A Role for Trunk Function in Elite Recumbent Handcycling Performance?

ABSTRACT

Handcycling classification considers trunk function, but there is limited scientific evidence of trunk involvement in recumbent performance. This study investigated the association between trunk function and recumbent handcycling performance of athletes without upper-limb impairments (H3-H4 sport classes). The study was divided into two parts. First, 528 time-trial results from 81 handcyclists with spinal cord injury (SCI) were obtained between 2014 and 2020. Average time-trial velocity was used as performance measure and SCI level as trunk function determinant. Multilevel regression analysis was performed to analyse differences in performance among SCI groups while correcting for lesion completeness, sex, and age. Second, in 26 handcyclists, standardised trunk flexion strength was measured with a handheld dynamometer. Peak and mean power-output from a sprint test and time-trial average velocity were used as performance measures. Spearman correlations were conducted to investigate the association between trunk strength and performance. Results showed that the different SCI groups did not exhibit significant differences in performance. Furthermore, trunk flexion strength and performance exhibited non-significant weak to moderate correlations (for time-trial speed: rs = 0.36; p = 0.07). Results of both analyses suggest that trunk flexion strength does not seem to significantly impact recumbent handcycling performance in athletes without upper-limb impairments.

KEYWORDS:

• Para-cycling
• paralympic classification
• trunk strength
• average velocity
• time trial
• sprint test

Introduction

Handcycling use varies from rehabilitation programmes, to a mode of transportation, recreational sport and elite competition (Hettinga et al., 2010). Handcycling has been a Paralympic sport since 2004 with more than 680 international athletes worldwide (Union Cycliste Internationale, 2020a). Handcyclists compete in two outdoor racing events, namely road races and time-trials. Road race events range from 37 km to 70 km for women, and from 45 km to 80 km for men. Time-trial events range from 10 km to 30 km for women, and from 12 km to 35 km for men (Union Cycliste Internationale, 2020b). As a Paralympic sport, handcycling must comply with the International Paralympic Committee (IPC) Athlete Classification Code, where it is stated that all Paralympic sports must be centred on evidence-based classification systems designed to minimise the impact of impairment on the performance (International Paralympic Committee, 2015^; Tweedy & Vanlandewijck, 2011). By minimising the impact of impairment, differences in performance between athletes from the same class are likely to be a result of genetic predisposition, talent and training intervention. From this classification perspective, athletes who augment their performance will not be disadvantaged by being moved towards a less impaired class or vice-versa (Tweedy & Vanlandewijck, 2011).

Handcycling is governed by the Union Cycliste Internationale (UCI) and its classification is based on a hierarchical system of five sport classes (H1-H5) since 2014. The H1-H4 classes compete in an arm-powered handbike in a recumbent position, with athletes in class H1 representing the athletes with the greatest impairments, e.g., athletes with lower-limb, trunk, and upper-limb impairments (Union Cycliste Internationale, 2020b). In contrast to the H1–H4 classes, the H5 class competes in an arm-trunk powered handbike in a kneeling position and represents athletes with the least impairments. These sports class profiles are described in a classification manual (Union Cycliste Internationale, 2020b), historically centred on people with spinal cord injury (SCI) as a reference of the functional limitation for comparison with other impairments. By comparing these class profiles, it is possible to identify the main physical determinants that characterise each one of them. In summary, the main characteristic that divides H1 and H2 is arm strength, H2 and H3 is hand strength and trunk function (no vs. minimal to limited), and H3 and H4 is trunk strength. H5 differs from H4 on full trunk and hip-leg function plus the ability to compete in an arm-trunk powered handbike (Union Cycliste Internationale, 2020b).

