Sage Journals: Discover world-class research

Abstract

Introduction

This study compared fixed and individualized speed zone methods for quantifying external match load (EML) across classification and playing levels in women’s wheelchair basketball (WB) and examined their relationship with internal match load (IML) measures.

Methods

EML of 31 well trained female players (age: 28 ± 7 years, experience: 11 ± 5 years, weekly training: 12 ± 4 h) were quantified across 13 international (INT) and 5 Women’s Premier League (WPL) games using two inertial measurement units. A linear mixed model compared the percentage time in fixed and individualized speed zones and correlations with IML (eTRIMP, sRPE) using Pearson and Spearman Rho.

Results

Individualized zones recorded more time (10%) in low speed (Z3) and less time (0.3–5.9%) in moderate to very high-speed zones (Z4-Z6) versus fixed (p < 0.0001). Differences and effect sizes were greater in INT games (+0.8-1.1 pp) and highest classifications (+9.5 pp). Both methods correlated poorly with IML, except weak Z3 correlations at the INT level (p ≤ 0.020).

Conclusions

Speed zone method influences high-intensity movement assessments, varying by playing level and classification. Neither method optimally related to IML, highlighting the need for calibrated, consistent thresholds.

Keywords

speed zones elite classification match load inertial measurement units

Introduction

Wheelchair basketball (WB) is a popular Paralympic sport, enabling athletes’ whose disability fits within one of the seven eligible impairment types¹ an opportunity to participate at the highest level. Examples of eligible disabilities that significantly affect an individual’s ability to play WB include spinal cord injuries (SCI), lower limb amputations, cerebral palsy, and other mobility impairments. Players are allocated a classification ranging from 1 to 4.5, based on the extent their physical impairment impacts their movement and function on the court; the sum of all 5 players on court must not exceed 14.0.¹ Internationally ineligible players may be classified as a 5.0 under the national classification system and play domestically, where teams are permitted a total of 14.5 on court at any one time.²

The sport is an intermittent, high-intensity team sport played over four 10 min quarters, and is characterised by frequent high-speed sprints, accelerations, decelerations, and changes of direction.^3,4 Quantifying the internal and external demands in wheelchair court sports is crucial to optimise training practices and athletes’ preparedness for competition. The external match load (EML) refers to the work performed on court, such as distance covered, time spent in speed zones, or number of high-intensity actions, while the internal match load (IML) reflects the physiological cost of that work, often measured using heart rate (HR) or rating of perceived exertion (RPE).^5,6 The IML can be quantified as a training impulse (TRIMP) score using the time in HR zones⁷ or as a session RPE (sRPE),⁸ the product of session duration and RPE. Since HR monitoring presents known limitations for wheelchair athletes with SCIs, due to altered autonomic regulation,⁹ and has demonstrated poor validity in intermittent sports,¹⁰ practitioners are encouraged to combine both IML and EML metrics when assessing load in wheelchair court sports.

Technologies such as inertial measurement units (IMUs), placed on the wheelchair frames and wheel hubs, have proven effective for profiling wheelchair mobility performance and EML demands in wheelchair sports such as tennis, rugby and basketball.^11–14 Wheelchair mobility performance refers to the kinematic output of the wheelchair-athlete system, including metrics such as speed, acceleration and movement patterns.¹¹ From these IMU-derived data, variables such as total distance covered, peak and average speed, number of efforts, and time spent in specific zones are typically extracted to estimate EML. The number of efforts or time spent (absolute or percentage) within specific speed zones, as seen in previous WB studies^{11–13,15,16} provides a more detailed understanding of EML intensity distribution compared to distance covered alone.

In non-disabled sport, speed zones are often expressed as a percentage of peak speed,^17–19 however wheelchair sports have used diverse zonal approaches. Some wheelchair rugby studies have employed individualized zones based on peak wheelchair propulsion speed,^15,16 whereas WB has relied on fixed speed thresholds,^3,12,13 with the highest zones ranging from 2.50 m.s^-13 to 3.50 m.s^-1.^11,13 However, these upper zones may only represent approximately 50 – 67% of an athlete’s peak speed, limiting the ability to measure high-intensity wheelchair propulsion. As a result, efforts of varying intensities may be grouped into a single, broad zone, reducing the sensitivity of the system to detect and differentiate high-intensity propulsion. This is particularly relevant given peak speeds of up to 5.22 m.s^-1 have been reported during WB match play.¹¹

Wheelchair mobility performance varies depending on player classification and playing level.^11,20 Players with lower wheelchair mobility performance, and therefore lower peak speed, may have fixed speed thresholds set too high, resulting in minimal time recorded in fixed speed high-intensity zones.¹¹ Conversely, those with higher peak speeds, due to their classification or playing level, may record more time in these zones, leading to an overestimation of high-intensity activities.^18,19 The appropriate selection of zonal methods is essential to accurately gauge match load across players with varying wheelchair mobility performance. Despite these observations, no investigations have examined the applicability of these zones across classification or playing level groups within WB.

