The Staircase Phenomenon: Implications for Monitoring of Neuromuscular Transmission

Thirty adult patients classified as American Society of Anesthesiologists physical status I or II (aged 18–61 yr) scheduled to undergo elective surgical procedures were included in the study. All patients were free from neuromuscular disease and had a body mass index of 17.5–27.5. The protocol was approved by our hospital’s Human Subject Review Committee (Saint Vincent’s Hospital and Medical Center, New York, NY), and informed consent was obtained. Anesthesia was induced with 15–40 μg/kg alfentanil plus 2.0–2.5 mg/kg intravenous propofol, and laryngeal mask placement or tracheal intubation was accomplished without the use of neuromuscular blocking drugs. Anesthesia was maintained with nitrous oxide (65–70% inspired), 50–75 μg · kg⁻¹· min⁻¹propofol, and intermittent doses of fentanyl if required. Ventilation was controlled, and end-tidal carbon dioxide tension (Pco²) was maintained between 34 and 40 mmHg.

After induction of anesthesia, the evoked response of the adductor pollicis muscle to ulnar nerve stimulation at the wrist was recorded in all subjects. The monitor–stimulator used was the TOF-Guard^®acceleromyograph. The acceleration transducer was taped to the volar aspect of the thumb at the interpharyngeal joint. The study arm was immobilized in the TOF-Guard^®arm board, and the thumb was placed under a small preload with a single strand of an elastic rubber band (approximately 3.3 mm wide × 2 mm thick). Although the early literature on acceleromyography suggested that movement of the thumb should not be impeded, 8,9better stability is achieved when a preload is applied because the thumb is unlikely to return to exactly the same resting position after each stimulus in the absence of this preload. 10,11All data was collected on a TOF-Guard^®Flash RAM memory card for later transfer to a desktop computer. After calibration (auto II mode) of single twitch (T¹) to 100% using supramaximal stimulation, patients were divided into three groups.

• Group 1 (n = 10): TOFs were administered at 15-s intervals for 25 min. A total of at least 400 stimuli were administered.

• Group 2 (n = 10): Single stimuli at 1-s intervals (1 Hz) were administered for 10 min (600 stimuli). The stimulus pattern was then changed to TOFs delivered at 15-s intervals for an additional 10–15 min.

• Group 3 (n = 10): Before calibration of the TOF-Guard^®unit, a 5-s, 50-Hz supramaximal tetanic stimulus (250 stimuli) was administered at the ulnar nerve. Immediately thereafter, the acceleration transducer was attached to the thumb, and calibration of T¹was performed. TOFs were then administered at 15-s intervals for 22.5 min. The total elapsed time from tetanic stimulation to T¹calibration did not exceed 60 s and averaged 30–45 s.

In groups 1 and 2, the initial recorded values of T¹averaged 101 ± 4 and 101.5 ± 4%, respectively. Because there is a 15-s hiatus with the TOF-Guard^®unit between T¹calibration and the initial TOF administered, the first recorded value of T¹only averaged 90 ± 7.7% (range, 79–105%) in group 3 because twitch calibration occurred in the immediate posttetanic period. In view of these variations in “control” twitch height, initial T¹values (in all three groups) were normalized to equal 100% at time 0 (first recorded TOF response after calibration).

The following comparisons were made:

• Group 1: The average T¹response (percent of control) after 25 min of stimulation was compared using a paired Student t  test to earlier values of T¹(at 1-min intervals). Observed differences were considered significant if P  was less than 0.05.

• Group 1 versus  group 2: The increase in T¹was plotted as a function of the number of preceding stimuli rather than elapsed time for groups 1 and 2. T¹values for any given number of stimuli (up to 400) were compared using an unpaired Student t  test. Observed differences were considered significant if P  was less than 0.05.

• Group 1 versus  group 3: The value of T¹as a function of time was compared using an unpaired Student t  test. Observed differences were considered significant if P  was less than 0.05.

• Group 3: The average T¹response (percent of control) after 22.5 min of stimulation was compared using a paired Student t  test to earlier values of T¹(at 0.5-min intervals). Observed differences were considered significant if P  was less than 0.05.

After calibration, there was a progressive increase in the evoked T¹response (fig. 1, table 1). After 25 min of TOFstimulation, T¹increased to an average of 158 ± 26% (SD) of control (range, 126–211%). The rate of increase slowed with time, but it is not certain that true stability was ever achieved, despite 25 min of continuous stimulation. Although T¹only increased by 2% of control in the last 2.5 min of observation, this change (paired t  test analysis) was statistically significant with P < 0.01. In contrast, the evoked TOF ratio remained quite stable with time (103 ± 4.6% at time 0, 106 ± 4.1% at 25 min).

Group 2 was included in the study in an attempt to determine whether the critical factor determining baseline stability was the number of preceding or preconditioning stimuli as opposed to the elapsed time from initial calibration. We found (table 1, fig. 2) that even after 600 stimuli (at 1 Hz), twitch height did not seem to have reached a stable plateau. However, there were no statistically significant differences in twitch height between group 1 (TOF every 15 s) and group 2 (1 Hz) stimulation for any given number of preconditioning stimuli. As noted previously, at 400 stimuli, T¹had increased to 158 ± 26% of control in group 1 and 172 ± 19% in group 2 (P > 0.15).

When the pattern of stimulation was changed from 1 Hz to TOFs every 15 s, T¹then decreased over the next 4–8 min by approximately 20% to an average value of 142 ± 19% of control. Twitch height then stabilized at this lower value. In four subjects in this group, after TOF twitch stabilization, the stimulus pattern was changed to single stimuli at 0.10 Hz. In each instance, T¹decreased again to a new and lower plateau, decreasing from an average value of 143% to 127% of control (fig. 2).