The involvement of the trunk in sports performance has received general interest from a diverse range of Paralympic sports over the last two decades. A higher level of SCI has been associated with decreased trunk active range of motion, slower postural reactions and lower acceleration in para-sports such as sit-skiing, wheelchair athletics and wheelchair rugby, with differences being smaller when comparing individuals with partial to full trunk function (Altmann et al., 2016, 2017; Connick et al., 2017; Rosso et al., 2019; Vanlandewijck, Verellen, Beckman et al., 2011; Vanlandewijck, Verellen, Tweedy et al., 2011). These sports suggest a direct trunk influence on performance given the upright position, e.g., balance in sit skiing and wheelchair manoeuverability in wheelchair rugby. In the case of a recumbent aerodynamic position as H1–H4 handcycling position (Mannion et al., 2018), research on the influence of trunk impairment is scarce. Recumbent positions with an upright backrest may evoke a higher contribution of the trunk muscles (Faupin et al., 2008^; Van Der Woude et al., 2000) including a low activation of the upper abdominal (Quittmann et al., 2019) when compared with more aerodynamic positions. However, these recumbent positions seem to evoke small trunk movement in the sagittal, frontal and transverse planes (Faupin et al., 2006; Quittmann et al., 2018; Stone, Mason, Warner, Goosey-Tolfrey et al., 2019; Stone, Mason, Warner, Tolfrey et al., 2019; Verellen et al., 2012a) as it was observed in able-bodied, H3 and H4 athletes. The small trunk movement was maintained throughout different intensities, and inter-individual variations decreased with increasing intensity among elite handcycling athletes (Stone, Mason, Warner, Goosey-Tolfrey et al., 2019^; Stone, Mason, Warner, Tolfrey et al., 2019).

The significance of trunk involvement in recumbent handcycling appears questionable, especially considering highly trained athletes. To the best of our knowledge, the only study investigating the direct association between SCI level and handcycling focused on force generation patterns and effectiveness, and did not find significant differences among groups of athletes with different SCI levels (Verellen et al., 2012b). However, according to the current para-cycling classification manual, athletes with impaired muscle strength are allocated to H3 and H4 classes mainly based on trunk muscle strength and with SCI levels as a reference (Union Cycliste Internationale, 2020b). There is evidence of the negative impact that higher lesion levels have on cardiovascular parameters, and consequently on endurance sports performance, as a result of an impaired autonomic nervous system (ANS) (West et al., 2015, 2014). This impact is predominant in athletes with autonomic complete cervical lesions, but athletes with high thoracic SCI may as well present an impaired ANS (West et al., 2015, 2014). However, in accordance with the IPC Athlete Classification Code (International Paralympic Committee, 2015^; Tweedy & Vanlandewijck, 2011) such physiological parameters are not taken into account during Paralympic classification. As there is no scientific evidence of the association between trunk strength and recumbent handcycling performance, the current classification system does not follow an evidence-based perspective as advocated by IPC.

The main aim of this study was to investigate the association between trunk function and recumbent handcycling performance in athletes from the H3 and H4 sport classes. To study this aim, two complementary analyses were performed. The first analysis aimed to investigate whether handcyclists’ trunk function (based on SCI level) is related to performance, using a retrospective design. The second analysis aimed to investigate the association between trunk flexion strength and handcycling performance, based on a cross-sectional cohort design. It was hypothesised that 1) different levels of trunk function do not present significant differences in recumbent handcycling performance, and 2) that trunk flexion strength is only weakly to moderately associated with recumbent handcycling performance.

Materials and methods

Design

Two complementary analyses were conducted to investigate the association between trunk function and recumbent handcycling performance. In the first analysis, 528 time-trial results of 81 athletes were grouped according to SCI level and multilevel regression analyses were performed to investigate the influence of lesion level on performance. In the second analysis, correlation coefficients were calculated between trunk flexion strength and measures of handcycling performance (time-trial average velocity and isokinetic sprint test) from a sample of 29 athletes.

SCI level & handcycling performance

Participants

Eighty-one elite handcycling athletes from H3 and H4 sport classes (64 males, 17 females) with SCI were included in this study after providing written informed consent to share their para-cycling classification data. All athletes were competing at world cups and world championships. The research was approved by the Scientific and Ethical Review Board of the Faculty of Behavioural and Movement Sciences, Vrije Universiteit Amsterdam, the Netherlands, No: VCWE-2018-093.

