A convenience sample of 16 young healthy participants (28.2 ± 4.8 years old; 8 females) was recruited from Université Laval student population for this single-day, repeated measures design study. Participants had to be naïve to the task and present no self-reported pain. The exclusion criteria were self-reported symptoms or movement limitations at the lower limb or any neurological impairment that could affect task performance. All participants read and signed a consent form describing the experimental procedure and their involvement in the study. This protocol was approved by the local Ethics Review Board (CIUSSS-CN, #2010-212). The experimental procedures were in accordance with the Declaration of Helsinki.

Participants took part in a 2-h laboratory session. After filling the Waterloo Footedness Questionnaire (WFQ; Elias et al., 1998), they walked at 4 km/h on a motorized treadmill (Biodex Gait Trainer 2) for four periods: a 5-min PRE-BASELINE period to familiarize with treadmill walking and to set individual painful stimulation intensity, a 3-min BASELINE period, where they walked without any stimulation, a 3-min PAIN period with stimulation on every gait cycle, and a 3-min POST-PAIN period with no stimulation. Short rest moments (<30 s) were given between the four walking periods. Participants wore shoes instrumented with pressure-sensitive insoles (Tekscan F-Scan, South Boston, MA, United States) to collect dynamic pressure distribution under the foot and temporal gait parameters. They were instructed to walk on the treadmill normally, and to keep walking as they would normally in the presence of pain.

During the PAIN period, they verbally rated ankle pain intensity every 15 s using a numeric Visual Analog Scale (VAS; range 0–10). Immediately after the PAIN period, they rated their global unpleasantness intensity using a modified numeric VAS incorporating the anchors “not bad at all” and “the most unpleasant imaginable” (Duncan et al., 1989), painful region size by pointing at circles of different diameters (Bennell et al., 2004), and pain location by pointing on a schematic shank and foot chart (Bennell et al., 2004). Upon completion of the treadmill walking test, participants completed the SF-MPQ-2. The SF-MPQ-2 consists of 22 pain descriptors divided into four sub-scales: continuous pain, intermittent pain, predominantly neuropathic pain, and affective (Dworkin et al., 2015).

Real-time heel-contact duration was measured using a pressure-sensitive foot switch located under the right heel. It served as stimulus trigger and stimulus duration control. There was no time delay between HC and actual electrical stimulus onset. A home-made electronic circuit interfaced the foot switch with two Grass s-88 stimulators (Grass Instruments, Quincy, MA, United States) wired in-series for signal generation, and a Digitimer DS7A stimulator (Hertfordshire, United Kingdom) for stimulus delivery to the participants. The painful electrical stimulation consisted of series of five pulses (pulse width: 500 μs; pulse frequency: 200 Hz) delivered in bursts at 30 Hz, for the duration of individual right HCs (Figure 1A). Kendall 2.2 cm² H69P disposable electrodes were placed on the right lateral malleolus and 2 cm further along the distal end of the fibula and used for stimulus delivery. Intensity required to reach 3/10 on the VAS was determined for each participant during the PRE-BASELINE period (steps of 5 mA increased every 10 s) and maintained constant throughout the experiment.

In a subgroup of five participants, a second walking test was performed on a separate day in the presence of non-painful stimulation, to assess the contribution of stimulation distraction on the gait biomechanical parameters. Walking periods duration, order, etc., remained the same except for stimulation intensity that was set at 1.4× perceptual threshold (PT), compared to approximately 3.0× PT for the pain experiment.

F-Scan pressure-sensitive insoles (Tekscan, South Boston, MA, United States) were used to collect dynamic pressure distribution under the foot and temporal gait parameters. Peak foot pressure magnitude and duration in the heel and metatarsal regions (see section “Materials and Methods” and Figure 1B) were quantified for each stride of the BASELINE, PAIN, and POST-PAIN walking periods, bilaterally.

Stride-to-stride duration of the nociceptive electrical stimulation was measured off-line from recordings of the pulse trains using a custom-made program written in MATLAB (Version R2018b, MathWorks, United States). The Tekscan data analysis software (F-Scan Research V7.5; Tekscan, United States) was used to set the ROIs on the pressure data from the insoles around the heels and the metatarsals for each participant (Figure 1B). These two ROIs were selected as they represent two functionally important outputs of movement strategies during gait: HC representing the initial contact at heel strike, weight acceptance and mid-stance phases of the gait cycle, and push-off a key part of the pre-swing phase associated with gait speed (Dean et al., 2000). These functional ROIs will be referred to as heel and metatarsal regions throughout this paper, respectively. Both peak pressure magnitude and contact duration of these ROIs were extracted and analyzed with a custom-made MATLAB program.

