Inter-Limb Hand Force Asymmetry in Front Crawl Swimming: Individual Interpretation Using BAI-1, MDD, and MID

Abstract

This study examined inter-limb hand force asymmetry in competitive swimmers during front crawl at two paces (200 m and 50 m). Eight national-level swimmers (18.3 ± 2.5 years) performed 25 m trials at both intensities using wearable force-sensing paddles. Asymmetry was assessed with a traditional percentage-based index (BAI-1) and distribution-based estimates, namely the Minimal Detectable Difference (MDD) and an exploratory Minimal Important Difference (MID) estimate. Mean hand force was significantly higher at sprint pace (50 m) compared to 200 m pace (p < 0.001), confirming adherence to the prescribed intensities. Using BAI-1, swimmers showed a wide range of asymmetry values (<2% to >10%), highlighting substantial inter-individual variability. Absolute BAI-1 was descriptively higher during the 50 m pace than during the 200 m pace, although this difference was not statistically significant. Distribution-based estimates refined the interpretation: MDD corresponded to 4.48% and 4.83% for the 50 m and 200 m paces, respectively, whereas the exploratory MID estimate corresponded to 2.68% and 2.89%. These findings illustrate how BAI-1, MDD, and MID may lead to complementary interpretations of asymmetry. The combined use of traditional indices and distribution-based estimates may support more individualized monitoring of inter-limb hand force asymmetry in swimming and potentially other bilateral sports.

1. Introduction

In swimming, specifically in the front crawl stroke, propulsion is primarily generated by the upper limbs, which are known to contribute the majority of forward thrust (i.e., forward propulsion) [1]. Evidence from coordination-based analyses also supports the dominant role of upper-limb actions in producing effective propulsion [2]. The ability to generate and sustain high levels of force has been positively linked to swimming velocity, efficiency, and overall performance [3]. Beyond its mechanical role, force application also interacts with energetic cost, as inefficiencies in force production may increase metabolic demands and compromise performance [4,5]. Accordingly, a precise evaluation of the force production of each upper limb is fundamental to understanding the determinants of competitive success.

In swimming, higher intensities, such as 50 m sprint pace, typically require greater mechanical output and higher force production than middle-distance paces (e.g., 200 m), reflecting differences in stroke rate, motor control demands, and energetic requirements [6,7]. These distinctions justify analyzing upper-limb hand force production across different paces to better understand how swimmers adjust their propulsion under varying performance constraints. In front-crawl swimming, asymmetries between the right and left upper limbs may impair technical efficiency, increase energy cost, and potentially elevate injury risk due to repetitive overload [8]. However, recent evidence also suggests that moderate contralateral asymmetry may, in certain contexts, contribute positively to stroke coordination and propulsion rather than hinder performance [9]. Therefore, the literature remains inconsistent: while some studies report strong associations between greater asymmetries and poorer performance outcomes [10,11,12,13], others have found negligible or unclear effects [14,15,16,17]. This lack of consensus underscores the need for more robust and standardized approaches to quantify and interpret asymmetry in aquatic contexts.

Beyond its potential biomechanical and energetic implications, inter-limb asymmetry has been increasingly debated as a marker of performance capacity and injury risk in elite sport [18,19]. Evidence from rehabilitation settings also illustrates that persistent inter-limb biomechanical differences may be associated with altered joint loading and internal tissue stress distribution following injury [20]. However, in swimming-specifically, available evidence does not clearly support that inter-limb asymmetry should be interpreted as universally detrimental; rather, its relevance appears to be primarily performance-related, influencing propulsion efficiency and energetic cost. Some evidence suggests that higher asymmetry levels may compromise propulsion efficiency and limit maximal velocity attainment, particularly in cyclical sports where symmetrical coordination is advantageous [21]. Conversely, other research indicates that moderate asymmetries may not necessarily impair performance and, in some cases, could even reflect functional adaptations to sport-specific demands [22,23]. This ongoing debate highlights a fundamental challenge for applied practitioners: without clear reference values or consistent methodologies, it remains difficult to determine when an asymmetry is detrimental, benign, or adaptive. In swimming, where small margins in propulsive efficiency can separate medalists from non-finalists [24], clarifying this distinction is particularly relevant for advancing both scientific knowledge and applied practice.

In most sport-related studies, asymmetry is quantified using the symmetry index (SI), the Bilateral Asymmetry Index (BAI), or similar percentage-based measures, typically based on the relative difference between limbs (e.g., dominant vs. non-dominant, right vs. left) [25]. The numeral in BAI-1 identifies the specific algebraic formulation used, distinguishing it from alternative formulations such as BAI-2. The equation applied in the present study is provided in the Inter-limb Asymmetry subsection. These indices are simple, widely used, and useful for describing the magnitude of inter-limb differences. When signed values are retained, they can also indicate the direction of asymmetry, allowing researchers and practitioners to determine whether the right or left limb yields higher values. This is particularly relevant in swimming, where dominance patterns may vary across task conditions and where group averages may mask individual responses.

Nevertheless, percentage-based asymmetry indices do not, by themselves, indicate whether an observed difference is statistically detectable or potentially meaningful within a given dataset. Their interpretation is often based on fixed thresholds, commonly around 10–15%, despite limited empirical support for these cut-offs [26,27,28]. Consequently, the same BAI-1 value may be interpreted differently depending on the athlete, task, measurement variability, and performance context. Recent reviews have therefore questioned the use of fixed cut-offs as universal decision rules, emphasizing that traditional indices may either overestimate or underestimate the practical relevance of between-limb differences when used in isolation [19,21,29].