Experimental protocol

Demographic information regarding age, sex and event distance and location was available from the official UCI website. Information regarding the athlete’s para-cycling classification and impairment characteristics was provided by the UCI. Based on muscle function and muscle innervation (Beaussier, 2013; McErlean & Migliozzi, 2017; Rozen et al., 2008; Sakamoto et al., 1996), participants were assigned to one of the four groups regarding their SCI level: Th1–Th5, including high thoracic lesions, from Th1 to Th5 lesion level, possibly affecting the chest wall and below, and the sympathetic cardiac innervation; Th6-Th9, injuries affecting the upper abdominal wall and defined as lesions from levels Th6 to Th9; Th10–L1, injuries affecting the lower abdominal wall and defined as lesions from levels Th10 to L1; and last, ≤L2, lesions from L2 down which do not affect the trunk muscles.

The completeness of their lesion was based on the ASIA impairment scale (AIS), i.e., motor complete (AIS A-B) versus motor incomplete (AIS C-D).

The official handcycling time-trial results from 2014 to 2020 of each athlete were obtained retrospectively from the UCI’s website (Union Cycliste Internationale, 2020c) for 24 events, resulting in a total of 528 time-trial results from 81 athletes (Table 1). As a key measure of handcycling performance, average time-trial velocity (m/s) was calculated by dividing the time-trial results by the race distance, following a previously described procedure (Muchaxo et al., 2020).

Table 1. Number of time-trial results included in the analysis per sex and spinal cord injury level group

SCI level groups	Time trials
	Women				Men				Total
	Age^a	AIS A/B^b	AIS C/D^c	Total	Age^a	AIS A/B^b	AIS C/D^c	Total
Th1–Th5 (high thoracic)	41 ± 10	17	0	17	42 ± 12	94	2	96	113
Th6–Th9 (upper abdominals)	33 ± 2	2	19	21	35 ± 6	79	16	95	119
Th10–L1 (lower abdominals)	40 ± 11	84	3	87	40 ± 5	132	57	189	277
≤L2 (full trunk)	26 ± 3	4	5	9	33 ± 11	0	10	10	19
Total		107	27	134		305	85	390	524^d

^aAveraged age and standard deviation of the group

^bAIS A or B, equivalent to complete motor lesion

^cAIS C or D, equivalent to incomplete motor lesion

^dThere are 4 additional participants not included in this table due to missing data regarding AIS. The total sample is representative of 81 unique handcycling athletes (64 males, 17 females).

Statistical analysis

Statistical analyses were performed using R software (version 3.6.3). All data were visually inspected for the presence of outliers. Following Tukey’s method (Komorowski et al., 2016), outliers were first identified based on the interquartile range (IQR), with outliers defined as 1.5 times the IQR below or above the first or the third quartile, respectively. Afterwards, the identified outliers were individually analysed and processed according to a previously described analysis (Muchaxo et al., 2020).

Multilevel analyses (fixed effect) were conducted to assess differences in average time-trial velocity among the four SCI groups of athletes. The SCI groups were dummy coded and three different models were built with the groups Th1–Th5, Th6–Th9, and Th10–L1 as reference (1). All models were adjusted for sex, age, distance, and motor completeness of the SCI. Potential systematic differences among events and athlete’s repeated measures were taken into account by building a two-level model with events as the first level and athletes as the second level. Significance level was set at α = 0.05, and effect sizes were calculated as Cohen’s d with the following cut-offs: d < 0.01 for a trivial effect; 0.01 ≤ d < 0.2 for a very small; 0.2 ≤ d < 0.6 for small; 0.6 ≤ d < 1.2 for moderate; 1.2 ≤ d < 2.0 for large; and d ≥ 2.0 for a very large effect (Hopkings, 2002).

Trunk flexion strength and handcycling performance

Participants

Twenty-nine elite recumbent handcycling athletes from H3 and H4 sport classes (26 athletes with impaired muscle power and three athletes with lower-limb deficiency; H3: (n = 17), age = 37 ± 9 years, body mass = 64 ± 11 kg; H4: (n = 12), age = 38 ± 9 years, body mass = 61 ± 12 kg) competing at the 2018 Para-cycling Road World Cup (Emmen, the Netherlands) and at the 2018 Para-cycling Road World Championships (Maniago, Italy) volunteered to participate in this study. Initially, all participants registered for the competition event received an email from the UCI with the information letter of the respective study. Participants had to fulfil the following inclusion criteria: 1) internationally classified in one of the existing handcycling sport classes; 2) registered for the time-trial competition of the respective event; 3) no upper-limb impairments; 4) no health condition or active medical treatment that could influence the testing outcomes; 5) being ≥18 years of age. When interested, participants could schedule a meeting with the researcher for more information and if they maintained the will to participate they had to sign an active informed consent prior to the measurement session. From the 29 athletes registered, three male participants did not perform the time-trial and were excluded from this analysis, leading to a sample of 26 athletes. Four male and two female participants did not perform the isokinetic sprint test and were excluded, leading to a sample of 23 athletes for this particular analysis. The research was approved by the Scientific and Ethical Review Board of the Faculty of Behavioural and Movement Sciences, Vrije Universiteit Amsterdam, the Netherlands, No: VCWE-2018-093.

