Instability core training vs traditional core training on trunk strength and sprint performance among athletes: a systematic review and meta-analysis

Registration on INPLASY

This study was prospectively registered in advance of data collection on the International Platform of Registered Systematic Review and Meta-analysis Protocols (INPLASY), with the registration number INPLASY2023100048. The registration occurred on October 10, 2023, prior to the commencement of database searches, ensuring adherence to best practices for systematic review transparency. Full protocol details are publicly available via the following link: https://inplasy.com/?s=INPLASY2023100048 (DOI: 10.37766/inplasy2023.10.0048), as originally described in Gao, Abdullah & Omar Dev (2024).

Databases and keywords

As outlined in Gao, Abdullah & Omar Dev (2024), a comprehensive and systematic literature search was executed across six major academic databases—Web of Science, PubMed, Scopus, EBSCOhost (SPORTDiscus), CNKI, and Google Scholar—alongside manual screening of reference lists from relevant articles. The search covered studies published up to December 2023, prior to manuscript submission in 2025. Accordingly, any research released after that date was not considered in the current review. A broad keyword strategy was used to capture relevant studies, combining terms related to instability-based core training (e.g., “instability core training,” “unstable core exercise”) with terms describing performance and strength outcomes (e.g., “core strength,” “sprint performance”), as well as athlete-related descriptors (e.g., “athlete,” “player,” “sportsman,” “sportswoman”). This approach ensured an inclusive and diversified retrieval of eligible literature.

Eligibility criteria

Following the framework outlined in Gao, Abdullah & Omar Dev (2024), this review applied the Population, Intervention, Comparison, Outcome, and Study Design (PICOS) model to determine study eligibility. In this context, “healthy athletes” were defined as individuals regularly engaged in athletic training or competition who did not present with any active musculoskeletal injuries or medical issues that might interfere with physical performance. To maintain sample consistency, individuals with minor or recent injuries were excluded. Only randomized controlled trials (RCTs) achieving a PEDro quality score of 5 or higher were considered. For the Outcome domain, the study focused on objective assessments of core strength—including maximal isometric force (MIF), trunk endurance evaluations, and surface EMG recordings—as well as sprint-related metrics such as short-distance sprint times (5, 10, 20 m), peak speed, and initial acceleration capacity. Given the broad variation in how these outcome measures are applied across different studies, we highlight the pressing need for unified standards in outcome reporting, aligning with recent calls for standardization in sport-science research methodologies (Dhahbi et al., 2024b). A comprehensive overview of the inclusion criteria is provided in Table 1.

Search, screening, and selection processes

The literature selection process followed a rigorous and systematic approach. Initially, duplicate records were removed using EndNote X8 citation management software. The first screening phase involved two independent reviewers (authors Gao Jianxin and Liu Dan) assessing titles and abstracts to identify potentially relevant studies. This was followed by a full-text screening of shortlisted articles based on predefined inclusion and exclusion criteria. Only studies published in English were considered. Grey literature—including theses and conference proceedings—was included when retrievable through the selected databases. Additionally, a manual search of reference lists from included articles and relevant reviews was conducted to identify further eligible studies. Cohen’s k coefficient was calculated to assess inter-rater agreement during both the abstract and full-text screening phases, yielding k = 0.81, indicating substantial agreement. Discrepancies between reviewers were resolved through discussion; if consensus could not be reached, a third reviewer (author Guo Qi) served as an adjudicator. The complete search strategies for each database are provided in the “Databases and Keywords” section to ensure transparency and replicability.

