Optimal neuromuscular electrical stimulation parameters after spinal cord injury

Abstract

Neuromuscular electrical stimulation (NMES) is often used to activate muscles impaired after spinal cord injury to elicit functional activities or to facilitate exercise. However, in addition to the cost and availability of NMES and the inherent muscle fatigue that is associated with its use may limit its widespread utilization. Optimizing stimulation parameters during NMES-induced contractions could maximize force production with less fatigue.

Purpose: To examine the interrelationship of pulse duration and pulse frequency on torque production and muscle fatigue in both impaired and non-impaired skeletal muscle of men and women.

Methods: Individuals with [n = 14 (6 females), 38 ± 13 yr; 175 ± 11 cm; 76 ± 20 kg] and without [n = 14 (6 females), 29 ± 8 yr; 175 ± 9 cm; 74 ± 14 kg] spinal cord injury (SCI) participated. Muscle torque was recorded during a series of NMES-induced isometric muscle contractions using different combinations of pulse durations and frequencies. Additionally, two different muscle fatigue protocols (20 and 50 Hz/200µs) were utilized to elicit repeat isometric muscle contractions (1s on and 1s off × 3 min).

Results: There was a statistically significant linear trend for pulse charge (the product of pulse frequency and pulse duration) on isometric torque production in participants without (p < 0.001, η²= 0.79), and in participants with SCI (p < 0.001, η²= 0.66), with higher total pulse charge generating higher torque values. Participants with SCI had significantly greater muscle fatigue for both muscle fatigue protocols (p < 0.05).

Conclusions: NMES protocols should consider using longer pulse durations with lower frequencies to maximize force production for individuals with SCI. However, because mechanisms of muscle fatigue may be different for impaired muscle when compared to non-impaired muscle, further studies on protocols to offset fatigue are warranted.

Introduction

There are approximately 300,000 people in the United States affected by spinal cord injury (SCI) and many more throughout the world¹. After SCI, it is well documented that affected skeletal muscles atrophy and become highly fatigable,^2–4 causing a reduced ability to produce and maintain force. Neuromuscular electrical stimulation (NMES) has been utilized to activate muscles impaired after SCI to produce functional movements to increase physical activity.^5–7 However, there are fundamental barriers associated with using NMES for the activation of skeletal muscles that preclude its widespread use, such as cost and availability of equipment.⁸ Additionally, the inherent muscle fatigue that has been associated with NMES is also a limiting factor⁹. For these reasons, people with SCI do not often incorporate NMES into activities that may improve their health or function.

Physical activity and exercise are essential to maintain fitness of cardiovascular and musculoskeletal systems, with current guidelines prescribing aerobic and resistance training exercise at least twice per week for people with SCI.¹⁰ NMES has shown great promise in providing people with SCI viable options for exercise that can improve their musculoskeletal system and overall fitness levels.¹¹ Studies have shown benefits when utilizing electrical stimulation to assist with aerobic (e.g. cycling, rowing) and resistance training activities in people with SCI.^11,12 A recent review concluded that electrical stimulation is an effective strategy to induce muscle hypertrophy and improve lean mass.¹³ However, the programs are often limited by expensive equipment and non-optimized NMES protocols to yield maximal benefit for participants with SCI.¹⁴

Due to the increased muscle fatigue that is known to occur with NMES-induced contractions, investigators have studied electrical stimulation parameters that impact muscle torque production to better understand their physiological consequences.^15–20 Numerous combinations of electrical stimulation parameters can be altered to impact force levels, some of which also impact muscle fatigue. Several investigators have reported that lower frequency of stimulation, in combination with longer pulse durations, maximizes performance during repetitive stimulation.^21,22 In addition, we have reported that the product of pulse duration and pulse frequency, defined as total pulse charge, is a good predictor of muscle torque production.^20,23 When comparing NMES-induced contractions with stimulation parameters that have equal total pulse charge, those with lower frequencies result in less fatigue.^20,23 These studies demonstrated the negative consequence of higher frequencies and shorter pulse durations as a contributor to fatigability, however, none of these studies included participants with SCI, a population that could substantially benefit from improved NMES protocols.^20,23