The primary aim of this study was to examine how fixed and individualized speed zone methods, which differ fundamentally in how speed thresholds are defined, influence the quantification of EML across classification groups and playing levels in women’s WB. Building on this, the secondary aim was to explore the relationship between EML quantified using these zones and measures of IML to evaluate how different approaches relate to physiological demand.

Methods

Participants

Thirty-one female WB players (age 28 ± 7 years, playing experience 11 ± 5 years, weekly court training 8 ± 3 h or 4 ± 2 sessions, weekly gym training 3 ± 1 h, weekly total training 12 ± 4 h) provided written informed consent and volunteered to participate in the study. Approval for the study was obtained by the Human participants sub-committee at Loughborough University (2022-8445-10161). For inclusion, players had to be a member of the GB national team (n = 11) or from four teams in the UK domestic female league (Women’s Premier League, WPL; n = 20). The WPL is a non-professional league, with each team permitted a maximum of four GB players. As a result, the standard of competition is lower than the international level. Players were excluded before data collection if their coaching staff identified they would not receive any playing minutes. All international and fourteen players from the domestic level held an international playing classification,¹ and six players had a classification from the National Governing Body.² All players used their own personalized sports wheelchair for all testing sessions.

Design

Competitive match play (excluding the warm-up periods) was monitored at two different playing levels, international level (INT) and the WPL. The INT matches took place over a four-month period (Aug – Oct 2022), during training camps leading up to the IWBF 2022 World Championships in November (which was later rescheduled to June 2023). Thirteen friendly international matches were used for data collection, with 100 observations collected. The WPL data was collected over a four-month period (Jan – Apr 2023) during the regular national league season. Data was collected from five league matches involving four teams (excluding the warm-up periods), totaling 31 observations. Teams ranked 2 to 4 were observed in games against the top ranked team, and the top ranked team was measured for two games, against the 2^nd and 3^rd ranked teams in the league. Prior to data collection, participants were familiarised with IMU technology on their wheelchair and wearing HR monitors.

Measurements

Wheelchair configurations (wheel diameter, chair width and camber angle) were measured during the study, for subsequent IMU data processing and the calculation of linear and rotational velocity, acceleration and displacement.^14,21 Prior to match play and training sessions involving maximal sprinting, participants completed a standardised 20-minute warm-up on a standard wooden-sprung WB court. Maximal sprinting speed was assessed using three 20m maximal sprints from a stationary start, with 3-min passive recovery.

To measure speed, two IMUs (Movesense Active, Vantaa, Finland; 52 Hz) were secured onto straps with stiff foam padding and tightly fitted to the sports wheelchairs - one on the hub of the right wheel and the center of the frame.²¹ Each IMU contained a gyroscope (2000 DPS), accelerometer, and magnetometer. Data from testing, training, and match play were transmitted via Bluetooth to an application on a handheld mobile device (Movesense Showcase, Version 11.0; iPhone SE, Apple Inc., California, USA).

For match play, EML was measured using the same IMU configuration. IML was collected using chest strap HR monitors (Polar Pro Chest Strap, Polar Electro Oy, Kempele, Finland) with data transmitted via Bluetooth to the Polar team pro tablet application (Polar Team Pro system, Polar Electro Oy, Kempele, Finland).

During all observations, substitutions and extended stoppages of time (time-outs, equipment delays, half-times, and quarter breaks) were noted. All data collection was performed by the lead researcher (embedded sports scientist, 6 years of applied sports science experience), a co-author (5 years of IMU data collection experience with wheelchair tennis) and supported by the sports strength and conditioning practitioner.

Data analysis

Data were excluded from the analysis if the athlete played less than 10 min of a match or if IMU Bluetooth connectivity disconnected for more than 30 s within a quarter.

IML

HR was recorded at ∼10 Hz and averaged at a rate of 1 Hz, with peak HRs determined for each participant at the end of the data collection period using the highest 5 s average HR during all observations. HR zones were split into Z1 (50-60% of peak HR), Z2 (60-70%), Z3 (70 – 80%), Z4 (80 – 90%) and Z5 (90 – 100%) and a training impulse (TRIMP) score was calculated using the summated HR zone method⁷:

eTRIMP = (Z 1 \min x 1) + (Z 2 \min x 2) + (Z 3 \min x 3) + (Z 4 \min x 4) + (Z 5 \min x 5)

(1)

For the INT players, after each game, an RPE score was recorded using the Borg CR-10 scale²² and used to calculate sRPE⁸:

sRPE = RPE score x duration of the session (\min)

(2)

Unfortunately, RPE could not be collected from the WPL players due to restricted court-side access imposed by the competition regulations.