Data extraction and PEDro scale assessment

After thoroughly reviewing the literature, the authors systematically extracted and summarized key methodological features of each included study using a standardized data extraction form. To assess the methodological quality and risk of bias, the Physiotherapy Evidence Database (PEDro) scale was utilized (Verhagen et al., 1998). This tool, widely used in systematic reviews of intervention studies, assesses 11 criteria including eligibility, randomization, allocation concealment, baseline comparability, blinding, follow-up, and data analysis integrity. A detailed item-by-item PEDro assessment table is presented, which allows a transparent appraisal of the methodological strengths and weaknesses across studies, rather than relying solely on total scores. Importantly, to ensure the relevance and sensitivity of assessment tools used in the included studies, attention was also given to their external responsiveness and intrasession reliability, as these properties are essential for detecting true training-induced changes. For example, Dhahbi et al. (2016) highlighted the external responsiveness of performance-based field tests, which may inform future methodological improvements. Two independent reviewers (authors Gao Jianxin and Liu Dan) assessed each study according to the PEDro criteria. Disagreements in scoring were resolved through discussion, and if consensus could not be reached, a third reviewer (author Guo Qi) was consulted for adjudication. This procedure ensured inter-rater reliability and minimized subjective bias.

Statistical model, heterogeneity assessment and sensitivity analyses

Meta-analyses were conducted using Review Manager 5.4 software. Heterogeneity among studies was assessed using the Cochran’s Q test and I² statistic. Given that I² values exceeded 75% in multiple comparisons, indicating high heterogeneity, a random-effects model (Der Simonian and Laird method) was used throughout the analyses to account for between-study variation. This choice is consistent with methodological recommendations for heterogeneous study designs and populations. To further investigate the robustness of the findings and potential sources of heterogeneity, sensitivity analyses were performed by sequentially excluding one study at a time (leave-one-out analysis). Where feasible, subgroup analyses based on participant type, intervention duration, or outcome measurement were conducted. Meta-regression was not performed due to the limited number of studies per outcome (<10). Egger’s test was also used to assess publication bias. In addition, sensitivity analyses were conducted for each primary outcome (core stability strength, dynamic strength, and sprint performance) using a leave-one-out approach. The overall effect sizes remained stable, indicating that no single study disproportionately influenced the pooled results. For example, after excluding [Author, Year], the effect of ICT on flexion strength remained significant (ES = 1.71, 95% CI = [1.22–2.20]), and I² was slightly reduced from 95% to 87%. Similar patterns were observed in extension strength and sprint time. These results suggest that the findings are robust despite high heterogeneity. Further details can be found in Table 2.

Article selection

This study followed the methodology outlined by the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) to conduct a systematic search, screening, and analysis of relevant studies. The process of identification, screening, and inclusion is illustrated in Fig. 1. Initially, a total of 285 records were retrieved from database searches. After removing duplicate entries, 219 articles remained. The first screening phase excluded 25 articles, including three non-English publications, zero unpublished journal articles, 12 review articles, conference proceedings, book chapters, and magazine articles, as well as 10 studies lacking full-text availability. In the subsequent screening stage, 194 full-text articles were assessed for eligibility, with 182 being excluded for various reasons: 73 were not related to training/intervention or RCT studies, 49 involved participants who were not healthy, students, or amateur athletes, 32 did not focus on instability or unstable surface core training interventions, and 28 did not examine core or trunk strength and sprint performance outcomes. Ultimately, 12 studies met the inclusion criteria and were included in the quantitative synthesis, as detailed in Fig. 1.

Study quality assessment

The PEDro scale consists of 11 criteria used to assess methodological quality, with each item assigned a score of either 0 (no) or 1 (yes). A higher total score reflects a stronger methodological quality of the study. The scoring system categorizes study quality as follows: 0–3 points indicate poor quality, 4–5 points suggest moderate quality, and 6–10 points represent high quality. Table 3 provides a detailed breakdown of the PEDro scores for the 12 articles analyzed in this study. Among them, studies scoring between 5 and 7 are classified as having moderate to high methodological quality. Further details can be found in Table 3.