Experimental protocol

Prior to testing in the field-lab, i.e., a standardised lab setting that was arranged at the competition venue, participants were interviewed regarding demographic information, sport experience and health conditions, including impairment characteristics. Each participant performed a set of two different measurements in the following order: 1) a Trunk Flexion Handheld-dynamometer Test and 2) a 20-sec Isokinetic Sprint Test.

Trunk flexion strength was assessed with the use of a handheld dynamometer (MicroFET2 Digital Handheld Dynamometer, HOGGAN Scientific, USA.) (Figure 1), with the participant supine in a semi-recumbent position on an examination table. The participant sat against a rigid backrest with a 55-degree angle with the horizontal and the legs extended and strapped with an adjustable Velcro belt to the table at thigh level. The dynamometer was placed at and perpendicular to the sternum. The participant was asked to perform a maximal voluntary contraction by executing trunk flexion during a “make” procedure, i.e., while the examiner provided a stable resistance (Figure 1). During the test, the hands of the participant rested on the legs and the participant was instructed to gradually increase the force exerted against the dynamometer without actively using the limbs. Following a familiarisation trial, each participant performed four testing trials where maximal force was collected. When the trial was not performed correctly, an additional trial was performed. The maximal force (N) of four correct trials was averaged and used as trunk flexion strength outcome for further analysis.

Figure 1. Handheld dynamometer device (MicroFET2 ProCare) (left), and position of the athlete during trunk flexion test (right)

The second measurement was a 20-sec isokinetic all-out handcycle sprint test as a measure of handcycling performance in a controlled test setting. The participant’s handbike was attached to a cycle ergometer (Cyclus 2, RBM Electronics, Leipzig, Germany) and the test started with a 5-min warm up at a self-selected power output, followed by a 2-min rest. Adapted from Zeller et al. (2015) (Zeller et al., 2015), the 20-s isokinetic sprint test started with an initial load of 20 N and had a cadence limitation of 130 rotations per minute. The highest power output (PO_peak (W)) and mean power output (PO_mean (W)) were determined after the 20-sec sprint test and used as outcome measures of handcycling performance.

In addition, official time-trial results of the respective events were collected from the UCI’s website (Union Cycliste Internationale, 2020c) and average velocity (m/s) was calculated (1) and used as a measure of handcycling performance in a field setting.

(1) $\begin{array}{rl}& Timetrial/,average/,velocity\left(m{s}^{-1}\right)\\ & =\frac{Distance\left(m\right)}{Time/,of/,completion\left(mm:ss:ms\right)\times 24\times 60\times 60}\end{array}$(1)

Statistical analysis

Statistical analyses were performed using R software (version 3.6.3). Prior to the analysis of the association between trunk flexion strength and handcycling performance, all data were visually inspected, tested for normality using the Shapiro–Wilk test, and an α level of 0.05 was set. Data were not normally distributed; therefore, Spearman correlation coefficients were calculated between trunk flexion strength and the handcycling performance outcomes, i.e., peak PO, mean PO, and average time-trial velocity. Correlation coefficients were considered: very small for rs<0.1; small for 0.1≤rs<0.3, moderate for 0.3≤rs<05, large for 0.5≤rs<0.7 and very large for rs≥0.7^{(Hopkings,2002)}.