EML

The accelerometer and gyroscope data of both IMUs were analysed to derive all variables using custom-written Python scripts, based on previously established methods (Python Software Foundation, Delaware, United States).^14,23,24 Speed was calculated using the gyroscope of the wheels, corrected for the camber angle using the gyroscope of the frame.²⁴ Distance covered and active minutes within each match were quantified accordingly, defined when the forward or reverse speed exceeded ±0.1 m.s^-1. The frequency of independent sprints efforts exceeding 3 m.s^-1, were measured as high intensity activations (HIA). Both percentage time and absolute minutes spent in five speed zones were subsequently calculated. These were computed using a modified fixed zone method¹³ and an individualized zone method, derived from the participants peak speed,¹⁷ Z1 (reverse propulsion) was excluded from the analysis (Table 1). The participant’s peak speed was recorded as the highest from the 20 m sprints, or the highest speed recorded in match play for 9 players (8 WPL, 1 INT) who were not available for sprint testing.

Table 1.

Cut-off values used in the fixed and individualized zone methods and example zone thresholds used in the individualized zones for the WPL and INT groups.

Description	Zone number	Fixed zones (m.s^-1)	Individualized zones
Description	Zone number	Fixed zones (m.s^-1)	Thresholds (% peak speed)	WPL (m.s^-1)	INT (m.s^-1)
Reverse	1	< 0	-	-	-
Very low	2	0 < 1	0 < 20	0 < 0.9	0 < 1.0
Low	3	1 < 2	20 < 50	0.9 < 2.3	1.0 < 2.5
Moderate	4	2 < 3	50 < 80	2.3 < 3.6	2.5 < 4.0
High	5	3 < 4	80 < 95	3.6 < 4.3	4.0 < 4.8
Very high	6	> 4	> 95	> 4.3	> 4.8

^aAbbreviations: INT, International; WPL, Women’s Premier League; m.s^‐1, metres per second.

Statistical analysis

Statistical analyses were conducted using IBM SPSS Statistics (Version 28, Chicago, Illinois, USA) and a custom R script (RStudio, Version 2024.12.1, Posit, Joseph Allaire). The Shapiro-Wilk test revealed the time in speed zones data to violate parametric assumptions; therefore, for aim 1 a linear mixed model was applied to each speed zone, incorporating logarithmic or square root transformations based on model fit. In the models, the relative time spent in the two zonal methods were fixed factors, and participant was included as a random effect, accounting for multiple observations per individual, with two performance levels (INT and WPL) and three classification groups (Group A: 1 – 1.5; Group B: 2 – 3; Group C: 3.5 – 5.0). Using the fitted model, differences between the two zonal methods were analysed using a Type II Wald Chi-Square ANOVA with post hoc pairwise comparisons. These were interpreted using a Hedge’s g score (trivial, g < 0.20; small, g = 0.20 < 0.50; medium, g = 0.50 < 0.80; large, g ≥ 0.80).²⁵

For aim 2, a linear mixed model was fitted to examine the relationship between EML and IML measures, with participant as a random effect and across two performance levels (INT and WPL), while classification group was not included as a level. Pearson correlations were conducted, and where transformed data that violated parametric assumptions, Spearman rho correlations were additionally used. The correlations were interpreted using correlation coefficient magnitudes²⁶ (negligible, r < 0.1; weak, r = 0.10 – 0.39; moderate, r = 0.40 – 0.69; strong, r = 0.70 – 0.89; very strong, 0.90 – 1.00). Statistical significance was set at p ≤ 0.05 for all analyses and Bonferroni corrections were used for multiple comparisons and correlations. All data are presented as estimated marginal means ± standard error.

Results

After the exclusion criteria were applied, 71 observations were included from the INT group (29 excluded) and 24 from the WPL group (7 excluded). The peak speeds for the whole cohort, performance-based subgroups, and classification groups are summarised in Table 2.

Table 2.

Estimated marginal means ± SE peak speed profiles for the whole cohort, subgroups and classification groups.

Groups	Group AClass 1 – 1.5 (m.s^-1)	Group BClass 2 – 3 (m.s^-1)	Group CClass 3.5 – 5 (m.s^-1)	All (m.s^-1)
Whole cohort	4.48 ± 0.11 (n = 8)	4.85 ± 0.11 (n = 10)	4.95 ± 0.08^d (n = 13)	4.77 ± 0.14b (n = 31)
Subgroups:
INT	4.81 ± 0.20 (n = 2)	5.05 ± 0.21 (n = 2)	5.10 ± 0.11 (n = 7)	5.00 ± 0.10^c (n = 11)
WPL	4.18 ± 0.20 (n = 6)	4.71 ± 0.10^d (n = 8)	4.76 ± 0.11^d (n = 6)	4.53 ± 0.06 (n = 20)

^aAbbreviations: INT, International; WPL, Women’s Premier League; m.s^-1, metres per second.