To measure the effects of pain on gait adaptations, two complementary analyses were performed; (1) evolution over time for the group data (Fortin et al., 2009; Bertrand-Charette et al., 2021) of the time course (Cumming and Finch, 2005); (2) multi-epochs (see below) for statistical comparisons (see Bertrand-Charette et al., for details).

The group was composed of 16 participants (mean age: 28.2 ± 4.8 years; 8 females; see Table 1 for participants’ characteristics).

The mean stimulus intensity necessary to obtain a phasic painful stimulation of 3/10 on the VAS during gait at the onset of the pain period was 14.4 ± 5.2 mA. While stimulus intensity was maintained fixed throughout the PAIN period, pain intensity perceived by the participants slightly decreased over the first 60 s, and then stabilized at 2.5 ± 0.4/10 (time course shown in Figure 2A). Pain tended to be centered on the right lateral malleolus with a mean diameter of 3.3 ± 1.3 cm. No radiating pain was reported by participants. The mean duration of the nociceptive stimulation was 300.7 ± 59 ms, 112.1 ± 12.1 ms shorter than the total HC phase measured with the insoles (Figure 3A).

Participants rated the general unpleasantness of the stimulation session at 4.3 ± 1.7/10. According to the results of their SF-MPQ-2 (Figure 2B), the electrical stimulus was described as throbbing (n = 10/16), sharp (n = 13/16), tender (n = 10/16), electric-shock (n = 16/16), and tingling (n = 15/16). Other qualities of pain reported were stabbing and numbness (n = 7/16). All perceived qualities of pain are presented in Figure 2B.

As mentioned in section “Materials and Methods,” gait is a complex multi-articular movement resulting in various movement strategies. Therefore, this section will present our results according to the two functionally important movement outputs identified by our ROIs (heel for weight acceptance; metatarsals for push-off), separately for peak pressure and contact duration.

The effect of non-painful electrical stimulation on heel-contact duration was also tested in a subgroup of five participants on a different day. Results show that the non-painful stimulation did not affect HC duration (see Figure 3B).

Participants modified their gait pattern in the presence of phasic nociceptive electrical stimulation. It is important to notice that not only did they modify their gait (HC duration and peak pressure) during the initial phase of the PAIN period (PAIN early), as hypothesized, but they also showed a persisting reduction of their HC peak pressure magnitude that stabilized for the rest of the stimulated period (PAIN late). These two periods will be discussed separately below.

During PAIN early, there was a significant reduction in HC peak pressure and HC contact duration, indicative of an unloading of the painful limb. Even though the stimulation protocol was not designed to generate a complete withdrawal response on the stimulated side (withdrawal caused by higher stimulus intensities such as those presented in Spaich et al. (2004), this unloading suggests that stimulation was painful enough to alter the gait pattern. This effect on the painful side was associated with an increased HC peak pressure on the non-stimulated side, indicative of a dynamic increase in weight transfer to that limb (loading). This bilateral change in motor strategy could be interpreted as a protective response, based on Hodges and Tucker (2011).

After approximatively 25 strides into the PAIN period, HC peak pressure stabilized below the lower CI_95%, i.e., at a significantly lower level than before PAIN period. This pressure level was then maintained stable for the rest of the pain period, i.e., for approximately 125 more strides (Figure 4A). Participants therefore showed a persisting reduction in HC peak pressure on the painful side. HC contact duration continued to tend toward baseline values until the end of the pain period. On the non-stimulated side, both pressure magnitude and contact duration returned to BASELINE values. Together these results suggest that the electrical nociceptive stimulus not only generated acute pain-avoidance gait modifications (as mentioned above), but also lead to the development of a persistent modification in gait pattern for the duration of the pain exposure. It is this modified gait pattern that will enable further studies on sensorimotor control or gait modifications with a well-controlled nociceptive stimulus in the future. The current protocol therefore brings a simple and powerful tool to further our knowledge in this field of research.

During the POST-PAIN period, statistically significant changes were observed, mainly for HC duration on the stimulated side. However, these changes were fairly small compared to the main effect observed during the PAIN period. As an example, on the painful side, an increase of 3.4 ± 0.9% relative to baseline duration is measured, compared to the 12.1 ± 0.9% decrease observed for PAIN early. Such small changes are therefore unlikely to be functionally or clinically meaningful.