Several studies have been conducted to develop optimal electrical stimulation parameters to reduce fatigue in individuals without sensorimotor impairment,^24,25, but few on impaired skeletal muscle after SCI.^26–28 Most importantly, protocols that are developed in individuals without SCI do not necessarily translate to people with SCI.²⁷ Recently, Ma et al.²⁹ stated that future studies on NMES in people with SCI should focus on optimizing stimulation parameters and we are only aware of a small study of people with SCI (n = 6) that used two different stimulation paradigms.²⁶ A recent review on the impact of electrical stimulation on body composition after SCI concluded that “future studies in people with SCI are recommended to compare between different ranges of parameters”.¹³ Therefore, the purpose of this study was to examine the interrelationship of multiple different combinations of pulse duration and pulse frequency on muscle torque production and fatigue in people with and without SCI.

Materials and methods

Participants. Twenty-eight individuals participated: 14 individuals with SCI and 14 without SCI completed this study. Additional characteristics about the participants are in Table 1. Criteria for participation included: (1) 19–60 years of age, (2) no known medical conditions that would result in a contraindication to receive NMES, (3) people with SCI were non-ambulatory and had motor impairment of the quadriceps femoris muscle (m. QF).

Table 1.

Participant characteristics.

Health Status	N	Age (yrs)	Height (cm)	Weight (kgs)	Sex	SCI status	AIS
Participants without SCI	14	29 ± 8	175 ± 9	74 ± 14	8 male	Not applicable	No known neuromuscular deficits
					6 female
Participants with SCI	14	38 ± 13	175 ± 11	76 ± 20	8 male	4 Tetraplegia	11 A
					6 female	8 Paraplegia	3 B

Research design : We utilized a within-subject experimental design approach to determine the effects of manipulating electrical stimulation parameters on muscle torque production and fatigue.

Procedures : Prior to participating in the study, written informed consent was obtained from all participants, as approved by the Institutional Review Board at the University of Alabama at Birmingham. Briefly, both the left and right m. QF of the participants without SCI were tested at separate times in a single visit using the following tests: maximum voluntary isometric contractions (MVIC), followed by NMES parameter testing on m. QF torque production, and then a fatigue test. The MVIC test was used to determine the stimulus amplitude required to produce 25% of the MVIC which was used during the NMES parameter test on m. QF torque production and the fatigue test. Participant with SCI underwent the same procedures as the participants without SCI with the exception of MVIC testing. The stimulus amplitude used for the NMES parameter test on m. QF torque production and fatigue test in individuals with SCI were determined by using the amplitude that produced a maximum stimulus force up to 20Nm torque. To minimize the effect of muscle fatigue due to repeat testing, we included a minimum 30s of rest between contractions for both groups.

Torque measurements: The m. QF was studied during electrically stimulated isometric contractions as described previously.^20,30,31 Participants were seated in a custom-built chair with the hip and knee secured at approximately 90° of flexion. A single leg was firmly secured to a rigid lever arm to ensure that m. QF would perform only isometric contractions. The moment arm was established by placing a load cell (Transducer Techniques, USA) parallel to the line of pull and perpendicular to the lever arm. Prior to data collection, participants without SCI were allowed to perform several m. QF warm-up contractions. Next, a value for MVIC of the m. QF for each leg was determined. MVIC was defined as the peak isometric torque achieved during 3 consecutive maximal efforts (∼5s contraction separated by 120s of rest). If the peak torque values differed by more than 5%, additional trials were conducted. Current intensity for subsequent NMES testing was determined by increasing stimulation amplitude to elicit 25% of the MVIC for each participant without SCI. For example, if a participant’s MVIC was 100 Nm, the investigators would increase the current intensity until 25 Nm of knee extension isometric torque was achieved and this value (mA) was used for all subsequent NMES-elicited contractions. We have found that 25% of MVIC is sufficient to answer questions related to muscle performance and is generally well-tolerated by those with sensation.²⁰ MVIC could not be tested in participants with SCI due to the lack of voluntary motor control of the m. QF. Thus, we utilized the NMES amplitude that produced the maximum stimulated force (using 50 Hz and 600µs) up to 20 Nm per m. QF to minimize risk of injury. For example, if we increased current amplitude and the force did not concomitantly increase or we reached 20 Nm, this value (mA) was used for all subsequent NMES-elicited contractions in participants with SCI. Other groups have used similar stimulation intensities when comparing protocols between people with and without SCI.²⁸