^bDenotes main effect of classification.

^csignificantly higher than WPL subgroup.

^dsignificantly higher than class A.

A significant main effect for classification was observed (F ₍₂₎ = 9.63, p < 0.001), with post-hoc tests revealing greater peak speeds in group C (+0.47 m.s^-1, p = 0.007, ES = 1.5) compared to group A. Peak speeds were also higher for the INT subgroup compared to the WPL subgroup (5.00 vs. 4.53 m.s^-1, p < 0.001, ES = 1.5).

Comparison of zonal methods

A main effect for the zonal method was observed for Z3 - 6 (p < 0.05, Figure 1), with more time recorded using the fixed method compared to individualized for Z4 to Z6 (+ 0.3 - 5.9pp., p < 0.0001, ES = 0.7 – 2.3), whereas the opposite was true for Z3 (- 10 pp., p < 0.0001, ES = 1.8).

Figure 1.

Estimated marginal means of percentage time spent in speed zones 2 – 6 across INT and WPL groups. Error bars indicate SE. Symbols denote statistical significance: # main effect of method (fixed vs. individualized), † main effect of group, * main effect of classification (p < 0.05 for all). Letters indicate posthoc differences between methods within groups: a for INT, b for WPL. *Abbreviations: Z, Speed zone; INT, International matches; WPL, Women’s Premier League matches.

The WPL group spent greater time in Z4 compared to the INT group (+3.8 pp., p = 0.0001, ES = 1.6). The fixed method resulted in greater time spent in Z2 at the WPL level (+3.4 pp., p < 0.0001, ES = 0.8) and Z4 at the INT level (+4 pp., p < 0.0001, ES = 1.3). This was also the case for Z5 and Z6, however larger differences were observed between methods at the INT level (+1.1 – 6.7 pp., p < 0.0001, ES = 1.7 – 2.3) compared to the WPL level (+ 0.3 – 5 pp., p ≤ 0.0002, ES = 0.7 – 1.4).

The effect of classification

A main effect of classification was found for Z3, Z5 and Z6 (p ≤ 0.046), with post hoc tests showing that group C spent more time in Z5 and Z6 (+0.2 - 1.4 pp., p ≤ 0.0018, ES = 0.9 – 3.3). Group B recorded more time in Z6 (+0.1 pp., p = 0.026, ES = 2.7) compared to group A (Figure 2).

Figure 2.

Estimated marginal means of percentage time spent in speed zones 2 – 6 across three classification groups (A, B and C). Error bars indicate SE. * Symbols indicates fixed speed zone method significantly higher than the individualized method (p < 0.05). *Abbreviations: Z, Speed zone; A, Classification group A (1 – 1.5); B, Classification group B (2 – 3); and C, Classification group C (3.5 – 5).

The fixed method resulted in more time spent in Z5 compared to the individualized methods; however, the differences were larger in group C (+16.4 pp., p < 0.0001, ES = 3.3) compared to group A (+6.9 pp., p < 0.0001, ES = 2.6) and B (+10.9 pp., p < 0.0001, ES = 2.4). Group C also had larger differences between methods (+ 1.1 pp., p < 0.0001, ES = 2.3) for Z6 compared to group B (+ 0.8 pp., p < 0.0001, ES = 1.5).

EML and IML

Strong positive correlations were observed between distance covered and HIA with eTRIMP and sRPE measures for INT observations (r = 0.77 – 0.90, p < 0.001) and WPL observations (r = 0.63 – 0.79, p ≤ 0.005) (Table 3).

Table 3.

Correlation coefficients (95% confidence intervals) between EML measures and IML measures at the international and WPL game level.