Results

SCI level & handcycling performance

Table 1 describes the number of time-trial results included in the multilevel regression model, based on SCI level group and sex. From the time-trial results sample (n = 528), 26% (n = 135) was from female athletes, and 21% (n = 112) was from athletes who had a motor incomplete lesion. The majority of athletes (52%; n = 277), had an SCI at the level of the lower abdominal wall (Th10–L1 group) and only 4% (n = 19) had an injury of L2 or below (≤L2 group)

Table 2 presents the multilevel regression results regarding differences in average time-trial velocity between SCI level groups after correction for age, sex and lesion completeness. Results showed that the different lesion level groups did not exhibit significant differences in handcycling performance (p > 0.05). In addition, all the comparisons exhibited very small effect sizes (d < 0.2). However, in this sample the Th10–L1 group tended (p = 0.051) to have a higher velocity than the ≤L2 group, despite the minimal effect size (d = 0.11).

Table 2. Multilevel models (Levels: Event and Athlete) assessing the differences in average time-trial velocity among pairs of SCI groups, with Th1–Th5, Th6–Th9 and Th10–L1 as a reference after controlling for age, sex (male = 1; female = 0), and AIS differences (AIS A/B = 1; AIS C/D = 0). Β = unstandardised beta values; SE = standard error; CI = confidence interval

Model with Th1–Th5 as reference
Fixed effect	β	SE	CI (95%)	z-value	p-value	Cohen’s d
Constant	7.96	0.62	[6.75; 9.17]	12.91	< 0.00001
Th1–Th5:Th6–Th9	0.11	0.32	[−0.51; 0.73]	0.35	0.73	0.02
Th1–Th5:Th10–L1	0.49	0.30	[−0.09; 1.08]	1.66	0.10	0.08
Th1–Th5:≤L2	−0.51	0.56	[−1.60; 0.59]	−0.91	0.36	0.08
Model with Th6–Th9 as reference
Fixed effect	β	SE	CI (95%)	z-value	p-value	Cohen’s d
Constant	8.07	0.58	[6.94; 9.21]	13.97	< 0.00001
Th6–Th9:Th1–Th5	−0.11	0.32	[−0.73; 0.51]	−0.35	0.73	0.02
Th6–Th9:Th10–L1	0.38	0.28	[−0.17; 0.94]	1.36	0.17	0.07
Th6–Th9:≤L2	−0.62	0.53	[−1.66; 0.42]	−1.16	0.25	0.10
Model with Th10–L1 as reference
Fixed effect	β	SE	CI (95%)	z-value	p-value	Cohen’s d
Constant	8.46	0.57	[7.34; 9.57]	14.91	< 0.00001
Th10–L1:Th1–Th5	−0.49	0.30	[−1.08; 0.09]	−1.66	0.10	0.08
Th10–L1:Th6–Th9	−0.38	0.28	[−0.94; 0.17]	−1.36	0.17	0.07
Th10–L1:≤L2	−1.00	0.51	[−2.00; 0]	−1.95	0.051	0.11
Variables Controlled^a
Parameter	β	SE	CI (95%)	z-value	p-value
Sex	1.00	0.28	[0.46; 1.55]	3.60	0.0003
Age	0.003	0.01	[−0.02; 0.03]	0.23	0.82
AIS (motor function)	−0.36	0.31	[−0.96; 0.25]	−1.16	0.25

^aThe values from the controlled variables did not change between models and therefore are presented separately for an easier visualization.

Trunk flexion strength and handcycling performance

In this athlete sample (n = 29) from two time-trial events, 25% (n = 7) were female, and 59% (n = 17) were classified as H3. Participants from the H3 sport class presented values of trunk flexion strength ranging from 0 N to 71 N (median = 37 N), whereas participants from H4 sport class presented values from 56 N to 344 N (median = 113 N). The medians of POpeak and POmean for H3 participants were 453 W and 358 W, respectively, and 435 W and 361 W for H4 participants. During the time-trial, median average velocity for H3 participants was 9.4 m/s (33.9 Km/h), ranging from 6.2 m/s to 10.6 m/s (22.2–38.1 Km/h), and for H4 participants was 10 m/s (35.9 Km/h), with a range between 7.3 m/s and 11.7 m/s (26.4–42.2 Km/h).