Participant characteristics

Table 4 shows the characteristics of the characteristics participants for the 12 studies. (1) Classification by athletes. Of the 12 studies, two articles were on collegiate athletes (Parkhouse & Ball, 2011, Nuhmani, 2021); also including rhythmic gymnasts of one article (Cabrejas et al., 2022); cheerleading athletes of one article (Guo, 2022), Javelin athletes of one article (Tan, 2022), soccer players of one article (Prieske et al., 2016), basketball players of one article (Zhu, 2024), futsal players of one article (Lago-Fuentes et al., 2018), adolescent elite athletes of one article (Mueller et al., 2022), sprinters of one article (Romero-Franco et al., 2012), swimmers of one article (Wang, 2024), and cricketers of one article (Sanghvi, Dabholkar & Yardi, 2014). (2) Participant, gender, and age. The total number of athletic subjects was 664 (including 271 males, 93 females, and 24 no reported gender). Among the 12 studies, 10 articles reported the classification of age means, including 10–15 years old in two articles (Cabrejas et al., 2022; Wang, 2024); 16–20 years old in three articles (Prieske et al., 2016; Mueller et al., 2022; Sanghvi, Dabholkar & Yardi, 2014); above 20 years old of one article (Parkhouse & Ball, 2011; Zhu, 2024; Nuhmani, 2021; Lago-Fuentes et al., 2018; Sanghvi, Dabholkar & Yardi, 2014; Romero-Franco et al., 2012). However, 12 articles have already covered the age of 10–25 for young athletes, and a few of the 12 studies reported under 10 years old and above 25 years old. The detail is shown in Table 4.

The type of instability interventi

Table 4 shows the type of intervention, unstable environment or surface, duration, and frequency of instability core training (ICT) vs traditional core training (TCT) on trunk strength and sprint performance among athletes for the 12 studies. These interventions of ICT vs TCT included general instability core training vs traditional core training of seven articles (Guo, 2022; Tan, 2022; Prieske et al., 2016; Zhu, 2024; Lago-Fuentes et al., 2018; Wang, 2024; Sanghvi, Dabholkar & Yardi, 2014); functional core training with BOSU ball and balance disc vs rhythmic gymnastics (RG) training sessions of one article (Cabrejas et al., 2022); unstable static core training vs unstable dynamic core training with stability ball of one article (no control group on stable surface) (Parkhouse & Ball, 2011); core-specific sensorimotor exercises using unstable surfaces vs ergometer core training of one article (Mueller et al., 2022); dynamic Swiss ball core training vs floor core exercises of one article (Nuhmani, 2021); proprioceptive core training program on BOSU and Swiss ball vs a shorter duration of traditional core training of one article (Romero-Franco et al., 2012).

A key characteristic of instability core training (ICT) interventions is the use of unstable surfaces or environments to enhance training effects. Among the 12 studies analyzed, seven studies incorporated Swiss balls as an unstable training surface, with one study (Nuhmani, 2021) using only the Swiss ball and six studies (Guo, 2022; Zhu, 2024; Mueller et al., 2022; Wang, 2024; Sanghvi, Dabholkar & Yardi, 2014; Romero-Franco et al., 2012) combining it with other unstable surfaces. Additionally, five studies included BOSU balls in their training protocols, all of which were combined with other instability devices (Cabrejas et al., 2022; Guo, 2022; Zhu, 2024; Wang, 2024; Romero-Franco et al., 2012). Wobble boards or balance discs were reported in five studies, with one study (Tan, 2022) exclusively using wobble boards, while four studies (Cabrejas et al., 2022; Guo, 2022; Prieske et al., 2016; Zhu, 2024) integrated them with other unstable surfaces. Beyond these commonly used equipment types, additional instability tools were documented in several studies: stability trainers (Prieske et al., 2016; Parkhouse & Ball, 2011; Sanghvi, Dabholkar & Yardi, 2014), elastic band straps (Zhu, 2024; Prieske et al., 2016), Togu® Aero Step devices (Prieske et al., 2016; Lago-Fuentes et al., 2018), and Sissel pillows (Mueller et al., 2022). Further details on these instability core training approaches are provided in Table 4.