	INT (n = 11)		WPL (n = 20)
EML measure	eTRIMP	sRPE	eTRIMP
Distance	0.90a (0.85 – 0.94)	0.85a (0.77 – 0.91)	0.79a (0.57 – 0.90)
HIA	0.86a (0.78 – 0.91)	0.77a (0.66 – 0.85)	0.63a (0.30 – 0.82)
Individualized zonal method
Z2	-0.05 (-0.28 – 0.18	0.04 (-0.19 – 0.27)	0.13 (-0.29 – 0.51)
Z3	0.41a (0.20 – 0.59)	0.24 (0.12 – 0.45)	0.16 (-0.26 – 0.53)
Z4	-0.11 (-0.33 – 0.13)	-0.19 (-0.40 – 0.05)	-0.32 (-0.64 – 0.10)
Z5	0.03 (-0.21 – 0.26)	-0.08 (-0.31 – 0.16)	-0.07 (-0.46 – 0.34)
Z6	-0.22 (-0.43 – 0.01)	-0.13 (-0.36 - 0.10)	-0.31 (-0.63 – 0.11)
Fixed zonal method
Z2	-0.17 (-0.39 – 0.07)	0.13 (-0.10 – 0.36)	0.07 (-0.35 – 0.46)
Z3	0.34a (0.12 – 0.53)	0.28 (0.05 – 0.48)	0.18 (-0.24 – 0.55)
Z4	0.12 (-0.12 – 0.34)	0.04 (-0.20 – 0.27)	-0.05 (-0.44 – 0.36)
Z5	0.03 (-0.20 – 0.26)	-0.28 (-0.48 – -0.05)	-0.26 (-0.46 – 0.34)
Z6	0.16 (-0.07 – 0.38)	-0.17 (-0.39 – 0.07)	0.02 (-0.39 – 0.42)

Abbreviations: EML, External match load; IML, internal match load; HIA, high intensity activations; Z, Speed zone; INT, International; WPL, Women’s Premier League; eTRIMP, Edward’s training impulse; sRPE, sessional RPE.

^aSignificant correlation at p < 0.05 level.

At the INT level, distance covered was also moderately correlated with time spent in Z3 for both methods (r = 0.46 – 0.53, p < 0.0001), and weak correlations were observed between HIA and time spent in individualized Z3 (r = 0.33, p = 0.02) and fixed Z6 (r = 0.30, p = 0.04). There were weak to moderate correlations between time spent in Z3 and the eTRIMP method for both the fixed (r = 0.34, p = 0.020) and individualized methods (r = 0.41, p = 0.002) at the INT level. No additional statistically significant correlations were identified.

Discussion

This study examined differences in EML quantification from two commonly used speed zone methods (individualized vs. fixed) in women’s WB. The aim was not to determine which is more accurate, since no gold standard exists, but to quantify the discrepancies between methods across two levels of competition (INT vs. WPL). The main findings were: (i) the fixed method classified more time recorded in higher-intensity zones (Z4, moderate speed; Z5, high speed; and Z6, very high-speed wheelchair propulsion) and less in Z3 (low speed) than the individualized method; (ii) these differences were more pronounced in INT players, who also achieved higher peak speeds; and (iii) classification level influenced EML, with higher-functioning athletes showing the greatest divergence between methods.

The fixed method led to a greater proportion of time spent in moderate (1–3% time), high (5-7% time) and very high-speed (0.2-0.7% time) zones compared to the individualized method. These differences were more pronounced at the higher level of competition (INT), likely reflected by the higher peak speeds observed in these athletes. Previous research has highlighted a difference of 0.43s in a 12m sprint between national and international WB athletes.²⁷ These higher peak speed capacities lead to substantially higher individualized speed thresholds in INT compared to WPL athletes, as illustrated in Table 1, thereby increasing the gap between the two zonal approaches. These findings suggest that fixed thresholds may insufficiently account for functional variability across competition levels.

The player’s functional capacity, which is associated with their level of impairment and therefore their classification group, was evidenced by the varying peak speeds between the groups. Our results showed that the lowest classification group (A), who typically have reduced trunk function limiting their propulsion speed,^1,27 spent less time in high intensity zones compared to groups B and C, regardless of the zonal method used. This pattern supports the idea that the zones may be less closely aligned for individuals with greater physical capabilities. Similar trends have been observed previously, with one study noting that ‘higher-class’ players (3.0 and above) spent slightly more time in the highest intensity zone compared to the overall sample.¹¹ The differences in peak speed capacity across classification groups are well established in WB, with particularly big differences between group A and group C athletes,^16,28,29 as observed in this study. The greatest discrepancies between methods observed in group C (8.4% time, Z5-Z6) compared to groups B and A further support the idea that the zones may be less closely aligned for individuals with greater functional capacity, and therefore impact EML quantification.

The individualized zones applied in this study were adopted from thresholds originally developed in running sports^17–19 and utilised in wheelchair rugby,¹⁵ which are recognized as arbitrary and not validated for WB. This likely explains the higher thresholds for high intensity zones in the individualized method, leading to less time being classified at very high speeds. These findings contrast with non-disabled team sports research, where fixed zonal methods typically underestimate EL.^30–32 While we acknowledge this limitation, these zones were applied as a practical approach to reflect relative intensity based on individual peak speeds. Given the absence of established methods for wheelchair sports, this study provides an important first step in exploring their application within this context. Future work should aim to develop sport-and-impairment specific zones to better capture the demands of WB.