Another unresolved question concerns how inter-limb hand force asymmetry varies with swimming intensity. Evidence based on coordination metrics (IdC) shows that swimmers reduce glide and adopt more continuous arm coordination at higher speeds, particularly in elite cohorts [30,31,32], but these studies did not quantify between-limb hand force and therefore cannot be interpreted as changes in force asymmetry. By contrast, studies that directly assessed limb-level force report mixed findings: some observed persistent or even greater force asymmetry under higher-intensity or sprint/tethered conditions [10,33], whereas others reported large asymmetries without clear performance effects [17]. A recent scoping review highlighted the heterogeneity of results and the limited direct evidence on how upper-limb force asymmetry varies with intensity [34]. These inconsistencies justify examining inter-limb hand force within subjects, across submaximal and maximal paces in competitive swimmers, rather than relying solely on group-level comparisons.

In this context, BAI-1 may be particularly useful because it allows quantification of the magnitude and direction of asymmetry for each swimmer and comparison across different pacing conditions. Therefore, beyond describing whether right- and left-hand force differ at the group level, BAI-1 can help identify whether individual asymmetry profiles remain stable or change between a 50 m sprint pace and a 200 m pace. This is relevant because an increase in swimming speed may not necessarily produce a more symmetrical force pattern. Higher stroke frequency could theoretically reduce the temporal window for side-to-side discrepancies, but sprint intensity may also increase mechanical and neuromuscular demands, potentially amplifying individual differences in force application. Thus, whether asymmetry decreases, increases, or remains stable across paces remains an open question.

Given the interpretative limitations of using percentage-based indices alone, complementary approaches widely used in other research areas may help refine the interpretation of asymmetry in sports performance contexts. In clinical, rehabilitation, and health sciences, distribution-based metrics have been extensively used to determine whether observed differences are large enough to be statistically detectable or potentially meaningful [35]. These include the Minimal Detectable Difference (MDD) [36], which can be used to evaluate whether between-limb differences are statistically detectable, and the Minimal Important Difference (MID), which estimates the smallest change with practical or clinical significance [37]. Although these metrics were originally developed and more commonly applied outside sports performance research, they may offer an interesting complementary framework for interpreting asymmetry in sport. Although MID is well established in clinical research, its application in sports performance and biomechanics appears to remain scarce. Available examples include its use to contextualize biomechanical changes during a 90° cutting task in collegiate soccer players and the recent establishment of an anchor-based MID for a visual analog scale assessing exercise-induced fatigue [38,39]. To the best of our knowledge, MID has not previously been applied to inter-limb hand-force asymmetry in swimming. Therefore, the present distribution-based MID estimate should be interpreted as exploratory and sample-specific and requires future validation against swimming-specific external anchors. In this context, MDD and MID do not replace traditional asymmetry indices such as BAI-1. Instead, they may complement BAI-1 by providing reference values that help assess whether the asymmetry quantified by BAI-1 exceeds the data’s variability or reaches a potentially relevant magnitude.

A further methodological gap is that no study has integrated BAI-1 with MDD and an exploratory MID estimate to evaluate inter-limb hand force asymmetry in swimming, a context where upper-limb force production is directly linked to velocity and efficiency and thus provides a unique setting to test this application. By bridging methodological frameworks from clinical research with sports performance, it becomes possible to generate new insights into the interpretation of asymmetry and its relevance for competitive swimmers. Therefore, the present study aimed to examine inter-limb hand force asymmetry in competitive swimmers across two front crawl intensities and, as its primary methodological contribution, to compare the interpretations provided by BAI-1, MDD, and an exploratory MID estimate. Specifically, the study sought to: (i) confirm whether the 50 m and 200 m paces represented distinct swimming intensities; (ii) compare hand force between paces and sides using a repeated-measures design; (iii) compare the magnitude and direction of asymmetry between paces using BAI-1; and (iv) explore whether asymmetries identified by BAI-1 were similarly interpreted when contextualized by MDD and an exploratory MID estimate. It was expected that the 50 m pace would elicit higher swimming speed, stroke frequency, and hand force than the 200 m pace. Given the exploratory nature of applying MDD and MID to swimming asymmetry, and the uncertainty about how asymmetry varies across intensities, no directional hypothesis was established regarding the agreement between metrics or changes in asymmetry across paces.

2. Materials and Methods

2.1. Participants

Eight competitive swimmers (7 males and 1 female; decimal age: 18.3 ± 2.5 years, range: 16.41–24.18 years; body mass: 74.5 ± 10.5 kg; height: 180.6 ± 8.7 cm; arm span: 181.8 ± 7.9 cm) were recruited for analysis. Hand dominance was determined by asking each swimmer which hand they habitually used for writing. Seven swimmers reported right-hand dominance and one reported left-hand dominance. They were selected from a national team that regularly competed at national and international events. At the time of data collection, their World Aquatic points were 560.71 ± 89.12 for 50 m and 567.33 ± 142.84 for 200 m in long course meter competitions. All swimmers held official national rankings in front crawl events, providing a reproducible indication of their competitive level. The sample included age-group national record holders, national champions, and swimmers who were part of a national talent identification program (Tier 3 athletes) [40]. All swimmers were actively competing at the national level, with regular participation in age-group and senior championships and followed a structured training program of six to nine sessions per week. To be eligible, swimmers were required to be proficient in the front crawl stroke, free of musculoskeletal injuries during the preceding six months, and accustomed to maximal effort testing protocols. Technical proficiency in front crawl was confirmed by their national-team coaches, as specialization in a single stroke is common in competitive swimming and does not guarantee equal proficiency across all techniques. Although the sample size is small, it reflects the typical availability of high-performance swimmers in force-based experimental studies. The study was designed using within-subject comparisons, which reduces inter-individual variability and increases the robustness of the statistical estimates despite the limited sample. All participants and their legal guardians (for those under 18 years old) received a comprehensive explanation of the study procedures and provided written informed consent prior to participation. The study complied with the ethical standards of the Declaration of Helsinki and was approved by the Ethics Committee of the Polytechnic Institute of Bragança (No. P547664-R678573-D2082555, 27 March 2025).