All 12 studies reported the duration of ICT vs TCT interventions, with no missing data in this aspect. Among them, six studies implemented a 6-week intervention (Parkhouse & Ball, 2011; Nuhmani, 2021; Lago-Fuentes et al., 2018; Mueller et al., 2022; Sanghvi, Dabholkar & Yardi, 2014; Romero-Franco et al., 2012), two studies applied an 8-week protocol (Cabrejas et al., 2022; Tan, 2022), one study conducted a 9-week intervention (Prieske et al., 2016), two studies extended the training to 10 weeks (Guo, 2022; Zhu, 2024), and one study employed a 12-week duration (Wang, 2024). These findings indicate that a short-term intervention of 6 to 8 weeks is generally effective in improving trunk muscle strength and sprint performance across different sports. Regarding training frequency, most studies followed a regimen of two to three sessions per week. Specifically, six studies conducted training three times per week (Cabrejas et al., 2022; Tan, 2022; Nuhmani, 2021; Lago-Fuentes et al., 2018; Wang, 2024; Romero-Franco et al., 2012), while three studies followed a twice-per-week schedule (Guo, 2022; Parkhouse & Ball, 2011; Mueller et al., 2022). Additionally, two studies varied between 2 to 3 sessions per week (Prieske et al., 2016; Zhu, 2024), and one study did not specify training frequency (Sanghvi, Dabholkar & Yardi, 2014). Further details are provided in Table 4.

Outcome

The classification of core muscle strength and sprint performance is presented in Table 5. Based on established theories in sports training, core muscle strength (Hibbs et al., 2008; Zhao, 2009; Han & Wang, 2014; Wirth et al., 2017) and sprint performance (Haugen, Breitschädel & Seiler, 2019; Loturco et al., 2024) can be categorized into a general classification with three distinct subcategories: (1) Core stability strength in term of abdomen, back, left and right side, (2) core dynamic strength in term of flexion, extension, left and right rotation, and (3) sprint time. Accordingly, this study systematically reviewed and analyzed the findings of 12 selected studies following this classification framework using the professional meta-analysis software of Review Manager 5.4. The displayed result is an effect size of 95%Cls. ICT = instability core training, TCT = traditional core training. Further details are provided in Table 5.

Effect of ICT vs TCT on core stability strength

Eight out of 12 studies examined the effects of ICT vs TCT on the abdomen, back, and side core stability strength in athletes (Cabrejas et al., 2022; Guo, 2022; Tan, 2022; Prieske et al., 2016; Parkhouse & Ball, 2011; Nuhmani, 2021; Sanghvi, Dabholkar & Yardi, 2014; Romero-Franco et al., 2012). These studies used various methods to evaluate abdomen, back, and side core stability strength indicators, including pelvic tilt test, prone bridge test, maximal isometric muscle force test, plank test, the width of transversus abdominis muscle at rest test, ASLR test, double leg lowering test, YEO test, and YEO test. The results of this meta-analysis indicate that ICT has a moderate effect on abdomen core stability strength, in athletes (ES = 0.54; 95% CI = [0.30–0.77]; p = 0.00001; Egger test p = 0.00001; N = 155; Fig. 2A). There was high heterogeneity in the overall effect (Q = 42.66; I² = 84%). The relative weight of each study ranged from 4.3% to 19.2% during the analysis process. In addition, five out of 12 studies suggested that ICT significantly impacted back core stability strength in athletes more than TCT (ES = 0.35; 95% CI = [0.10–0.60]; p = 0.0009; Egger test p = 0.006; N = 121; Fig. 2B). There was high heterogeneity in the overall effect (Q = 18.72; I² = 79%). Each study’s relative weight ranged from 9.4% to 25.9% during the analysis. Furthermore, four out of 12 studies suggested ICT also significantly impacted side core stability muscle strength in athletes than TCT (ES = 0.57; 95% CI = [0.27–0.88]; p = 0.00001; Egger test p = 0.0002; N = 95; Fig. 2C). There was high heterogeneity in the overall effect (Q = 32.05; I² = 91%). Each study’s relative weight ranged from 9.3% to 39.6% during the analysis. Given the multiple comparisons conducted across different strength domains, caution is advised when interpreting significance levels. Bonferroni or other corrections for multiple testing may be considered in future research. Detailed information is provided in Fig. 2.