In addressing the relationship between EML and IML measures, moderate to strong positive relationships were found between distance covered and HIA with the eTRIMP measure of IML at both competition levels, consistent with prior wheelchair rugby research.¹⁵ Additionally, moderate to strong correlations with sRPE were observed in INT players, which contrasts with previous research reporting weaker associations.¹⁵ While researchers have recommended combining RPE and HR-based measures to quantify IML in wheelchair court sports,⁹ due to logistical constraints sRPE data were not collected for WPL participants, limiting our ability to compare these relationships fully across competition levels. There are well-established limitations to the use of HR monitoring in intermittent sports, often underestimating the physiological load of short-duration high-intensity actions.¹⁰ In individuals with an SCI, blunted autonomic responses⁹ can reduce the ability to reach higher HR zones and lead to a greater proportion of time classified within moderate zones. These considerations might influence the strength of relationships observed between IML and EML measures across different impairment groups, presenting implications not only for the validation of speed zone methods but also for longitudinal monitoring purposes.

The predominance of time spent in the moderate intensity zone (Z3) across both methods explains much of the observed relationships with IML. This pattern reflects the intermittent nature of WB, where high intensity efforts are interspersed with periods of lower intensity propulsion and coasting, the latter of which complicates the interpretation of speed-based zones.²⁷ Consequently, IML measures such as eTRIMP and sRPE are more strongly associated with time spent in Z3, distance covered and HIA, as they represent global measures of load and capture the demands of the entire session rather than the most intense periods, unlike the time spent in high intensity zones. Whilst these measures of EML are essential for practitioners seeking to understand the load demands of WB, it remains vital to consider the distribution of intensity within this total load, as this may significantly influence the athlete’s perceived effort as well as their subsequent adaptations or fatigue responses. The relatively small amount of time recorded in the highest intensity zones using individualized thresholds in this study suggests that these zones may need adjustment, such as combining the upper zones or lowering the lower limits to 70-80% of peak speed to better capture meaningful high intensity efforts.¹⁵

Importantly, our analysis did not determine which zonal method has a stronger relationship with IML, and the generally weak associations observed indicate neither method clearly outperforms the other in representing internal load, as previously observed in soccer players.³⁰ Future studies employing within-individual correlations may provide more robust insight into these relationships.¹⁵

Limitations

Due to logistical constraints, not all participants completed peak speed sprint testing, resulting in some peak speeds being derived from gameplay, training or sprint trials. This may have introduced variability in individualized zone thresholds via the underestimation of true peak speed and increased the likelihood of WPL athletes (n=8) recording time in the highest zones. Therefore, caution is warranted when interpreting differences in time spent in the highest intensity zones across competition levels and practitioners should aim to adopt a standardised method for determining peak speed in future analyses. There is, however, value in confirming whether an athlete’s peak speed measured during linear sprint testing truly represents their maximal capacity, or if higher speeds are achieved during training or competition.

Maximal HR was not collected through laboratory testing, potentially affecting the accuracy of eTRIMP calculations. Additionally, sRPE was not collected from the WPL subgroup, limiting comparisons between IML between competitions. Our sample was limited to a single nation’s female squad, with some classification groups underrepresented, which may limit generalizability. Finally, given the narrower variance in peak speeds among male athletes due to higher professionalisation, fixed zones may be more relevant in that population, but this requires empirical verification.

Practical recommendations and future directions

Practitioners should be cautious when applying speed zones derived from running sports¹⁷ to WB athletes due to the fundamental differences in propulsion kinematics and the impact of impairment. Both fixed speed and individualized zonal methods have limitations; fixed zones do not account for functional variability between athletes, while individualized zones rely on able-bodied running.^17,18 Assessing additional physiological markers such as blood lactate or oxygen consumption during gameplay and laboratory testing could allow practitioners to anchor these speed zones to established physiological thresholds, including ventilatory and lactate thresholds. This would offer a more objective basis for establishing individualized zones, rather than relying on arbitrary percentages of peak speed adapted from running sports.

Until more accurate, validated zones are developed, we recommend the consistent use of a single method with an awareness of these limitations for monitoring EML in WB. It would be valuable for practitioners and researchers to retrospectively analyse existing IMU datasets from WB and apply these individualized zones, particularly when comparing intensity distributions across classification groups or playing levels. Based on the minimal time observed in Z6 in this study, we recommend combining Z5 and Z6 to represent very high intensity propulsion.

Given these limitations, future studies should focus on developing and validating wheelchair specific, impairment informed speed zones tailored to women’s WB; examining within individual EML–IML relationships longitudinally; and include male and mixed nationality cohorts to improve generalizability.

Conclusion

Fixed and individualized speed zone methods produce meaningfully different EML profiles, especially in highspeed activities (Z4-Z6). The magnitude of these differences were greater at higher competition and classification levels. Neither method demonstrated clear superiority in relation to internal measures of match load (eTRIMP, sRPE), underscoring the need to develop and validate wheelchair-specific zonal thresholds that account for the diverse functional abilities of athletes.