2.2. Study Design

This study followed a repeated-measures design conducted in a 25 m swimming pool, in which each swimmer completed two 25 m front crawl trials under controlled conditions: one at submaximal intensity, corresponding to the individual’s 200 m pace, and one at maximal intensity, corresponding to the 50 m sprint pace. Following a standardized warm up routine commonly used in competitive swimming [41], swimmers performed the two 25 m front crawl trials under controlled conditions. For the submaximal condition, swimmers were instructed to swim at a pace consistent with their typical 200 m front crawl intensity. For the sprint condition, swimmers were instructed to swim at maximal effort. All trials were supervised by the national team coach, who monitored pacing and technical execution to ensure adherence to the prescribed intensities. The submaximal trial was performed first and also served as an additional warm up stimulus before the sprint trial. After a 30 min recovery period, swimmers completed the maximal trial. This fixed order was selected to avoid residual fatigue from the maximal effort affecting the submaximal condition. Randomization or counterbalancing was therefore not adopted because assigning the maximal trial first to some swimmers could have introduced residual fatigue into the subsequent submaximal trial. All trials were initiated with a push-off from the pool wall, with swimmers instructed to begin stroking immediately without performing an underwater glide phase. Data collection was restricted to the 10–20 m section of the 25 m pool to minimize the influence of the push-off and the approach to the opposite wall. Swimmers were instructed to perform the stroke cycles while holding their breath to avoid disruptions in stroke coordination and to maintain consistent technique [42]. The following variables were measured during both swimming paces: swimming speed, stroke frequency (SF), force of the right (F_right) and left (F_left) upper limbs, and overall mean hand force (F_average). The average of three consecutive stroke cycles selected within the 10–20 m section was used for further analysis.

2.3. Kinetic Data

Force values derived from the hand-worn sensors were referred to as hand force to distinguish the measured/estimated output from the total propulsive force generated by the whole upper limb. Hand force (N) was assessed using wearable force-sensing paddles (PoolShark, Tampere, Finland), which represent the updated generation of wearable paddles derived from earlier Trainesense technology. The paddles were secured to the swimmers’ hands with adjustable silicone straps integrated into the device, ensuring a firm, comfortable fit without restricting natural hand movement or altering stroke mechanics. The system captures in-water force production during swimming strokes using two integrated pressure sensors and a 9-axis IMU at a sampling rate of 100 Hz.

The PoolShark Session Manager mobile application was used to record data and upload it to the Analysis Center (https://sharksensors.com/; accessed on 13 September 2026), where the recordings were processed and stored. The processed data were generated using the manufacturer’s proprietary algorithm. Similar wearable force-sensing systems have been used in swimming biomechanics research to monitor in-water force variables and to examine their relationship with swimming velocity [43]. For each participant, forces of the right (F_right) and left (F_left) upper limbs were recorded across three consecutive stroke cycles in both testing conditions (50 m sprint and 200 m middle-distance pace). The mean values of F_right and F_left were retained for analysis, as averaging several consecutive cycles reduces intra-cycle noise and yields a more stable and reliable estimate of limb force production, a standard practice in swimming biomechanics. An overall mean hand force (F_average) was also computed as the arithmetic mean of both limbs.

2.4. Kinematic Data

Swimming speed (m·s⁻¹) and stroke frequency (Hz) were measured using a mechanical speedometer (SpeedRT, ApLab, Rome, Italy) attached to the swimmers’ waist by a string, sampling at 100 Hz. The speedometer continuously recorded displacement and velocity throughout the trials. These recordings were used to verify that participants performed the sprint (50 m) and submaximal (200 m pace) trials at the intended intensities. To confirm pacing, swimming speed values obtained from the speedometer were compared between conditions, ensuring higher velocity during the sprint trial and a controlled velocity consistent with the typical 200 m training pace during the submaximal trial. Stroke frequency (SF) was also examined, as lower SF is characteristic of middle-distance pacing and higher SF is expected at maximal sprint intensity. Together, these metrics allowed us to confirm that swimmers adhered to the prescribed intensities. In addition, two cameras were employed to assist with stroke identification and synchronization of data sources: one positioned outside the pool (GoPro Hero 7, GoPro, San Mateo, CA, USA) to capture the sagittal plane, and one positioned underwater to record hand entry and exit [44]. Both cameras were synchronized with the speedometer by means of a light-emitting diode flash embedded in the system, which provided a visible timestamp in the video. During post-processing, the video data were used to determine stroke entry and exit points and to align these with the force–time curves from the wearable sensors. Stroke frequency (SF) was computed as the number of stroke cycles per unit of time (1/t), based on the three consecutive cycles, and expressed in Hz. Swimming speed and SF were then used to confirm trial pacing, while the synchronized video recordings ensured accurate alignment of hand force data with stroke events. The stroke frequency values used in the analyses were those automatically computed by the PoolShark system. The synchronized video recordings served only to validate these values and to ensure accurate identification of stroke events for alignment with the force–time curves.

2.5. Statistical Analysis

Data processing and statistical analyses were performed using IBM SPSS Statistics version 29 and Microsoft Excel (Microsoft Corporation, Redmond, WA, USA; https://www.microsoft.com/microsoft-365/excel; accessed on 13 September 2026). Descriptive statistics (mean ± standard deviation) were computed for each variable under both swimming paces (50 m: sprint; 200 m: middle distance pace). Normality of the paired differences and repeated measures contrasts was assessed using the Shapiro–Wilk test. Swimming speed, stroke frequency, and stroke length were compared between paces using paired sample t tests, as the same swimmers performed both conditions. These comparisons were used as manipulation checks to verify whether swimmers performed the trials at the intended intensities. Cohen’s d was reported as the standardized effect size for paired comparisons and was calculated as the mean paired difference divided by the standard deviation of the paired differences. The effect size was interpreted as: (i) trivial if 0 ≤ d < 0.20; (ii) small if 0.20 ≤ d < 0.60; (iii) moderate if 0.60 ≤ d < 1.20; (iv) large if 1.20 ≤ d < 2.00; (v) very large if 2.00 ≤ d < 4.00; and (vi) nearly distinct if d ≥ 4.00 [45].

