2016; Murphy et al., 2018). The reason for the discrepancy may be associated with the calibration process. The maximal value, the current that gives 100% response, is acquired with a calibration mode called CAL2, and 110% of maximal value is suggested as the supramaximal reference (Martin-Flores et al., 2012; Schreiber et al., 2011). Compared with TOFscan, which delivers uniform current intensity of 50 mA at 2 Hz, the calibration process in TOF-Watch provides 10 stimuli at 1 Hz until the optimal supramaximal current is detected. Such higher initial frequency may affect the gradual potentiation process—in other words, a staircase phenomenon (Zhou et al., 2013). The staircase phenomenon influences the onset time and duration of twitch depression by increasing T1 values (Martin-Flores et al., 2011). Nonetheless, some researchers argue that, despite the staircase phenomenon, T1 through T4 increase in the same proportion, therefore not affecting the TOF ratio (Suzuki et al., 2006). However, others reported the need for an extra-stabilization period after the staircase phenomenon to obtain a stable baseline (Martin-Flores et al., 2011). Therefore, calibration process may either effect T1 size or time to reach a stable T1 to T4 baseline. As such, further investigation is required to determine the reliability of the measurement devices at TOF count 0.

This study has some limitations. First, we did not compare acceleromyography with mechanomyography, despite acceleromyography itself not differing significantly from mechanomyography, especially when calibrated and normalized (Bowdle et al., 2019; Claudius et al., 2010). Second, no randomization was performed between the right and left arm, but previous research indicates that arm-to-arm variations did not display significant bias gap (Claudius et al., 2010; Hohenauer et al., 2017). Third, preload and device size were not individualized. However, acceleromyography is well reported to be precise with preload application ranging from 75–120 g (Claudius et al., 2009). Although we used a pediatric sensor, the TOFscan’s sensor might be too big for very small children. To correctly measure acceleration, the index finger should fit in the sensor’s hole after the curvature. Our youngest patient was a 16-month-old child whose finger length fit the sensor. However, to expand TOFscan’s validity to infants and neonates, a smaller sensor is required. Fourth, we did not identify the potential effect of calcium in neuromuscular monitoring. In motor neurons, presynaptic voltage-gated calcium channel activation produces neuromuscular junction’s synaptic vesicle to release acetylcholine, which eventually causes muscle contraction. Some studies report that increased ionized calcium levels decrease sensitivity to non-depolarizing NMBA and enhance neuromuscular recovery (Ju et al., 2017; Munir et al., 2003). When comparing TOF ratio, equalizing the use of calcium between groups would be helpful. Finally, our results reliably indicated concordance and lack of bias but no agreement between the devices. The limits of agreement’s absolute range exceeded 0.2, and this difference in TOF ratio indicates poor agreement. Careful interpretation is required since a 0.2 difference practically results in diverse decisions regarding extubation time and reversal dose. In conclusion, TOFscan demonstrated good concordance and is unbiased with TOF-Watch SX in children’s neuromuscular recovery, especially when normalized.