Electrical Stimulation: Bipolar self-adhesive neuromuscular stimulation electrodes (7.5 cm x 10 cm) were placed over the distal-medial and proximal-lateral portion of the m. QF group.³¹ Monophasic square-wave stimulation pulses were delivered using a Grass S88 stimulator with a Grass Model SIU8 T stimulus isolation unit (Grass Technologies, West Warwick, RI). The intensity of stimulation to elicit ∼25% of the participants without SCI MVIC was determined using a monophasic pulsed current with the following parameters: 50 Hz and 600µs pulse train of 500 ms duration. We utilized a relatively high frequency (50 Hz) and long pulse duration (600µs) to elicit initial force, knowing we were going to lower these parameters during the subsequent testing protocols. Following determination of desired stimulation intensity, 6 stimulation trains (150 total pulses) were delivered at the aforementioned settings to ensure potentiation of the m. QF group, as done previously.³² Immediately following determination of stimulation intensity and potentiation, the torque elicited using different combinations of frequencies and pulse durations was measured on a single m. QF (parameter testing). We delivered, in random order, eight different combinations of frequencies (20, 30, 40, and 60 Hz) and pulse durations (200, 300, 400, and 600 µs) at the pre-determined current amplitude level. Specifically, the eight combinations are: 20 Hz and 400µs, 20 Hz and 600µs, 30 Hz and 600µs, 40 Hz and 600µs, 40 Hz and 200µs, 60 Hz and 200µs, 60 Hz and 300µs, 60 Hz and 400µs. We limited the parameter testing to these combinations and provided a minimum of 30s of rest between contractions based on our prior studies and to limit the number of contractions for participants in an effort to minimize fatigue.^32,33 We used the exact procedure for participants with SCI but the current amplitude was kept at the maximum electrically induced force for each individual with SCI up to 20Nm.

Fatigue Protocols: Fatigue protocols were conducted using an initial starting force equal to ∼25% of each participant’s MVIC, or the maximum torque evoked from participants with SCI (up to 20Nm). The 25% of MVIC intensity was selected because it allows for recruitment of a sufficient number of motor units in the m. QF is generally well tolerated by participants without SCI.²⁰ A single fatigue test was performed on each leg, using different parameter settings of equal total charge (50 Hz and 200µs or 20 Hz and 500µs) at the same amplitude used previously. Contractions were 1-s long with 1-s rest between contractions for three min or 180-s resulting in 90 total contractions, essentially as done previously.³² It should be noted that while there are different relative intensity levels, the motor units that are being activated between those with and without SCI are equivalent because they use a consistent amplitude, frequency, and pulse duration throughout the fatigue protocol.

Data Processing: Torque data were analyzed using a commercially available software package (Chart v5.0, ADInstruments). Torque values obtained from the eight combinations of frequencies (20, 30, 40, and 60 Hz) and pulse durations (200, 300, 400, and 600 µs) at the pre-determined amplitude level, resulted in 8 variables (repeated measurement occasions) of pulse charge. The combination produced two settings of equal total pulse charges with four levels: 24,000 µs–Hz (24k), 18,000 µs–Hz (18k), 12,000 µs–Hz (12k) and 8000 µs–Hz (8k). The parameter combinations for set A were: 20 Hz and 400 µs (8k), 20 Hz and 600 µs (12k), 30 Hz and 600 µs (18k), and 40 Hz and 600 µs (24k). The parameter combinations for set B were: 40 Hz and 200 µs (8k), 60 Hz and 200 µs (12k), 60 Hz and 300 µs (18k), and 60 Hz and 400 µs (24k) (Table 2). We selected these parameters in an effort to match total pulse charge between Set A and B using different combinations of frequency and pulse duration. We analyzed each setting of the same pulse charge separately. Fatigue scores were obtained from two different parameter settings of equal total charge (50 Hz and 200µs and 20 Hz and 500µs) at the same amplitude used previously. Muscle fatigue was operationalized as the percent drop in muscle torque production from contraction one (which equals to 1 after normalization) to the last contraction during the fatigue tests.