Mean hand force was analyzed using a two-way repeated measures ANOVA with “pace” (50 m and 200 m) and “side” (right and left) as within-subject factors. The model tested the main effect of pace, the main effect of side, and the pace × side interaction. Since both within subject factors included only two levels, the assumption of sphericity was automatically satisfied. The overall mean force of both limbs (F_average) was computed for descriptive purposes only and was not included in the ANOVA model. The level of significance was set at α = 0.05. For the ANOVA effect size index, partial eta squared (η_p²) was computed and interpreted as: (i) without effect if 0 < η_p² < 0.04; (ii) minimum if 0.04 < η_p² < 0.25; (iii) moderate if 0.25 < η_p² < 0.64; and (iv) strong if η_p² > 0.64 [46].

2.6. Inter-Limb Asymmetry

Inter-limb asymmetry of hand force between the right and left upper limbs was calculated for each swimming pace using the Bilateral Asymmetry Index 1 (BAI-1), as proposed by Bishop et al. [19]. The numeral identifies the specific algebraic formulation used and distinguishes BAI-1 from alternative formulations such as BAI-2. The asymmetry was expressed as the relative percentage difference between the right and left limbs, calculated as:

$$\mathrm{A}\mathrm{s}\mathrm{y}\mathrm{m}\mathrm{m}\mathrm{e}\mathrm{t}\mathrm{r}\mathrm{y}\left(\mathrm{\%}\right)=\left({\displaystyle \frac{{\mathrm{F}}_{\mathrm{r}\mathrm{i}\mathrm{g}\mathrm{h}\mathrm{t}}-{\mathrm{F}}_{\mathrm{l}\mathrm{e}\mathrm{f}\mathrm{t}}}{{\mathrm{F}}_{\mathrm{r}\mathrm{i}\mathrm{g}\mathrm{h}\mathrm{t}}+{\mathrm{F}}_{\mathrm{l}\mathrm{e}\mathrm{f}\mathrm{t}}}}\right)\times 100$$

where F_right and F_left represent the mean force values of the right and left hands, respectively. Positive values indicate greater force production by the right hand, whereas negative values indicate greater force production by the left hand. In the present study, signed asymmetry values were retained to preserve the direction of asymmetry. Absolute BAI-1 values were also computed to describe the magnitude of asymmetry independently of direction, since signed group means may mask individual asymmetry patterns when swimmers present opposite directions of asymmetry.

To examine whether the interpretation obtained from BAI-1 was consistent with distribution-based estimates, the Minimal Detectable Difference (MDD) and an exploratory Minimal Important Difference estimate (MID) were calculated separately for each swimming pace. These distribution-based estimates were not used as definitive diagnostic cut-offs for individual swimmers, but as exploratory reference values to determine whether asymmetries identified by BAI-1 were similarly interpreted when contextualized by statistical detectability and potential practical relevance.

The MDD was calculated as the minimum right-left difference that would be statistically detectable at each pace, following the general concept that MDD represents the minimum difference required to detect a statistically significant effect given the variability of the data [47]. Although right- and left-hand force values were analyzed as separate side conditions, they were not statistically independent because both values were obtained from the same swimmers. Therefore, side was treated as a within-subject factor, and the MDD was calculated using a paired formulation [36]:

$$\mathrm{M}\mathrm{D}\mathrm{D}=t_{1-\mathsf{\alpha }/2,\mathrm{n}-1}{\displaystyle \frac{S_D}{\surd n}}$$

where MDD is the Minimal Detectable Difference, t_{1−α/2,n−1} is the critical value of the t distribution for a two-tailed test, $S_D$ is the standard deviation of the paired right-left hand-force differences, and n is the number of independent paired observations (swimmers). The MDD was interpreted as a statistical detectability threshold for the right–left contrast, not as a direct estimate of instrument measurement error. Minimal Detectable Change was not calculated because the present study did not include repeated testing sessions or test–retest reliability data.

The exploratory MID estimate was calculated using a distribution-based approach:

$$\mathrm{MID}=0.5\times S_D$$

where $S_D$ is the standard deviation of the paired right-left force differences within each pace. This value was interpreted as an exploratory estimate of potential practical relevance, not as a validated minimal important difference. This interpretation was adopted because no swimming specific anchor-based threshold was available, and distribution-based estimates should not replace empirically established anchor-based MID values [37].

MDD and MID were first calculated in newtons (N). To allow direct descriptive comparison with BAI-1, both values were additionally expressed as percentages using the same denominator structure as the BAI-1 equation. Specifically, MDD and MID were divided by the mean bilateral hand-force sum for the corresponding pace and multiplied by 100. This represents a scale-harmonization procedure specific to the present analysis:

$$\mathrm{M}\mathrm{D}\mathrm{D}\left(\mathrm{\%}\right)={\displaystyle \frac{\mathrm{M}\mathrm{D}\mathrm{D}\left(\mathrm{N}\right)}{\overline{F_{\mathrm{r}\mathrm{i}\mathrm{g}\mathrm{h}\mathrm{t}}+F_{\mathrm{l}\mathrm{e}\mathrm{f}\mathrm{t}}}}}\times 100$$

$$\mathrm{M}\mathrm{I}\mathrm{D}\left(\mathrm{\%}\right)={\displaystyle \frac{\mathrm{M}\mathrm{I}\mathrm{D}\left(\mathrm{N}\right)}{\overline{F_{\mathrm{r}\mathrm{i}\mathrm{g}\mathrm{h}\mathrm{t}}+F_{\mathrm{l}\mathrm{e}\mathrm{f}\mathrm{t}}}}}\times 100$$

where MDD(N) and MID(N) represent the values originally calculated in newtons, and mean(F_right + F_left) represents the mean bilateral hand-force sum for the corresponding pace. This conversion was used only for descriptive comparison with BAI-1 and was not intended to define new asymmetry thresholds. Finally, the agreement between BAI-1 and the distribution-based estimates was explored descriptively. Each swimmer’s absolute BAI-1 value was compared with the corresponding MID and MDD values expressed on the BAI-1 percentage scale. Four interpretive outcomes were considered: (i) below both reference values; (ii) exceeding only the exploratory MID estimate; (iii) exceeding only the MDD value; or (iv) exceeding both values. No inferential test was applied to these classifications, as they were used only to describe whether BAI-1, MDD, and MID provided convergent or divergent interpretations of inter-limb hand force asymmetry.