Although (as in other groups) there was wide individual variability, the average  evoked T¹remained within a very narrow range (fig. 3, table 2). Even after 22.5 min of TOF stimulation, the mean T¹did not increase above 105% of control. This value was not statistically different from the evoked response seen at 2.5 min (99% of control, P > 0.05). In contrast to the still increasing T¹average seen in figure 1, T¹increased by only 3.5% during the interval from 5 to 22.5 min and by only 1.1% from 10 to 22.5 min. The average evoked TOF ratio recorded in this group was 112 ± 6.6%.

It has been known for many years that repetitive indirect stimuli of skeletal muscle may cause potentiation of evoked twitch tension. Posttetanic potentiation was clearly described in the mid 1930s. 12The positive staircase phenomenon seen after long trains of low frequency was well-known by the mid 1950s. 3In 1981, Krarup 2demonstrated in the rat that as the number of indirectly applied stimuli increases (up to a maximum of approximately 250), the evoked mechanical response but not the compound action potential increases in amplitude. This increase amounted to approximately 75% at 2 Hz, 85% at 3 Hz, and 95% at 5 Hz.

Therefore, although the staircase phenomenon is well-known to physiologists, its implications regarding the monitoring of neuromuscular function have been virtually ignored in the anesthesia literature. This is strange because most clinical investigators who use isometric force transducers (mechanomyography) have learned this themselves by trial and error. A common trick that has been passed on by word of mouth from one generation of investigators to the next is that a brief tetanus or a few minutes of stimulation at 1 Hz shortens the period of time necessary to achieve “baseline stabilization.” The results of this investigation support these anecdotal suggestions.

There have been several recent studies that address the issue of the need for a period of baseline stabilization before commencing studies of neuromuscular function. 5,6,13However, none of these studies suggest that repetitive stimulation could actually affect muscle contractility. We have been able to find only one clinical investigation that addressed this issue.

Lee et al.  14delivered TOF stimuli at 12-s intervals for 20 min (total = 400) under nitrous oxide–isoflurane anesthesia to three groups of healthy volunteers (n = 6, 8, and 6, respectively) and measured the evoked mechanical response. They found that T¹increased to 133 ± 12, 145 ± 23, and 149±15% of control. These values are not greatly different than the value of 158 ± 26% that we report in group 1 or 142 ± 19% in group 2. Although these authors did not directly study the influence of a 5-s, 50-Hz tetanus on baseline stability, they were able to show that when a 5-s tetanus preceded 2 min of TOF stimulation, onset/offset times for vecuronium were not different from those after 20 min of TOF stimulation alone.

Although we are reluctant to speculate on the mechanisms responsible for the potentiation of twitch response that we observed with repetitive stimulation, we believe our results have several practical implications. First, the staircase phenomenon does not seem to effect the evoked TOF fade ratio (fig. 1). Second, in view of the similarities between our results and those of Lee et al. , 14the staircase effect seems to be equally applicable to acceleromyographic as well as mechanomyographic monitoring techniques. We are not aware of any evidence to indicate that tetanic preconditioning will have any utility when electromyographic monitoring is used. As noted by Krarup, 2repetitive stimuli do not increase the size of the compound action potential. In addition, current evidence suggests that the compound evoked action potential usually decreases by 20–25% during the first half hour of anesthesia despite repetitive stimulation. 15 

Finally, the magnitude of twitch augmentation seems to be directly related to the frequency of stimulation. Sixty stimuli per minute (1 Hz) produces a larger effect than does 16 stimuli per minute (TOF every 15 s), which in turn produces a greater effect than six stimuli per minute (0.10 Hz). Therefore, when stability at any given stimulus frequency is established, any alteration in this pattern results in a baseline shift from “control.”

It is common for investigators to allow only 1–3 min to elapse between initial transducer calibration and drug administration. If the magnitude of the staircase effect or other sources of signal “drift” are not appreciated by the investigator, it may be difficult for the reader to evaluate an author’s final results. To give one example, Miguel et al.  16recently reported a multicenter comparison of the onset/offset characteristics of rapacuronium, mivacurium, and succinylcholine. At two sites, electromyographic monitoring was used; at two other sites, mechanomyographic recordings were used. The authors state, “. . . after obtaining baseline readings for at least 60 sec . . . patients received study medications.” Let us assume that electromyographic twitch height only returned to 80% of the initial control value, as is common with the monitor used by these authors. Let us also assume that the mechanomyographic values ultimately returned to 140% of control. If the initial twitch height values (just before drug administration) are used to define 100% response, investigators at two different sites studying the same drug might easily obtain dissimilar values for the 25 and 75% recovery intervals where no real difference actually exists. It is for this reason that, when recovery intervals need to be measured, normalization to the final T¹value (at a TOF ratio ≤ 0.85) rather than the initial twitch height is generally recommended. 17 

How then should the investigator who wishes to rapidly establish baseline stability proceed? We agree with the suggestion of Lee et al.  14that a 5-s, 50-Hz tetanus administered before initial twitch calibration considerably shortens the time required to achieve baseline stability. This suggestion applies to acceleromyography as well as to mechanomyography. If time permits, a 2–5 min period of TOF stimulation at 12- to 15-s intervals after tetanic preconditioning should optimally precede final twitch calibration. However, tetanic preconditioning produces what seems to be acceptable baseline reliability, even if this abbreviated stabilization period is omitted, and its use should be standard practice when acceleromyography is used as a research tool. In group 3, the average  value of T¹in the 22.5 min after instrument calibration ranged only between 97 and 105% of control. Nevertheless, it should also be recognized that although tetanic preconditioning reduces individual variability, it does not eliminate it. Even with tetanic preconditioning, final individual T¹values ranged from 87 to 136% of the initial control value after 22.5 min of TOF stimulation.