The scatterplot (Figure 2) displays trunk flexion strength versus average time-trial velocity for male/female and H3/H4 classes. Spearman’s correlation coefficients between trunk flexion strength and average time-trial velocity showed a non-significant moderate association (rs = 0.36; p = 0.07). In addition, trunk flexion strength showed non-significant small to moderate associations with PO_mean (W) (rs = 0.31; p = 0.15) and PO_peak(W) (rs = 0.27; p = 0.22) (Figure 3).

Figure 2. Scatterplot of trunk flexion strength and average time-trial velocity. Data points are identified by sex and by handcycling class. (H3: spinal cord injury with lesion levels between Th1 and Th10; H4 lesion levels below Th11 or amputations)

Figure 3. Scatterplots of trunk flexion strength and handcycling performance during a 20-sec isokinetic sprint: Mean PO (upper graph) and Peak PO (lower graph). data points are identified by sex and by handcycling class. (H3: spinal cord injury with lesion levels between Th1 and Th10; H4 lesion levels below Th11 or amputations)

Discussion

The current study presents two research approaches to investigate the influence of trunk function on recumbent handcycling performance. We retrospectively investigated the association between the handcyclists’ level of SCI and handcycling performance in a sample of international H3/H4 elite handcyclists (n = 81). In addition, a cross-sectional experimental approach focussed on the association between trunk flexion strength and handcycling performance in (n = 29) international elite H3/H4 athletes. The findings of both research approaches suggested that trunk function does not play a major role in recumbent time trial handcycling performance.

SCI level & handcycling performance

As hypothesised, differences in average time-trial velocity between athletes with more or less trunk function, based on their SCI level, were not significant in the sports classes investigated. Our findings corroborate earlier para-sport research that found only small differences in trunk performance between athletes with different levels of trunk function, e.g., in wheelchair racing (Vanlandewijck, Verellen, Beckman et al., 2011), although in our handcycling study, as a significant distinct discipline, the influence of trunk function is even less prominent. The lack of differences among the trunk function groups H3/H4 is in line with previous research that suggested no differences regarding handcycling force effectiveness between individuals with different paraplegia levels (Verellen et al., 2012b). Verellen et al. (2012) found that handcyclists with a high thoracic SCI (Th1–Th8), i.e., with lack of trunk strength but no upper-limb impairments, presented values of fraction of effective force during recumbent handcycling comparable to handcyclists with low thoracic SCI (below Th10), i.e., who have partial to full trunk strength (Verellen et al., 2012b). This is particularly evident when comparing our Th1–Th5 group, which included athletes with minimal trunk function, with groups with lower SCI levels. The non-significant differences and negligible effect sizes indicates comparable performance across all groups.

Given the large aerobic component of a handcycling race (Finch & Baggish, 2016), athletes with lesion levels above Th6 might have a disadvantage due to a possible impaired response of the ANS to exercise (Krassioukov & West, 2014^; Theisen, 2012^; West et al., 2015). Reduced cardiac output and deficient thermoregulation may eventually affect the performance outcome (Krassioukov & West, 2014^; Theisen, 2012^; West et al., 2015). However, in the present study, the group Th1–Th5 demonstrated comparable performance, with minimal effect sizes, even when compared to groups with greater discrepancy regarding trunk function. Such findings may contribute to different assumptions: a) that impaired trunk function and/or impaired ANS, among high thoracic SCI, presents limited influence on time-trial handcycling performance; b) that the sample may reflect a greater absence of athletes with impaired ANS competing at an elite level because they will not make it to this level. The influence of an impaired ANS varies among high thoracic SCI (Krassioukov & West, 2014), but literature has shown a general negative impact on sport performance, with emphasis on sports with greater demand for aerobic exercise capacity, like handcycling (West et al., 2015). However, it should be noted that previous research has shown limited association between motor/sensory completeness of the lesion and autonomic completeness (West et al., 2015). Although the results may suggest an underrepresentation of these athletes, we do not have sufficient data to verify the ANS assumption solely based on the motor completeness of the lesion and additional measurements to verify this are recommendable.