Table 2.

Set A and Set B of electrical stimulation parameters.

	Set A		Set B
Total pulse charge (Hz–µsec)	Frequency (Hz)	Pulse duration (µsec)	Frequency (Hz)	Pulse duration (µsec)
8000	20	400	40	200
12,000	20	600	60	200
18,000	30	600	60	300
24,000	40	600	60	400

Data Analyses: Both graphical (histogram and Q-Q plot) and statistical (Shapiro–Wilk tests, skewness and kurtosis) methods were used to determine whether the data set (torque and fatigue scores) was modeled for normal distribution. Only 10% of the data were considered not normally distributed, justifying the use of parametric statistical tests for further analysis.³⁴ A three-way repeated measures analysis of variance (ANOVA) was performed to determine the effects of two between-subject factors and one within-subject factor on torque production. The two between-subject factors were SCI (with or without) and sex (male or female), and the within-subject factor was total pulse charge (it had four levels). Trend analyses were conducted to test for various trends of muscle torque production across four levels of total pulse charge. A repeated measure ANOVA was used to determine if differences existed between torque values elicited by a series of same total charge yet varying parameter settings (A and B) within the participant groups. Another three-way repeated measures ANOVA was performed to determine the effects of two between-subject factors (with/without SCI and sex) and one within-subject factor (same pulse charge of two different parameter settings) on muscle fatigue.

When the Mauchly’s test statistic is significant (i.e. sphericity assumption has been violated with unequal variances) in repeated measures ANOVAs, a correction factor (Greenhouse-Geisser or Huynh-Feldt estimate) to the degrees of freedom of the F-distribution was applied to produce a more valid critical F-value with a corrected Type I error rate. Data were analyzed by SPSS v28 (SPSS Inc., Chicago, IL), and the level of significance was set at α = 0.05. The effect size was reported using partial eta squared (partial η²). The magnitude of effect size was reported as small (0.01–0.05), moderate (0.06–0.13) and large (≥0.14) ³⁵ .

Results

The values for muscle torque production from different parameter combinations in one female participant without SCI and the muscle fatigue scores from one parameter combination in one male participant with SCI were missing.

Effects of different parameter combinations with equal total pulse charges on muscle torque production for set A:

Mauchly’s test indicated that the assumption of sphericity for total pulse charge had been violated, χ²(5) = 38.07, p < 0.001, therefore the degrees of freedom were corrected using Greenhouse-Geisser estimates of sphericity (epsilon, ϵ = 0.49). There was no interaction between total pulse charge and sex [F(1.5, 33.9) = 1.32, P = 0.27, partial η² = 0.05], suggesting that the profiles of the four repeated measurement occasions on torque production are parallel across sex. There was no evidence of an interaction between health condition (with or without SCI) and sex [F(1,23) = 0.04, P = 0.84, partial η² = 0.002]. In addition, there was no main effect for sex [F(1,23) = 1.43, P = 0.25, partial η² = 0.06], suggesting that any differences for males and females on the torque production was due to sampling error. Hence, sex was excluded from the model.

The interaction between total pulse charge and health condition were statistically significant [F(1.5, 33.9) = 23.38, p < 0.001, partial η² = 0.5], and indicated that the variation in the means of the pulse charge over the four repeated measurement occasions on torque production varied as a function of health condition group. The main effect of health condition group on the average total pulse charge across repeated measurement is statistically significant on torque production, [F(1,23) = 95.9, p < 0.001, partial η² = 0.81], with the mean torque production from persons without SCI being significantly higher than that of people with SCI at each level / occasion (after adjusting for multiple comparisons using Bonferroni correction, all p values <0.001). Effect sizes (partial eta squared, η²) of the health condition indicated that its effect on torque production was substantial.