3. Results

The manipulation check confirmed that the two trials represented distinct swimming intensities (

Table 1. Manipulation check for the 50 m and 200 m swimming paces.

Variable	50 m Pace Mean ± SD	200 m Pace Mean ± SD	Mean Difference	t(7)	p	d
Swimming speed (m·s−1)	1.69 ± 0.14	1.36 ± 0.18	0.33	5.75	<0.001	2.03
Stroke frequency (Hz)	0.98 ± 0.07	0.63 ± 0.06	0.35	9.03	<0.001	3.19
Stroke length (m)	1.73 ± 0.15	2.17 ± 0.29	−0.45	−7.20	<0.001	−2.54

Note: Values are presented as mean ± SD. Mean difference refers to the difference between the 50 m and 200 m paces. d = Cohen’s d for paired comparisons, calculated using the standard deviation of the paired differences.

). Compared with the 200 m pace, the 50 m pace showed significantly higher swimming speed (1.69 ± 0.14 vs. 1.36 ± 0.18 m·s⁻¹; t(7) = 5.75, p < 0.001, d = 2.03) and stroke frequency (0.98 ± 0.07 vs. 0.63 ± 0.06 Hz; t(7) = 9.03, p < 0.001, d = 3.19). Conversely, stroke length was significantly lower during the 50 m pace compared with the 200 m pace (1.73 ± 0.15 vs. 2.17 ± 0.29 m; t(7) = −7.20, p < 0.001, d = −2.54).

Mean hand force differed significantly between swimming paces (

Table 2. Descriptive statistics of hand force during the 50 m and 200 m paces and results of the two-way repeated measures ANOVA.

Descriptive Statistics (Mean ± SD)
Variable	50 m Pace		200 m Pace
Right hand force (N)	45.73 ± 3.01		41.09 ± 5.67
Left hand force (N)	50.24 ± 6.89		41.37 ± 5.39
Average hand force (N)	47.99 ± 4.66		41.23 ± 4.99
Two-Way Repeated Measures ANOVA (Pace × Side)
Effect	F(1, 7)	p	ηp2
Pace	31.31	<0.001	0.817
Side	2.65	0.148	0.274
Pace × Side	4.94	0.062	0.414

Note: Values are presented as mean ± SD. η_p² = partial eta squared. Average hand force was computed for descriptive purposes only and was not included as a dependent variable in the repeated measures ANOVA.

). The two-way repeated measures ANOVA showed a significant main effect of pace (F(1, 7) = 31.31, p < 0.001, η_p² = 0.817), with the estimated marginal mean decreasing from 47.99 N in the 50 m pace to 41.23 N in the 200 m pace. In contrast, no significant main effect of side was observed (F(1, 7) = 2.65, p = 0.148, η_p² = 0.274), indicating that, when both paces were considered together, right- and left-hand force did not differ significantly. The pace × side interaction was not statistically significant (F(1, 7) = 4.94, p = 0.062, η_p² = 0.414). However, the descriptive values showed a larger difference between the right and left hands during the 50 m pace, with the left hand presenting higher force than the right hand (50.24 ± 6.89 vs. 45.73 ± 3.01 N). During the 200 m pace, right and left-hand force values were very similar (41.09 ± 5.67 vs. 41.37 ± 5.39 N).

Inter-limb hand force asymmetry was further examined using BAI-1, MDD, and the exploratory MID estimate (

Table 3. Individual BAI-1 values and comparison with MDD and MID estimates for hand force asymmetry at both swimming paces.

Panel A. Distribution-Based Estimates by Pace
Pace	${\mathit{S}}_{\mathit{D}}$ (N)	MDD (N)	MID (N)	Mean Bilateral Sum (N)	MDD (%)	MID (%)
50 m pace	5.14	4.30	2.57	95.97	4.48	2.68
200 m pace	4.77	3.98	2.38	82.47	4.83	2.89
Panel B. Summary Comparison of BAI-1 Between Paces
Variable	50 m Pace Mean ± SD		200 m Pace Mean ± SD	t(7)	p	d
BAI-1 signed (%)	−4.37 ± 5.12		−0.37 ± 6.00	−1.99	0.087	−0.70
Absolute BAI-1 (%)	5.50 ± 3.67		4.26 ± 3.92	0.72	0.493	0.26
Panel C. Individual BAI-1 Values and Interpretation
Swimmer ID	BAI-1 Signed 50 m (%)	Absolute BAI-1 50 m (%)	50 m Interpretation	BAI-1 Signed 200 m (%)	Absolute BAI-1 200 m (%)	200 m Interpretation
1	−1.32	1.32	Below both	8.12	8.12	Exceeds both
2	−3.68	3.68	Exceeds MID only	−3.23	3.23	Exceeds MID only
3	−1.35	1.35	Below both	0.10	0.10	Below both
4	−10.09	10.09	Exceeds both	−11.91	11.91	Exceeds both
5	4.51	4.51	Exceeds both	3.57	3.57	Exceeds MID only
6	−4.09	4.09	Exceeds MID only	−3.37	3.37	Exceeds MID only
7	−10.13	10.13	Exceeds both	0.55	0.55	Below both
8	−8.81	8.81	Exceeds both	3.25	3.25	Exceeds MID only

Note: BAI-1 = Bilateral Asymmetry Index 1; $S_D$ = standard deviation of the paired differences between right- and left-hand force; MDD = Minimal Detectable Difference; MID = exploratory Minimal Important Difference estimate. MDD and MID were first calculated in newtons and then expressed as percentages using the same denominator structure as BAI-1 to allow direct descriptive comparison with absolute BAI-1. “Below both” indicates that absolute BAI-1 was below both MID and MDD values; “Exceeds MID only” indicates that absolute BAI-1 exceeded MID but not MDD; “Exceeds both” indicates that absolute BAI-1 exceeded both MID and MDD. These interpretations were used for descriptive purposes only and should not be considered diagnostic cut-offs.