Even if participants modified their contact duration to modulate pain duration (similar to what has been seen in Gallina et al. (2021) with task-relevant modulation of perceived pain intensity), they still maintained their unloading of the painful limb (reduced HC peak pressure) to possibly continue to “protect” the painful limb. We suggest that this persistent behavior could represent a maladaptive gait modification, according to Hodges and Tucker (2011). Using nociceptive electrical stimulation is therefore a powerful pain model that mimics possible adaptation to modulate pain as shown in Gallina et al. (2021) and by our results.

Regarding painful area size and location, the electrical stimulation used in our protocol was localized around the stimulation site (3.3 ± 1.3 cm), similar to the result of Gallina et al. (2021). This is a major improvement compared to saline or adenosine injections and ischemic block (Graven-Nielsen et al., 2003), with participants reporting pain in various location below their knee.

As a means of further describing the qualities of the pain generated, previous studies using other experimental pain models simulating lower limb MSK pain (saline injections, ischemic block) have used the McGill Pain Questionnaire (versions 1 or 2). They have reported various qualities from the sensory subgroups of the SF-MPQ-2 including Aching and Throbbing (Bennell et al., 2004; Smith et al., 2020) or Stabbing, Cramping, Burning, Heavy, and Exhausting (Graven-Nielsen et al., 2003).

In the present study, Throbbing, Sharp, Tender, Electric-shock, and Tingling were the most frequently reported pain qualities. Sharp and Electric-shock have been categorized as intermittent pain descriptors, Throbbing and Tender as continuous, and Tingling as predominantly neuropathic (Dworkin et al., 2009). These subcategories were suggested by Dworkin et al. (2009, 2015), for participants suffering from chronic pain and acute low back pain. It has been shown that people suffering from acute pain tend to use sensory subgroups (continuous, intermittent, and predominantly neuropathic descriptors) more frequently compared to participants with chronic pain, that tend to score more frequently the affective descriptors (Reading, 1982). Due to the electrical nature of the pain used in our protocol, it is not surprising to have a higher number of participants reporting Electric-shock and Tingling over Burning or Cramping (frequent with the other models). However, in the present study, participants mainly reported sensory subgroups pain descriptors, and only rarely affective ones. This further supports the fact that the electrical nociceptive stimulation represents a good acute pain model.

This study has some limitations. First, the relatively young adult group and relatively small sample size that was recruited might limit the generalizability of the results. Also, participants walked on a treadmill, which may not be as functional as walking overground, but was necessary for our stimulation setup. Regarding the pain intensity, following the first minute of pain exposure, the perceived pain level stabilized around 2.5/10 (compared to the 3/10 initially reached). This could be partly explained by the electrical nature and parameters of the stimulation, close to what is used for transcutaneous electrical nerve stimulation (TENS), known for its hypoalgesic effects in healthy participants (Chesterton et al., 2002, 2003). Using monophasic square waveform at 300 Hz with an intensity of 14.4 ± 5.2 mA [compared to the 7.4 ± 2.2 mA sinusoidal waveform at 4 Hz in Gallina et al. (2021)] show similarities to usual TENS parameters (Chesterton et al., 2003). Importantly, this 0.5-point reduction in pain score had no major impact on the objectives of the study, that were to study gait modifications in the presence of pain. Now knowing that these stimulation parameters can induce changes in pain intensity, it will be possible to conduct future studies using parameters similar to Gallina et al. (2021). In addition, adjusting pain intensity to reflect the amount of pressure on the heel could be an improvement to even better represent the MSK-like aspect of our protocol. Finally, collecting electromyographic data would’ve made it possible to determine if flexion reflex were elicited following painful stimulation.

This study also has several strengths. It presents an original protocol to elicit pain experimentally that shares characteristics similar to actual MSK pain, in order to study human adaptation to nociceptive stimulation. One of the highlights of this phase-specific pain model is its easily adjustable nature (in terms of phase, duration, and intensity at any moment of the experiment), that allowed being present at a functionally relevant moment of the gait cycle. Another highlight is the longer-lasting gait modifications that quickly stabilized to obtain a robust modified gait pattern during the painful condition. Moreover, this model is non-invasive making it safe and easy to use in many settings and populations. Finally, it is possible to recreate gait adaptations that are found in other validated pain models and actual MSK injuries.