Polynomial contrasts indicated a statistically significant linear trend for pulse charge (the product of pulse frequency and pulse duration) of 24k, 18k, 12k and 8k cycles (µs - Hz) on muscle torque production in participants without SCI (F(1,12) = 44.12, p < 0.001, η²= 0.79), and in participants with SCI (F(1,13) = 25.19, p < 0.001, η²= 0.66), with higher total pulse charge generating higher torque values (see Figure 1).

Figure 1.

Torque values for set A and set B for individuals without (upper panel) and with spinal cord injury (lower panel). Significant linear trend for pulse charge on muscle torque production for both sets in both participant groups (P < 0.01).

Effects of different parameter combinations with equal total pulse charges on muscle torque production for set B:

Since Mauchly’s test indicated that the assumption of sphericity for total pulse charge had been violated, χ²(5) = 73.90, p < 0.001, the degrees of freedom were corrected using Greenhouse-Geisser estimates of sphericity (epsilon, ϵ = 0.41). There was no interaction between total pulse charge and sex [F(1.2, 28.0) = 0.07, P = 0.84, partial η² = 0.003], suggesting that the profiles of the four repeated measurement occasions on torque production were parallel across sex. There was no evidence of an interaction between health condition and sex [F(1,23) = 0.01, P = 0.91, partial η² = 0.001]. In addition, there was no main effect for sex [F(1,23) = 1.54, P = 0.23, partial η² = 0.06], suggesting that any differences for males and females on the torque production was due to sampling error. Hence, sex was excluded from the model.

The interaction between total pulse charge and health condition were statistically significant [F(1.2, 28.0) = 34.87, p < 0.001, partial η² = 0.6], and indicated that the variation in the means of the pulse charge over the four repeated measurement occasions on torque production varied as a function of health condition group. The main effect of health condition group on the average total pulse charge across repeated measurement is statistically significant on torque production, [F(1,23) = 79.8, p < 0.001, partial η² = 0.78], with the mean torque production from persons without SCI being significantly higher than that of people with SCI at each level / occasion (after adjusting for multiple comparisons using Bonferroni correction, all p values <0.001). Effect sizes (partial eta squared, η²) of the health condition indicated that its effect on torque production was substantial.

Polynomial contrasts indicated a statistically significant linear trend for pulse charge (the product of pulse frequency and pulse duration) of 24k, 18k, 12k and 8k cycles (µs - Hz) on muscle torque production in participants without SCI (F (1,12) = 70.81, p < 0.001, η²= 0.86), and in participants with SCI (F (1,13) = 11.85, p < 0.004, η²= 0.48), with higher total pulse charge generating higher torque values (see Figure 1). A quadratic trend (F (1,12) = 48.18, p < 0.001, η²= 0.80) for pulse charge on torque production in participants without SCI was also observed.

Comparing muscle torque production between parameters of equal charge between two sets

For a given total pulse charge, a combination of longer pulse duration and lower frequency induced a greater average torque production than a combination of shorter pulse duration and higher frequency in participants without SCI when the total pulse charge was below the 24k (µs–Hz) (P = 0.10), with the p values < 0.002 on each of the three remaining parameter combinations in the two settings of total pulse charges. However, for individuals with SCI, no significant difference in muscle torque production between each of the four parameter combinations in these two settings of total pulse charges after adjusting for multiple comparisons using Bonferroni correction, with all four p values slightly larger than 0.05.

Effects of different parameter combinations with equal total charge on muscle fatigue

There was no interaction between fatigue and sex [F(1,23) = 2.67, P = 0.12, partial η² = 0.1], suggesting that the profiles of the two repeated measurement occasions on fatigue were parallel across sex. There was no evidence of an interaction between health condition and sex [F(1,23) = 2.01, P = 0.17, partial η² = 0.08]. In addition, there was no main effect for sex [F(1,23) = 0.01, P = 0.92, partial η² = 0.00], suggesting that any differences for males and females on muscle fatigue was due to sampling error. Hence, sex was excluded from the model.

