Influence of time since injury and physical activity level on upper limb kinematics and muscle activation during wheelchair propulsion in complete T12/L1 spinal cord injury

Table of Contents

Participants

Eleven participants (7 males, 4 females; age range: 25–63 years) with T12/L1 SCI classified under American Spinal Injury Association (ASIA) Impairment Scale A score (complete injury) were recruited. Participants were recruited through outreach via national networks affiliated with the Korea Spinal Cord Injury Association. To minimize the potential influence of pain on propulsion biomechanics, only individuals who reported no shoulder pain during wheelchair propulsion were eligible for participation. Participants’ anthropometric data were collected prior to the experiment, with seated heights varying between 70 and 90 cm and weights ranging from 40 to 80 kg. PA levels were determined using the International Physical Activity Questionnaire (IPAQ) [3], where participants reported PA at work, in daily life, and in leisure and professional sports. The IPAQ is a widely used self-report instrument designed to assess PA across multiple domains and to categorize individuals into low, moderate, or high activity levels. Although originally developed for the general population, recent studies has supported its use and comparability in populations with SCI [10, 18]. Based on their responses, participants were categorized into three PA levels: Low (inactive or minimal PA required in daily life, equivalent to < 600 MET-minutes/week; n = 4), Moderate (regular engagement in recreational PA or sports, 600–3000 MET-minutes/week; n = 3), and High (high-intensity PA required at work or engagement in professional sports, > 3000 MET-minutes/week; n = 4). Additionally, participants were categorized based on the time since injury: Short (< 10 years; n = 3), Medium (10–25 years; n = 4), and Long (≥ 25 years; n = 4). Although no universally accepted criteria exist for categorizing time since injury in SCI populations, previous studies have adopted time-based classifications guided by clinical reasoning or research objectives [14, 21, 39]. In this study, participants were categorized into three groups: < 10 years, 10–25 years, and ≥ 25 years post-injury. These groupings were selected to align with the study’s objective of examining long-term patterns in wheelchair propulsion strategies within the context of SCI rehabilitation in South Korea. Unlike Western countries where community reintegration typically occurs within several months, individuals with SCI in South Korea often experience prolonged hospitalization—averaging approximately 30 months, and in some cases extending up to 10 years [12, 13, 30]. During this extended period, patients receive little to no wheelchair training. As a result, most individuals begin learning to use a wheelchair only after discharge, often through informal peer support. Accordingly, the < 10-year category in this study represents an early post-discharge phase of wheelchair use and skill development, distinguishing it meaningfully from the longer-term experience captured by the other groups.

Data collection

Participants in this study propelled their wheelchairs using a wheelchair roller ergometer (Wheely-X, Kangsters Inc., Republic of Korea) and participated in two tasks: 1) 15 s of forward propulsion at their maximum speed (MAX), and 2) 30 s of forward propulsion at a self-selected speed (SEL). The MAX task was designed to simulate not just sprinting but also challenging conditions like uphill gradients and rough road surfaces. This approach aimed to observe the propulsion strategies employed by participants under various situational demands. Participants had ample time to acclimatize on the roller ergometer until they indicated they were ready to perform the task. They performed each task three times. The order of the task trials was randomized for each participant. If any participant reported pain or discomfort during the trial, the task was immediately terminated to prevent risk of aggravation.

Participants performed the tasks using their personal wheelchairs, all of which were manual and equipped with 24-inch wheels. To maintain consistency in the experimental setup and minimize the potential impact of wheelchair size variability on the collected data, each wheelchair was adjusted to ensure that the participants’hands were aligned with the center of the wheel when seated upright.