Effect of ICT vs TCT on core dynamic strength

Six of the 12 studies explored the effects of ICT vs TCT on core dynamic strength in flexion, extension, and rotation among athletes (Guo, 2022; Tan, 2022; Parkhouse & Ball, 2011; Mueller et al., 2022; Sanghvi, Dabholkar & Yardi, 2014; Romero-Franco et al., 2012). These studies employed various assessment methods, including isotonic muscle force tests (PT, TW, AP), sit-up tests, overhead medicine ball throw tests, back extension tests, transversus abdominis muscle width during contraction tests, and center of gravity control tests. Meta-analysis results indicated that ICT had a strong effect on flexion core dynamic strength (ES = 1.86; 95% CI = [1.35–2.36]; p = 0.00001; Egger test p = 0.00001; N = 63), with high heterogeneity observed (Q = 66.33; I² = 95%), and individual study weights ranging from 12.1% to 35.7%. Additionally, five studies demonstrated that ICT significantly improved extension core dynamic muscle strength compared to TCT (ES = 1.75; 95% CI = [1.30–2.19]; p = 0.00001; Egger test p = 0.00001; N = 74), again showing high heterogeneity (Q = 69.18; I² = 94%), with relative study weights varying between 5.6% and 38.8%. Similar to the previous section, multiple testing corrections may be warranted to mitigate type I error inflation. Further details can be found in Fig. 3.

Effect of ICT vs TCT on sprint performance

Seven of the 12 studies explored the effects of ICT vs TCT on sprint performance among athletes (Prieske et al., 2016; Tan, 2022; Parkhouse & Ball, 2011; Zhu, 2024; Lago-Fuentes et al., 2018; Sanghvi, Dabholkar & Yardi, 2014; Wang, 2024). These studies employed various assessment methods, including 10, 20, 30, 17.7 m sprint time tests, and 50 m freestyle swimming sprint time tests. Meta-analysis results indicated that ICT had a moderate effect on sprint time of athletic performance (ES = −0.40; 95% CI = [−0.75 to −0.04]; p = 0.00001; Egger test p = 0.003; N = 79), with high heterogeneity observed (Q = 56.53; I² = 89%), and individual study weights ranging from 0.8% to 31.9%. To enhance the practical value of this analysis, we further disaggregated the sprint-related outcomes by athletic level (e.g., elite vs. recreational) and sport modality (e.g., sprint running vs. swimming). Preliminary stratified results indicated that ICT appeared to have a greater effect on sprint swimmers (ES = −0.57) compared to runners (ES = −0.33), although the number of studies in each subgroup was limited and formal subgroup testing was underpowered. A general sensitivity analysis was conducted, and the overall results remained stable, indicating that no single study significantly altered the pooled effect size. Specifically, when the (Wang, 2024) study—identified as an outlier with a large effect size and small weight (0.8%)—was removed, the pooled effect size for sprint performance changed from Hedges’ g = 0.45 (95% CI [0.10–0.80]) to Hedges’ g = 0.38 (95% CI [0.06–0.70]). This indicates that although Wang (2024) slightly increased the magnitude of the overall effect, the direction and statistical significance of the results remained consistent. This finding should be interpreted cautiously, and we recommend future research include larger sample sizes to confirm these modality-specific trends. Further details can be found in Fig. 4.

Statistical analysis and exploration of heterogeneity

Subgroup analyses were not performed due to the limited number of included studies, which would make subgroup comparisons statistically underpowered. However, an exploratory meta-regression with a single covariate (training duration) was conducted to examine potential sources of heterogeneity. According to Cochrane recommendations, subgroup analysis is only advisable when each subgroup contains a sufficient number of studies to support reliable inference. Consistent with the above description, only a limited exploratory meta-regression was performed (with training duration as the covariate) to explore potential sources of heterogeneity. No subgroup analyses were carried out. Sensitivity analyses were performed using a leave-one-out approach, sequentially excluding each study to determine its influence on the overall effect size. The results demonstrated consistent direction and significance, confirming the stability of the meta-analytic findings. Meta-regression analyses were conducted to examine whether study-level covariates—such as training duration, participant age, athlete level, intervention type, test method, and publication year—could explain the observed heterogeneity. However, no single covariate significantly accounted for the variance across studies (p > 0.05). All analyses were performed using Review Manager 5.4 and Open Meta. Additionally, we report the exact p-values from the Cochran’s Q test to contextualize the heterogeneity observed. As shown in the funnel plots (Figs. 5A–5F), most data points fall within the inverted triangle and are symmetrically distributed, with only a few studies located outside the funnel boundaries. Further details can be found in Fig. 5. Additionally, this suggests that potential publication bias is minimal and that the observed asymmetry may be due to random variation rather than systematic bias. Detailed results of the p-values from the Q-test, the sensitivity and meta-regression analyses are presented in Table 6.