Previous evidence has shown that the lesion level, including tetraplegia and paraplegia, may be a good predictor of the maximal aerobic power in handcycling (De Groot et al., 2019); however, it has also been shown that the level alone may not be a good predictor of trunk function (Bjerkefors et al., 2007). According to the UCI classification, athletes in class H3 present impairments corresponding to an SCI between Th1 and Th10, while athletes in class H4 present SCI below Th11 or equivalent (Union Cycliste Internationale, 2020b). The classes are further detailed on the level of trunk stability and abdominal strength, separated by limited to very limited in H3 and normal to almost normal in H4. In the current study, we did not assess trunk stability as we do not consider trunk vertical stability to play a major role in a recumbent position, where athletes are often positioned aerodynamically and strapped to the handbike. The findings of this first analysis suggest that handcycling athletes with good upper-limb function exhibit comparable time-trial average velocity at group level despite differences in trunk function.

Trunk flexion strength and handcycling performance

The findings in our cross-sectional analysis matched our initial hypothesis regarding weak to moderate associations between trunk flexion strength and handcycling performance. The association between trunk flexion strength and average time-trial velocity showed a non-significant moderate correlation. A similar correlation was observed between trunk flexion strength and PO_mean during the isokinetic sprint test. In this controlled test setting, a weaker association was found between trunk flexion strength and PO_peak.

H3 and H4 class profiles are similar to T53 and T54 from wheelchair athletics. The classification of these athletes is comparable, as both classes include athletes with partial to full trunk stability (International Paralympic Committee, 2017^; Union Cycliste Internationale, 2020b), although it should be noted that the T54 sport class includes a broader range, compared to the H4 sport class (Th11 and down). More specifically, T53 includes a range of SCI levels from Th1 to Th7, while T54 includes a broad range of SCI levels corresponding from Th8 and down (International Paralympic Committee, 2017). In a previous study, it was shown that athletes classified in T54 with different levels of trunk strength do not differ in performance based on acceleration test from standstill (Vanlandewijck, Verellen, Beckman et al.,2011). In recumbent handcycling, elite athletes adopt a laying down aerodynamic position (Stone, Mason, Bundon et al., 2019), which is expected to reduce the involvement of trunk more than in sitting upright positions (Faupin et al., 2006, 2008; Quittmann et al., 2018; Stone, Mason, Warner, Goosey-Tolfrey et al., 2019; Stone, Mason, Warner, Tolfrey et al., 2019; Verellen et al., 2012a). In addition, abdominal muscle activation has been found to be low in able-bodied participants performing recumbent handcycling, and expected to be lower in elite athletes with SCI (Quittmann et al., 2019), as the upper limbs are mainly responsible for force generation and steering (Verellen et al., 2012). Hence, it could be hypothesised that between H4 and H3 sports classes, trunk strength plays a minor role and differences between these classes may be less than between T54 and T53.

The findings of the present study are in line with previous studies and lead us to assume that the impact of trunk flexion strength on performance will be considerably diminished when compared to the other para-sport modalities in which a more upright position is necessary. In a recent study, stronger relationships between wheelchair racing performance and isometric strength were found when combined arm–trunk strengths were included as predictors instead of trunk strength only (Connick et al., 2017). In our study, trunk flexion strength was measured in an isolated manner and future research should focus on the measurement of multiple joints and muscle functions, such as arm pulling while controlling for trunk position. Nevertheless, the same wheelchair racing study has suggested four different clusters associated with different performance levels. Aside from the clusters that included athletes with upper-limb impairments, it was remarkable to observe that the second least impaired cluster included athletes with SCI levels from Th2 to Th12, comprising nearly the total range of athletes in H3 and H4. Differences in performance were not observed between these different levels of trunk function. Instead, differences were found between athletes who did and did not have hip–leg connection (Connick et al., 2017).

In the present study, the results from the trunk flexion strength (second analysis) are in agreement with the results from SCI level (first analysis), as well as with previous findings on small differences between H3 and H4 time-trial handcycling performance (Muchaxo et al., 2020). Although H3 and H4 athletes present differences in trunk function and flexion strength, the present study suggests that trunk flexion strength is not an essential determinant of recumbent handcycling performance in these classes, as similar values for average velocity and sprint power-output were found. These results suggest that isolated trunk flexion strength should not be included as a discriminative factor in classification, contrary to the current trunk perspective of the classification system, which assesses trunk strength as an important method of class allocation.