Footnotes

Acknowledgements

The authors would like to acknowledge all the athletes and support staff within British Wheelchair Basketball and the teams from the BWB Women’s Premier League for their participation in this research.

ORCID iDs

Pippa Bailey

Richard C. Blagrove

Thomas Rietveld

Rienk M. A. Van Der Slikke

Ethical considerations

Ethical approval was received from the Human Participants sub-committee at Loughborough University (2022-8445-10161). All procedures were conducted in accordance with the declaration of Helsinki.

Consent for publication

All participants provided written informed consent prior to data collection.

Author contributions

Conceptualisation: PB, RB, TR, VGT; Methodology: PB, RB, TR, RvdS, VGT; Data collection: PB, TR; Data analysis: PB, RB, TR, RvdS; Writing (original draft): PB; Writing (review & editing): PB, RB, TR, RvdS, VGT.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This research was joint funded by the Peter Harrison Centre for Disability Sport and British Wheelchair Basketball.

Declaration of conflicting interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Data Availability Statement

Anonymized data are available from the corresponding author upon reasonable request.*

References

International Wheelchair Basketball Federation (IWBF) . IWBF Player Classification Rules. https://cdn.prod.website-files.com/66a7ad387e177bc7c659bb87/67950bd14f43b4e97b4d45ae_2021-IWBF-Classification-Rules-Version-2022-12-12-vers-2%20small.pdf (2021, accessed 1 October 2024).

British Wheelchair Basketball . Eligibility Criteria for British Wheelchair Basketball. https://britishwheelchairbasketball.co.uk/resource-category/rules-and-classification/ (2023, accessed 1 October 2024).

Grossmann

Flueck

Roelands

, et al. Characteristics of official wheelchair basketball games in hot and temperate conditions. Int J Environ Res Public Health 2022; 19: 1250. https://doi.org/10.3390/ijerph19031250

De Witte

AMH

Hoozemans

MJM

Berger

MAM

, et al.

Do field position and playing standard influence athlete performance in wheelchair basketball?

J Sports Sci 2016; 34: 811–820. https://doi.org/10.1080/02640414.2015.1072641

Iturricastillo

Yanci

Granados

, et al. Quantifying wheelchair basketball match load: a comparison of heart-rate and perceived-exertion methods. Int J Sports Physiol Perform 2016; 11: 508–514. https://doi.org/10.1123/ijspp.2015-0257

Marszalek

Gryko

Prokopowicz

, et al. The physiological response of athletes with impairments in wheelchair basketball game. Hum Mov 2019; 20: 1–7.

Edwards

. High performance training and racing. In: Edwards

(ed). The Heart Rate Monitor Book. 8th ed. Feet Fleet Press, 1993, pp. 113–123.

Foster

Florhaug

Franklin

, et al. A new approach to monitoring exercise training. J Strength Cond Res 2001; 15: 109–115.

Paulson

Goosey-Tolfrey

. Current perspectives on profiling and enhancing wheelchair court sport performance. Int J Sports Physiol Perform 2017; 12: 275–286. https://doi.org/10.1123/ijspp.2016-0231

10.

Akubat

Abt

. Intermittent exercise alters the heart rate-blood lactate relationship used for calculating the training impulse (TRIMP) in team sport players. J Sci Med Sport 2011; 14: 249–253. https://doi.org/10.1016/j.jsams.2010.12.003

11.

Van der Slikke

RMA

Berger

MAM

Bregman

DJJ

, et al. Wearable wheelchair mobility performance measurement in basketball, rugby, and tennis: lessons for classification and training. Sensors 2020; 20: 1–13. https://doi.org/10.3390/s20123518

12.

Mason

Van der Slikke

RMA

Hutchinson

, et al. The effect of small-sided game formats on physical and technical performance in wheelchair basketball. Int J Sports Physiol Perform 2018; 13: 891–896. https://doi.org/10.1123/ijspp.2017-0500

13.

Van der Slikke

RMA

Mason

Berger

MAM

, et al. Speed profiles in wheelchair court sports; comparison of two methods for measuring wheelchair mobility performance. J Biomech 2017; 65: 221–225. https://doi.org/10.1016/j.jbiomech.2017.10.040

14.

Rietveld

Vegter

RJK

Van der Slikke

RMA

, et al. Wheelchair mobility performance of elite wheelchair tennis players during four field tests: Inter-trial reliability and construct validity. PLoS One 2019; 14: e0217514. https://doi.org/10.1371/journal.pone.0217514

15.