There was no interaction between fatigue and health condition [F(1,23) = 0.8, P = 0.38, partial η² = 0.03], suggesting that the profiles of the two repeated measurement occasions on fatigue were parallel across health condition. The main effect of health condition group on the average total pulse charge across repeated measurement is statistically significant on fatigue, [F(1,23) = 22.09, p < 0.001, partial η² = 0.49], with the mean fatigue from persons without SCI being significantly lower than that of people with SCI at each level / occasion (after adjusting for multiple comparisons using Bonferroni correction, all p values <0.001). Effect sizes (partial eta squared, η²) of the health condition indicated that its effect on fatigue was substantial.

Both groups demonstrated a significant decline in torque during the fatigue tests, when parameter combinations of 20 Hz and 500 µs were used compared to 50 Hz and 200 µs; however, they were not significantly different (Without SCI: P = 0.07 and with SCI: P = 0.12). At 50 Hz and 200 µs, the mean ± SD of the decline in muscle torque production was 58.0 ± 19.1% for participants without SCI and 78.8 ± 11.6% for participants with SCI; whereas, at 20 Hz and 500 µs, the mean ± SD of muscle fatigue was 48.0 ± 8.8% for participants without SCI and 72.8 ± 15.2% for participants with SCI. Because the muscle fatigue scores for the two parameter combination variables (20 Hz and 500 µs for participants with SCI, and 50 Hz and 200 µs for participants without SCI) were considered not normally distributed, a sensitivity analysis using non-parametric statistical test was conducted. Results of the Wilcoxon signed ranks tests (after adjusting for multiple comparisons using Bonferroni correction) were congruent with that of the paired test on the muscle fatigue scores between the two parameter combinations. Individuals with SCI had significantly greater declines in torque as compared to individuals without SCI during the fatigue protocol regardless of the frequency of activation (p < 0.002).

Discussion

The novel aspect of this study is that the skeletal muscle of individuals with or without SCI responds in a similar manner to altering the parameters of NMES. In this study, we compared the muscle torque production from different combinations of stimulation parameters that had equal total charge and it was determined that torque increases as total charge increases. We have previously demonstrated this in individuals without SCI³², but this was the first study to determine that impaired muscle after SCI responds similarly despite the absolute torque production being significantly lower for SCI due to the significant muscle impairments such as muscle atrophy, conversion to a fast-fatigable fiber types, and reduced ability to generate ATP that are known to occur.^2,36,37 The muscle of individuals with SCI also showed significantly greater muscle fatigue as compared to individuals without SCI regardless of frequency of stimulation.

As indicated in both set A and set B, increasing total pulse charge results in increased torque production. Longer pulse duration and higher frequencies (up to a point) can each independently produce a larger torque. However, when comparing set A and B, even with the same total pulse charge, pulse duration seemed to have a larger impact on torque production than pulse frequency. For example, when looking at parameters that have the same total pulse charge, a longer pulse duration with a lower frequency produces a larger torque than a higher frequency with a shorter pulse duration. This disparity is more obvious when the pulse duration is at the low end (such 200 µs), and the frequency is at the higher end (such as 60 Hz), compared to when the pulse duration is at the higher end (600 µs) and the frequency is at the lower end (such as 20 Hz); when both settings produce the same total pulse charge of 12k.

