I read with interest the recent publication by Joo et al. comparing posttetanic counts by accelerometry as compared with electromyography.  They demonstrated that accelerometry counted more twitches than did an electromyography system. However, while their work was meticulously conducted and analyzed, there is a slight but fundamental misstatement in their conclusion. This stems from a nearly universal misunderstanding as to the meaning of the term “twitch” in an era of quantitative technology.

Since the introduction of peripheral nerve stimulators in the late 1950s we have relied on visible and/or tactile assessment of the movement of the thumb in response to ulnar nerve stimulation (many other nerves can be stimulated and distal responses assessed; for the current discussion, I will restrict my comments to stimulation of the ulnar nerve). Since 1970, with the introduction of the “train of four” by Ali et al., this changed from estimating the “perceived strength” of a single response (relative to baseline) or perhaps watching the response to a tetanic stimulus, to counting the number of responses between 4 and 0 as a means of assessing the depth of neuromuscular blockade. This concept was extended by Viby-Mogensen et al. in 1981 with the posttetanic count – the number of responses generated by stimuli given after a tetanic stimulus.  However, both train-of-four and posttetanic count, at least in the hands of clinicians (rather than in the research setting), depended on the direct observation of responses—responses that we commonly call “twitches.”

However, what quantitative accelerometry and electromyography systems report is closely related to but fundamentally different from what providers see or feel. They record an electrical signal from the accelerometer attached to the thumb, or from the depolarization of the adductor pollicis or dorsal interosseous muscles in the hand. How does a device decide what to display as a twitch? This is not a black and white issue. Every electrical response cannot be called a twitch. For example, measurable responses are recorded on the TwitchView (Blink, USA) after each ulnar nerve stimulus—even in an unambiguous “no visible twitch” situation electromyography (personal observation, based on a datalogging system provided by the manufacturer). There is essentially never a “no response,” although the responses may be very, very small. So how do these devices decide when to display a twitch and which ones to ignore?

All available quantitative neuromuscular blockade monitors use internal algorithms to make such decisions. For example, they presumably set some “threshold” for the response. During the development of those algorithms, the manufacturers make an effort to “calibrate” their monitors against observed twitches. However, there is substantial variation from individual to individual—and there is even variability in visible counts reported by different observers, particularly in the face of substantial degrees of neuromuscular blockade and “faint” twitches. Baseline electromyography signals with the TwitchView (before the administration of any neuromuscular blocking agent) can vary by as much as sevenfold in apparently normal individuals (personal observation). Presumably this is due to differences in muscle mass, skin impedance, exact electrode placement, and other factors. Therefore, choosing an absolute response threshold amplitude may not always work. If that threshold is set too high, the device will “undercount” the number of twitches. If set too low, it may “overcount.”

None of these issues are of any importance when evaluating quantitative monitors for their most important feature—verifying the adequacy of full reversal before extubation (i.e., a train-of-four ratio of 0.9 or greater with electromyography or greater than 1.0 with uncalibrated accelerometry). These measurements are made in the face of relatively vigorous twitches and high-amplitude electrical responses. However, as the responses progressively diminish (and when visible twitches begin to disappear), this issue becomes manifest. Clearly a “twitch is not always a twitch” when we are comparing quantitative devices with our own eyes—and we should not expect perfect concordance. This also applies when comparing “weak responses” between different devices.

Therefore, the precise conclusion from the data obtained by Joo et al. is that “the Phillips Intellivue with software version X installed displayed more post-tetanic counts than did the TwitchView electromyography monitor with software version Y installed…” The results apply only to these two devices with the software used at the time of the study—and do not necessarily apply to accelerometry or electromyography in general or to devices other than those used. This does not reduce the value or validity of the study by Joo et al. in any way—but clinicians should be aware of this issue. Just because the twitches counted by a quantitative device differ from what the eye perceives does not mean that the device is working incorrectly.