). The magnitude of inter-limb hand force asymmetry, expressed as absolute BAI-1, was descriptively higher during the 50 m pace than during the 200 m pace, although this difference was not statistically significant (5.50 ± 3.67% vs. 4.26 ± 3.92%; t(7) = 0.72, p = 0.493, d = 0.26). Signed BAI-1 values showed a more negative mean value during the 50 m pace compared with the 200 m pace, indicating a greater left-hand bias at the sprint pace; however, this difference also did not reach statistical significance (−4.37 ± 5.12% vs. −0.37 ± 6.00%; t(7) = −1.99, p = 0.087, d = −0.70). At the individual level, left-hand force was higher in seven of eight swimmers during the 50 m pace and in three of eight swimmers during the 200 m pace. When hand dominance was considered descriptively, all swimmers produced greater mean hand force with the non-dominant hand during the 50 m pace, whereas no consistent dominance-related pattern was observed during the 200 m pace. This post hoc observation was not subjected to inferential testing.

The distribution-based estimates differed slightly between paces. For the 50 m pace, the MDD was 4.30 N, corresponding to 4.48% on the BAI-1 scale, whereas the exploratory MID estimate was 2.57 N, corresponding to 2.68%. For the 200 m pace, the MDD was 3.98 N, corresponding to 4.83%, whereas the exploratory MID estimate was 2.38 N, corresponding to 2.89%.

The comparison between absolute BAI-1 and the distribution-based estimates showed that the interpretation of asymmetry varied across swimmers and paces. During the 50 m pace, two swimmers were below both estimates, two exceeded only the exploratory MID estimate, and four exceeded both MID and MDD values. During the 200 m pace, two swimmers were below both estimates, four exceeded only the exploratory MID estimate, and two exceeded both values. These findings indicate that some asymmetries identified by BAI-1 were also supported by both distribution-based estimates, whereas others exceeded only the exploratory MID estimate or remained below both values.

4. Discussion

The present study examined inter-limb hand force asymmetry in competitive swimmers across two front crawl intensities using BAI-1, MDD, and an exploratory MID estimate. The main findings were: (i) the 50 m pace was clearly distinguished from the 200 m pace by higher swimming speed and stroke frequency, but lower stroke length; (ii) mean hand force was significantly higher during the 50 m pace than during the 200 m pace; (iii) no statistically significant main effect of side or pace × side interaction was observed, although the descriptive pattern suggested a larger difference between right and left hand force during the 50 m pace; and (iv) the interpretation of asymmetry varied across swimmers when absolute BAI-1 values were compared with MDD and MID reference values. Overall, these findings suggest that group-level comparisons alone may be insufficient to interpret inter-limb hand force asymmetry in competitive swimmers, particularly when individual responses differ in both magnitude and direction.

As expected, swimmers produced greater mean hand force during the 50 m pace than during the 200 m pace. This result is consistent with the increased mechanical demands of sprint swimming, where higher velocity and stroke frequency require greater force production by the upper limbs [7]. Previous research has shown that swimming velocity is strongly associated with force application, stroke mechanics, and the capacity to generate propulsion effectively during front crawl [7,48]. More recent hydrodynamic perspectives also emphasize that front crawl velocity emerges from the interaction between propulsive and resistive forces, and that swimmers typically increase stroke frequency to increase velocity, although this strategy is constrained by the ability to maintain effective hand orientation and propulsive force production [49]. In addition, wearable sensor-based assessments have shown that propulsive force, stroke length, and drag-related variables contribute to swimming velocity, supporting the relevance of monitoring kinetic outputs during free swimming conditions [43]. In the present study, the manipulation check confirmed that the two trials represented distinct intensities, with the 50 m pace characterized by higher speed and stroke frequency, and the 200 m pace characterized by longer stroke length. Therefore, the observed increase in mean hand force at the 50 m pace supports the validity of the experimental contrast and confirms that swimmers modulated their force output according to task intensity.

A secondary objective was to explore whether inter-limb hand force asymmetry changed with swimming intensity. The initial rationale was that higher swimming speed and stroke frequency could reduce the temporal window for side-to-side discrepancies and, therefore, promote a more symmetrical hand force pattern. This rationale was partly supported by coordination-based studies showing that front crawl swimmers modify arm coordination as velocity increases, generally moving from catch-up patterns toward more continuous opposition or superposition patterns [30]. However, the present findings did not support a statistically confirmed reduction in asymmetry during the 50 m pace. The pace × side interaction did not reach statistical significance, and absolute BAI-1 did not differ significantly between paces. Nevertheless, the descriptive values suggested that differences between the right and left hands were more apparent during the 50 m pace, with the left hand presenting higher force than the right hand at sprint intensity, whereas both sides were very similar during the 200 m pace. This pattern should be interpreted cautiously, but it does not support the assumption that higher speed necessarily leads to greater inter-limb symmetry in hand force production.

One possible explanation is that higher swimming intensity may amplify individual force application strategies rather than homogenize inter-limb behavior. High-intensity front crawl is accompanied by increased stroke rate and changes in the power phase of the stroke cycle, which may modify the timing and magnitude of force application [33,49]. In addition, upper-limb kinetic imbalances have been reported during maximal front crawl swimming, supporting the possibility that some swimmers may rely more strongly on one limb under maximal conditions [48]. This may help explain why some swimmers showed greater asymmetry during the 50 m pace, despite the higher stroke frequency. Previous research has shown that bilateral asymmetry in front crawl may depend on exercise intensity and stroke phase, suggesting that asymmetry is not a fixed property but may vary according to task demands [33]. Similarly, tethered swimming research has shown that propulsive force asymmetries can be detected between body sides, and that better-performing swimmers may present lower asymmetry than their counterparts [10]. However, the literature remains heterogeneous, with some studies reporting meaningful relationships between asymmetry and performance, while others have reported unclear or non-significant associations [11,17,34]. Therefore, the present findings are consistent with the idea that the relationship between speed, force production, and asymmetry is complex and may depend on the swimmer, the testing condition, and the method used to quantify asymmetry.