Comparable performance between H3 and H4, i.e., comparable average velocities and sprint power-outputs despite differences in trunk function and flexion strength imposes a threat to the currently used hierarchical classification system. However, notwithstanding the recent findings on the minor involvement of trunk function and flexion strength in recumbent time-trial handcycling performance, the scientific evidence does not necessarily support merging of the H3 and H4 classes since other factors may play a role in performance. Determinants such as the ability to perform an active closed-chain with the lower limbs (Kouwijzer et al., 2018), combined joint function, core co-contraction, or impact of strapping may play more important roles in determining handcycling performance and should be included in future studies.

Limitations

To investigate associations with a time trial event, which implies a significant endurance component, aerobic measures of performance capability could be used or maybe even preferred (Van Der Zwaard et al., 2018). For example, literature suggests significant strong correlations between peak aerobic PO (W) and handcycling performance (Lovell et al., 2012^; Stone et al., 2020). Correlations between endurance events and anaerobic measures of performance might be less strong (Van Der Zwaard et al., 2018). However Janssen et al. (1993) and De Groot et al. (2021) (De Groot et al., 2019) found strong associations between sprint PO and peak aerobic capacity (oxygen uptake and aerobic PO_peak) in individuals with SCI during wheelchair propulsion and arm cranking, respectively. De Groot et al. (2012) (De Groot et al., 2012) found that individuals with neurological impairment presented strong associations between sprint PO and oxygen uptake and aerobic PO_peak. Quittmann et al. (2018) showed strong associations between sprint PO and lactic power and Nooijen et al. (Nooijen et al., 2021) found a significant association (r²= 0.61) between the 20-sec isokinetic PO_mean and time-trial average velocity in elite handcycling athletes competing at world cups and world championships.

A limitation of the present study is the sample size. While in the first analysis a relatively large sample was used, the size of the most functional group is small and may influence the comparison with the other groups. Although undesirable, this is a reflection of the competition, as athletes with hip–leg connection are distributed among recumbent and kneeling handbikes (H4 and H5 classes), dependent on additional functional characteristics. Similarly, in the second analysis, it was not possible to include a larger sample, despite the fact that the participants were elite athletes tested on-site at the World Cup and the World Championships. Research at events is often complicated by the obligations of the event for the athlete, while otherwise elite handcyclists are difficult to reach worldwide in a standardised experimental protocol.

The assumptions made from the multilevel regression analysis may be limited due to the absence of data regarding the athlete’s fitness level. However, we do not have reasons to believe there is a significant discrepancy in the number of unfit athletes between H3 and H4 classes, which composed the SCI groups included in this analysis. In addition, there was no information on whether the response of the ANS to exercise was limited in the athletes. Finally, rigid standardised strength testing setups may provide higher reliability and accuracy in the assessment of trunk strength, compared with handheld dynamometry, as it reduces the measurement errors from the tester. However, besides the impracticality of rigid strength setups during research at sporting events, we do not believe the use of a handheld dynamometer has significantly impacted the assumptions formulated in this current study as the trunk flexion trials showed similar ranges per participant. The differences in trunk flexion strength were observed in alignment with trunk function, and in addition, the use of a handheld dynamometer has been considered a valid and reliable assessment tool for muscle strength in different body segments, including trunk flexion (De Blaiser et al., 2018^; Febrer et al., 2010; Karthikbabu & Chakrapani, 2017).

Conclusion

The involvement of the trunk in Paralympic sports is complex, and trunk function is not solely explained by trunk flexion strength and lesion level. However, based on the two analyses conducted in this study, it can be concluded that trunk function and flexion strength seem to have a minor impact on recumbent handcycling time-trial performance of elite H3/H4 athletes. The first analysis showed that handcycling performance was not significantly different among groups based on SCI level. These findings were supported by the second analysis, where only non-significant small to moderate associations were found between trunk flexion strength and handcycling performance. While generalisation of these findings should be addressed with caution, the insights of this study exposed the need to pursue evidence-based research on other potential determinants of handcycling performance that should be considered for the classification system.

Disclosure of potential conflicts of interest

No potential conflict of interest was reported by the author(s).

Additional information

Funding

This project has been carried out with the support of UCI (Union Cycliste Internationale). The funding body was not involved in decisions concerning the design of this study, data analysis, interpretation of data or in reporting and publishing this project.