We suspect the reason that pulse duration can have a greater impact on torque production is because the increase in torque when pulse duration increases is due to the recruitment of additional motor units. Whereas the mechanism for increasing torque due to an increase in pulse frequency is through the summation of actional potentials to get additional torque from motor units that have already been recruited. Thus, considering we see that the difference is more obvious for pulse duration at the lower end and frequency at the higher end, adding additional motor units that are contracting at a tetanic level is going to have a greater impact on force production than adding motor units at the sub-tetanic level. Some studies have suggested that a long pulse duration recruits additional preferentially fast motor units¹⁸ but others have found the opposite, suggesting longer pulse durations result in more of the smaller, fatigue-resistant motor units.³⁸ We suggest that the differences in adding additional motor units are likely not due to a difference in motor unit type (fast vs. slow) because it is widely accepted that electrical stimulation recruits motor units in a nonselective, spatially fixed, and temporally synchronous pattern.⁹

We were surprised that frequency did not have a significant effect on muscle fatigue as one would expect.^20,31 Our fatigue tests utilized a lower frequency (20 Hz and 500 µs) protocol that most would have expected to demonstrate significantly less fatigue than the higher frequency (50 Hz and 200 µs) protocol, however, our findings were not statistically significant. We suspect this was a statistical power issue in the group without SCI because the percent declines in torque were 48% and 58% for the 20 and 50 Hz protocols, respectively. However, it is less clear as to why the individuals with SCI had fatigue levels that were similar at 72% and 78% declines for the 20 and 50 Hz protocols, respectively. This could be that muscle fatigue in response to a 3-minute stimulation protocol is likely to due to muscle metabolism in the participants without SCI but multi-factorial for the SCI group.³³ Additionally, in this study, we maintained an equal pulse charge despite differences in both pulse duration and frequency. Previous studies have typically only adjusted the frequency while maintaining a constant pulse duration. We expected that lower frequency would result in less fatigue but cannot be certain how maintaining pulse charge may impact fatigue.

Much of the work that has been done on the topic of optimizing electrical stimulation parameters to improve muscle performance has been done in participants without SCI. More specifically, there have been several studies that have investigated the influence of altering pulse duration and pulse frequency on muscle torque output and fatigue.^18,21,32,38 Gonnelli et al. recently compared a short pulse duration / high intensity protocol to a long pulse duration / low intensity protocol.²⁶ They found that both protocols resulted in similar torque outputs and fatigue for individuals with and without SCI. It should be noted that they only had six individuals with SCI but their findings are consistent with our study that had fourteen people with SCI.

Finally, it needs to be noted that over 40% of the participants enrolled into this study were female. There has been considerable work done that demonstrates known sex differences in muscle fatigue of participants without SCI.³⁹ The differences appear to be related to contraction type, with women being less fatigable than men when using isometric contractions similar to what was utilized in this study. We included sex in our preliminary statistical models and found no evidence for a sex effect, so we excluded it from further analyses. However, it should be noted that because of the many differences between the affected skeletal muscle in people with and without SCI that biological sex differences may be more difficult to uncover. Future studies may consider utilizing combinations of both voluntary – and electrically – elicited contractions to determine if sex differences are due to more central vs. peripheral mechanisms.

Limitations: This study had a relatively small sample size (n = 28, 14 each for individuals with and without SCI). It is challenging to recruit large samples of individuals with SCI at a single site. However, we, and others, have conducted similar studies that investigated the impact of parameters of stimulation using samples of similar size.^20,32,40 This study also had a high degree of variability between participants including age in both groups and duration of injury and level of injury in individuals with SCI. The small sample size coupled with the high variability would negatively impact the statistical power for some comparisons. However, this variability may allow greater generalization. Finally, another point to consider is that we utilized isometric contractions in this study. While there is evidence to support using isometric contractions in people with SCI^37,41 many clinical applications of NMES for people with SCI utilize dynamic muscle contractions. Therefore, the generalizability of the findings of our study must be considered with caution until further investigated during dynamic activities.

In conclusion, this study supports using longer pulse durations and lower frequencies to elicit greater force production than higher frequencies and short pulse durations. This would allow for greater motor unit recruitment and possibly less fatigue, thus leading to improved performance. It appears that the skeletal muscle after SCI responds in the same manner as non-impaired muscle despite not being able to generate as much torque with NMES.

Disclaimer statements

Contributors None.

Conflicts of interest The authors declare no conflict of interest.

Funding This work was supported by the National Institutes of Health 5R03 HD057388.

Ethics approval Ethical approval was granted by the University Institutional Review Board.