This complexity may also be explained by the fact that coordination symmetry and force symmetry are not necessarily equivalent. A swimmer may adopt a more continuous or synchronized arm coordination pattern at higher speed while still producing unequal hand force between limbs. Studies on front crawl coordination show that swimmers adapt their arm coordination with changes in velocity, stroke rate, and stroke length [30], while research on inter-limb coordination and energy cost suggests that different coordination patterns may be adopted according to task constraints and individual preferences [32]. However, these studies mainly address coordination and energetic organization, not direct side-to-side hand force production. Thus, they should not be interpreted as direct evidence of force symmetry. Instead, they help explain why changes in swimming intensity may alter the motor organization of the stroke without necessarily producing symmetrical kinetic outputs. Therefore, the present findings suggest that the relationship between speed and asymmetry is more complex than initially expected.

The lack of statistical significance for the pace × side interaction and BAI-1 comparisons may also be partly explained by the short analyzed segment and the limited number of stroke cycles retained for analysis. The use of three consecutive stroke cycles is consistent with previous swimming biomechanics research and can provide a stable estimate of mean force production within a controlled segment [3,48]. However, asymmetry may fluctuate across a full trial or race distance, particularly when swimmers adjust breathing, maintain velocity, manage fatigue, or alter coordination strategies over time. This is relevant because recent work using continuous analysis has shown that discrete average values may not fully capture changes occurring at specific moments of the arm-pull, with Statistical Parametric Mapping providing more sensitive outputs than average-based comparisons [3]. Thus, while the present data provide a controlled snapshot of hand force asymmetry, they may not capture the full temporal variability of asymmetry across longer distances or across the complete force–time curve. Future studies using longer continuous recordings, a greater number of analyzed cycles, repeated segments across the same trial, or continuous analyses of the force–time curve may help determine whether the descriptive tendency observed here becomes more pronounced under longer or more fatiguing conditions.

The present findings also reinforce the importance of considering individual asymmetry profiles rather than relying solely on group-level averages. Although the main effect of side was not statistically significant, individual BAI-1 values showed considerable variability in both magnitude and direction (Table 3, Panel C). Some swimmers presented very small asymmetries, whereas others exceeded 10% in at least one pace. Moreover, the direction of asymmetry was not uniform across swimmers, and in some cases differed between intensities. An additional exploratory observation was that all swimmers produced greater mean hand force with the non-dominant hand during the 50 m pace, whereas no consistent dominance-related pattern was observed during the 200 m pace. Previous maximal front-crawl research reported higher mean thrust for the dominant upper limb but a non-significantly higher peak thrust for the non-dominant limb, which was cautiously interpreted as a possible compensatory response to lower propelling efficiency [48]. Given that this post hoc observation concerns mean hand force and was derived from a small sample including only one left-handed swimmer, it should not be interpreted as evidence of a systematic dominance effect. Future studies should examine whether dominance-related force patterns vary with swimming intensity and with the specific kinetic variable considered. This is consistent with previous asymmetry literature showing that inter-limb asymmetry is often task-specific, athlete-specific, and sensitive to the metric used for its calculation [19,25,29,50]. In swimming, this individual variability is particularly relevant because upper-limb force production depends on hand trajectory, hand orientation, coordination, and swimmer-specific technical strategies [49,51]. Therefore, a non-significant group-level side effect does not necessarily imply that asymmetry is irrelevant at the individual level.

The interpretation of BAI-1 values was further refined by comparing them with MDD and the exploratory MID estimate. This was a central contribution of the present study. Traditional percentage-based asymmetry indices are simple and widely used, but their interpretation often depends on arbitrary cut-offs, commonly around 10–15%, with limited empirical support [25,52]. Previous methodological work has emphasized that the calculation and interpretation of inter-limb asymmetry require caution, as different equations, reference limbs, and reporting strategies can yield different conclusions [19,25]. Recent evidence from maximal front-crawl swimming further supports this distinction. Morais and Pyne [36] reported a mean BAI-1 of 7.3% across the sample, which exceeded the sample-specific MDD of 5.8%, indicating that the observed upper-limb force asymmetry was greater than the minimum difference statistically detectable under the sampling conditions. However, BAI-1 was not retained as a significant predictor of swimming velocity, suggesting that a statistically detectable asymmetry does not necessarily imply a clear association with performance. The present results further illustrate this distinction. In the 50 m pace, two swimmers were below both reference values, two exceeded only the exploratory MID estimate, and four exceeded both MID and MDD. In the 200 m pace, two swimmers were below both values, four exceeded only the exploratory MID estimate, and two exceeded both values. Therefore, the same BAI-1 value may lead to different interpretations depending on whether it is considered only as a percentage magnitude or contextualized using distribution-based estimates.

Importantly, MDD and MID should not be interpreted as diagnostic cut-offs in the present study. The MDD was used as a statistical detectability reference for the paired right-left hand force contrast, rather than as a direct estimate of instrument measurement error. Minimal Detectable Change was not calculated because the present study did not include repeated testing sessions or test–retest reliability data. Similarly, the MID was treated as an exploratory distribution-based estimate of potential practical relevance, not as a validated minimal important difference for swimming asymmetry. This distinction is essential because MID is traditionally used in clinical and longitudinal contexts, often supported by anchor-based information, and it cannot be assumed that distribution-based estimates alone reflect true practical importance [37]. In the present study, no swimming-specific anchor-based MID was available. Therefore, the MID estimate was used only to support descriptive interpretation and to illustrate how a lower reference value may identify asymmetries that do not exceed the higher MDD reference value. Repeated-session reliability has been examined for related in-water hand-force measurement systems. Campos et al. [53] reported excellent test–retest reliability for peak and mean force measured using SmartPaddle technology during tethered swimming (ICC = 0.96 and 0.95, respectively), whereas Santos et al. [54] reported similarly high reliability and measurement-error estimates for hand resultant force measured using the Aquanex pressure-sensor system during free swimming (ICC = 0.92–0.97). Marinho et al. [55] also reported good agreement between SmartPaddle and Aquanex measurements. However, agreement between systems does not make their reliability or measurement-error estimates interchangeable, particularly across different devices and experimental protocols. The present study was designed as a single-session comparison of two swimming paces rather than as a reliability study. Repeating the maximal and submaximal conditions within the same session could have introduced residual fatigue and order effects and would primarily have assessed within-session repeatability. Therefore, a dedicated assessment involving the same PoolShark free-swimming protocol on separate testing occasions would be required to calculate a reliability-based Minimal Detectable Change.