Motion capture system (Optitrack Prime13, NaturalPoint Inc., USA) was used to collect kinematic data of upper limb joint and trunk angles. Twenty-two passive retroreflective markers (18 plug in gait marker set with four tracking markers) were placed on anatomical landmarks on the torso and upper extremities [28, 40], and wireless EMG system (Trigno TM, Delsys Inc., USA) was used to collect unilateral activity of six muscles: biceps brachii (Biceps), triceps brachii (Triceps), anterior deltoid (Ant.Delt.), posterior deltoid (Post.Delt.), upper trapezius (Up.Trap.), and lower trapezius (Low.Trap) (Fig. 1). These muscles were selected based on their primary roles in wheelchair propulsion: the Biceps and Triceps control elbow flexion/extension, the Ant.Delt. and Post.Delt. contribute to shoulder movement during the push and recovery phases, and the Up.Trap. and Low.Trap. muscles are involved in scapular stability and trunk control during propulsion. Additionally, the whole experimental process was video-recorded from sagittal and frontal plane with iPhone 13 for propulsion velocity analysis.

Data analysis

Marker trajectories were processed in Optitrack software and modeled with Visual3D (C-motion, USA) software to acquire six upper limb joint and trunk angles: shoulder (referred as sho_sag in Fig. 2), elbow, wrist, and trunk angles in the sagittal plane (Fig. 2(a)); shoulder (referred as sho_fro in Fig. 2) in the frontal plane (Fig. 2(b)).

The cycles were segmented at the point of hand contact with the wheel and then normalized on a scale from 0 to 100%. From the total cycles recorded, those associated with starting and stopping strokes were excluded to negate the effects of acceleration, and the central five cycles were selected for detailed analysis. The maximum and minimum peak angles for the shoulder, elbow, wrist joints, and trunk were identified. Subsequently, the observed range of joint angle (ORJA) was determined by calculating the difference between these maximum and minimum angles. We analyzed the propulsion strategy for each task with five indicators from the peak joint angle analysis. These indicators were selected to capture both the absolute kinematic characteristics during each task and the relative changes in movement patterns between maximum and self-selected speeds, which can suggest adaptive strategies employed by wheelchair users. The five indicators are as follows:

The ORJA during a wheelchair propulsion cycle in the MAX task (ORJA_MAX),
The ORJA in the SEL task (ORJA_SEL),
The difference between ORJA_MAX and ORJA_SEL (MAX-SEL ORJA),
The difference between the maximum joint angle positions for MAX and SEL (MAX-SEL Peak), and
The difference between the minimum joint angle positions for MAX and SEL (MAX-SEL Valley).

The EMG data was subjected to a bandpass filter, incorporating low-pass filtering at 450 Hz and high-pass filtering at 20 Hz, to minimize noise interference. For normalization within a dynamic movement cycle (DMC), sinusoidal rectification was implemented over a 0.01-s interval to accurately identify the maximum EMG value. The normalization process was then carried out using the following formula:

$${\%DMC }= {\text{EMG}}_{\text{Cycle}} / {\text{EMG}}_{\text{DMC}}$$

where EMG_Cycle represents the average of maximum EMG value of five cycles each, and EMG_DMC denotes the maximum EMG value recorded during the propulsion phase. We analyzed the propulsion strategy for each task with three indicators. These EMG indicators were selected to quantify muscle activation patterns during different propulsion intensities and to examine how participants modulate their muscle recruitment strategies when transitioning between self-selected and maximum speed conditions. The three indicators are as follows:

The average of the peak %DMC EMG amplitudes during a wheelchair propulsion cycle in the MAX task (%DMC_MAX),
The average of the peak %DMC EMG amplitudes in the SEL task (%DMC_SEL), and
The difference between %DMC_MAX and %DMC_SEL, (MAX-SEL %DMC).

The collected data met the conditions of normality and homogeneity of variances, thus a One-Way analysis of variance (ANOVA) followed by a Tukey post hoc test was performed to determine the effect of each of the time since injury and PA level factors on the propulsion strategy indicators.

In addition, to measure the propulsion velocity of the five extracted cycles, we calculated the revolutions per minute (RPM) of a 24-inch wheel from the recorded video. The collected data met the conditions of normality and homogeneity of variances, thus a One-Way analysis of variance (ANOVA) followed by a Tukey post hoc test was performed to determine the effect of each of the time since injury and PA level factors on the propulsion strategy indicators.

In addition, to measure the propulsion velocity of the five extracted cycles, we calculated the RPM of a 24-inch wheel from the recorded video.

link