Paulson

TAW

Mason

Rhodes

, et al. Individualized internal and external training load relationships in elite wheelchair rugby players. Front Physiol 2015; 6: 388. https://doi.org/10.3389/fphys.2015.00388

16.

Rhodes

Mason

Paulson

TAW

, et al. A comparison of speed profiles during training and competition in elite wheelchair rugby players. Int J Sports Physiol Perform 2017; 12: 777–782. https://doi.org/10.1123/ijspp.2016-0218

17.

Cahill

Lamb

Worsfold

, et al. The movement characteristics of English Premiership rugby union players. J Sports Sci 2013; 31: 229–237. https://doi.org/10.1080/02640414.2012.727456

18.

Murray

Gabbett

Townshend

. The use of relative speed zones in Australian football: are we really measuring what we think we are? Int J Sports Physiol Perform 2018; 13: 442–451. https://doi.org/10.1123/ijspp.2017-0148

19.

Reardon

Tobin

Delahunt

. Application of individualized speed thresholds to interpret position specific running demands in elite professional rugby union: a GPS study. PLoS One 2015; 10: e0133410. https://doi.org/10.1371/journal.pone.0133410

20.

Van der Slikke

RMA

Berger

MAM

Bregman

DJJ

, et al. From big data to rich data: The key features of athlete wheelchair mobility performance. J Biomech 2016; 49: 3340–3346. https://doi.org/10.1016/j.jbiomech.2016.08.022

21.

Van Dijk

Van der Slikke

RMA

Rupf

, et al. Obtaining wheelchair kinematics with one sensor only? The trade-off between number of inertial sensors and accuracy for measuring wheelchair mobility performance in sports. J Biomech 2022; 130: 110879. https://doi.org/10.1016/j.jbiomech.2021.110879

22.

Borg

. Borg’s Perceived Exertion And Pain Scales. Human Kinetics, 1998.

23.

De Klerk

. Worklab: a wheelchair biomechanics mini-package. https://research.hanze.nl/en/publications/worklab-a-wheelchair-biomechanics-mini-package/ (2019, accessed 2 April 2022).

24.

Van der Slikke

RMA

Berger

MAM

Bregman

DJJ

, et al. Opportunities for measuring wheelchair kinematics in match settings; reliability of a three inertial sensor configuration. J Biomech 2015; 48: 3398–3405. https://doi.org/10.1016/j.jbiomech.2015.06.001

25.

Cohen

. Statistical Power Analysis. Curr Dir Psychol Sci 1992; 1: 98–101. https://doi.org/10.1111/1467-8721.ep10768783

26.

Hopkins

Marshall

Batterham

, et al. Progressive statistics for studies in sports medicine and exercise science. Med Sci Sports Exerc 2009; 41: 3–12. https://doi.org/10.1249/MSS.0b013e31818cb278

27.

Van der Slikke

Berger

Bregman

, et al. Push characteristics in wheelchair court sport sprinting. Procedia Eng 2016; 147: 730–734. https://doi.org/10.1016/j.proeng.2016.06.265

28.

Snyder

Goods

Peeling

, et al. Contribution of physical characteristics to game performance in male wheelchair basketball athletes at the Tokyo Paralympic Games. J Strength Cond Res 2024; 38: e579–e586. https://doi.org/10.1519/JSC.0000000000004858

29.

García-Fresneda

Carmona

. Short distance sprint performance in elite wheelchair basketball women players: influence of functional classification. Apunts Educ Fis Deporte 2021; 2: 75–80.

30.

Padrón-Cabo

Solleiro-Duran

Lorenzo-Martínez

, et al. Application of arbitrary and individualized load quantification strategies over the weekly microcycle in professional soccer players. Biol Sport 2024; 41: 153–161. https://doi.org/10.5114/biolsport.2024.129481

31.

Rago

Brito

Figueiredo

, et al. Application of individualized speed zones to quantify external training load in professional soccer. J Hum Kinet 2020; 72: 279–289. https://doi.org/10.2478/hukin-2019-0113

32.

Murray

Gabbett

Townshend

. The use of relative speed zones in Australian Football: Are we really measuring what we think we are? Int J Sports Physiol Perform 2018; 13: 442–451. https://doi.org/10.1123/ijspp.2017-0148

Comparative analysis of fixed and individualised speed zones for quantifying external match load in female wheelchair basketball

Abstract

Introduction

Methods

Results

Conclusions

Keywords

Introduction

Methods

Participants

Design

Measurements

Data analysis

IML

EML

Statistical analysis

Results

Comparison of zonal methods

The effect of classification

EML and IML

Discussion

Limitations

Practical recommendations and future directions

Conclusion

Footnotes

Acknowledgements

ORCID iDs

Ethical considerations

Consent for publication

Author contributions

Funding

Declaration of conflicting interests

Data Availability Statement

References