The conversion of MDD and MID from newtons to percentages was performed only for scale harmonization with BAI-1. Because BAI-1 is expressed relative to the sum of right- and left-hand force, MDD and MID were divided by the mean bilateral hand force sum for the corresponding pace and multiplied by 100. This allowed direct descriptive comparison between metrics. In the present data, MDD corresponded to 4.48% at the 50 m pace and 4.83% at the 200 m pace, while the exploratory MID estimate corresponded to 2.68% and 2.89%, respectively. These values were lower than commonly used arbitrary asymmetry thresholds, suggesting that potentially relevant or statistically detectable differences may occur below traditional cut-offs. However, because these values are sample-specific, they should not be generalized as universal thresholds for swimmers.

From an interpretative perspective, these findings illustrate that the interpretation obtained from a single percentage-based index may differ when the same observation is contextualized using sample-derived reference values. This aligns with broader asymmetry literature showing that asymmetry-performance relationships are not always consistent and may depend on the task, population, and physical quality assessed [21,29,50]. However, the MDD and exploratory MID values obtained here should be treated only as descriptive, sample-specific reference points and not as criteria for technical evaluation, monitoring decisions, or intervention. This multi-metric interpretation may help contextualize individual asymmetry profiles without relying solely on a single percentage-based index.

Nevertheless, several limitations should be acknowledged. The sample size was small, although this is common in experimental studies involving competitive swimmers of national or international level. The marked sex imbalance, together with the small sample size, also precluded meaningful age- or sex-specific interpretation of the findings. Accordingly, the findings should be interpreted as specific to this cohort and should not be generalized to male and female swimmers separately. The within-subject design reduced inter-individual variability for the main comparisons, but statistical power remained limited, particularly for detecting interaction effects and individual asymmetry patterns. Given this limitation, the non-significant pace × side interaction and BAI-1 comparisons cannot distinguish confidently between the absence of an effect and insufficient statistical power and should not be interpreted as evidence that pace-related differences are absent. Although the fixed trial order may have influenced pace-specific force or asymmetry values through order effects, it does not directly compromise the primary methodological comparison of BAI-1, MDD, and MID, because all three approaches were applied to the same observations within each pace. Nevertheless, future studies primarily designed to compare asymmetry between swimming intensities should consider counterbalanced or repeated-session protocols. The study included only two front crawl intensities, and the findings may not generalize to other strokes, distances, breathing strategies, or competitive contexts. Swimmers were instructed to hold their breath during the analyzed segment to minimize disruptions in stroke coordination and signal interpretation, but this does not fully reproduce natural breathing patterns during competition. Because breathing can alter upper-limb kinematics and coordination [42], the non-breathing condition may have reduced breathing-related variability but may also have modified the asymmetry that would occur during natural swimming. Future studies should compare inter-limb hand force asymmetry under breathing and non-breathing conditions and across different breathing strategies. Moreover, hand force was assessed using wearable sensors and a proprietary processing algorithm. Although similar wearable systems have been used to monitor in-water force variables and examine their relationship with swimming velocity [43], the measured variable in the present study should be interpreted as hand force rather than total propulsive force. Finally, the MDD and MID values were derived from the present sample and should be interpreted as exploratory reference values rather than validated thresholds.

Future research should examine whether asymmetry magnitude and direction remain stable across longer swimming distances, larger numbers of stroke cycles, and repeated testing sessions. This would allow researchers to determine whether the descriptive tendency for greater asymmetry during sprint pace becomes more evident under extended or more fatiguing conditions. Future studies should also include reliability assessment to compute MDC and distinguish measurement error from true individual change. In addition, continuous analyses of the force–time curve may provide more detailed information about when asymmetries emerge during the arm-pull, rather than relying only on discrete average values. Longitudinal and intervention studies are needed to determine whether reducing inter-limb asymmetry improves swimming performance or whether certain asymmetry profiles reflect functional and stable individual adaptations. Until such evidence is available, asymmetry should be interpreted as an individualized monitoring variable rather than as an automatic indicator of technical deficiency.

5. Conclusions

This study showed that the interpretation of inter-limb hand force asymmetry in competitive swimmers depends strongly on how asymmetry is quantified and contextualized. Mean hand force was higher during the 50 m pace than during the 200 m pace, confirming that swimmers increased force output with swimming intensity. However, asymmetry magnitude did not differ significantly between paces, and the data did not support a clear reduction in asymmetry at higher speeds. Instead, the descriptive pattern suggested that differences between the right and left hands may become more apparent during sprint swimming in some swimmers. By comparing BAI-1 with MDD and an exploratory MID estimate, the present study showed that interpretations of individual asymmetry vary depending on whether asymmetry is treated as a percentage magnitude, a statistically detectable difference, or a potentially relevant practical difference. These findings support the potential value of a multi-metric and individualized approach to asymmetry monitoring in swimming, while emphasizing that the proposed interpretative framework remains exploratory and sample-specific. Future research with larger samples, repeated testing, and longer continuous recordings is needed to determine whether greater inter-limb symmetry is consistently related to performance or whether inter-limb asymmetry reflects individual technical strategies.