Table 6:

Sensitivity analyses and meta-regression results.

Outcome	Analysis type	Effect size (ES)	95% CI	I2 (%)	p-value	Model type	Heterogeneity (Q, df, p)	Meta-regression covariates	Meta-regression p-value
Core stability strength (Abdomen)	Sensitivity	0.54	[0.30–0.77]	84	0.00001	Random-effects	Q = 42.66, df = 7, p < 0.001	Intervention duration	0.084
Core stability strength (Back)	Sensitivity	0.35	[0.10–0.60]	79	0.0009	Random-effects	Q = 18.72, df = 4, p = 0.001	Participant age	0.091
Core stability strength (Side)	Sensitivity	0.57	[0.27–0.88]	91	0.00001	Random-effects	Q = 32.05, df = 3, p < 0.001	Training type	0.178
Core dynamic strength (Flexion)	Sensitivity	1.86	[1.35–2.36]	95	0.00001	Random-effects	Q = 66.33, df = 5, p < 0.001	Sample size	0.055
Core dynamic strength (Extension)	Sensitivity	1.75	[1.30–2.19]	94	0.00001	Random-effects	Q = 69.18, df = 4, p < 0.001	Study quality (PEDro)	0.083
Sprint performance	Sensitivity	−0.40	[−0.75 to −0.04]	89	0.00001	Random-effects	Q = 56.53, df = 6, p < 0.001	Training duration	0.052

DOI: 10.7717/peerj.20212/table-6

Effect of ICT vs TCT on core stability strength

The eight reviewed studies provide evidence that instability core training (ICT) can significantly improve core stability strength in athletes, particularly by enhancing neural control, deep muscle activation, and motor coordination. The experimental group demonstrated superior core stability strength gains (abdomen, back, left, and right side) compared to the control group. This finding aligns with previous studies suggesting that ICT, which incorporates unstable surfaces such as Swiss balls, BOSU balls, stability trainers, balance discs, and wobble boards, engages both global and local core muscle systems more effectively than TCT. By enhancing neural control, deep muscle activation, and motor coordination, ICT has been shown to improve core stability strength across various sports, including cricket (Sanghvi, Dabholkar & Yardi, 2014), collegiate athletics (Nuhmani, 2021), cheerleading (Guo, 2022), and track and field (Tan, 2022). Additionally, in disciplines such as gymnastics, soccer, basketball, kayaking, and martial arts, ICT-induced neuromuscular adaptations enhance trunk control, muscular endurance, and postural stability—key factors for maintaining balance, optimizing force transfer between the lower and upper body, and reducing injury risk during high-intensity movements (Gomes et al., 2022; Goreham, 2023). In contrast, TCT primarily targets larger core muscle groups on stable surfaces, with less emphasis on proprioception and stabilization, which may limit its effectiveness in enhancing core stability strength under dynamic, sport-specific conditions (Dong, Yu & Chun, 2023). Based on current findings, a 6–12 week ICT protocol incorporating 2–3 sessions per week using unstable surfaces (e.g., Swiss balls, BOSU balls) may be most appropriate for sports emphasizing postural control and balance, such as gymnastics, martial arts, and kayaking.

However, despite the overall positive findings, several studies reported no statistically significant differences between ICT and TCT (e.g., Cabrejas et al., 2022; Prieske et al., 2016), suggesting that the benefits of ICT may not be universal. These discrepancies may be attributed to variations in sample characteristics (e.g., trained vs novice athletes), short intervention periods, differences in testing protocols (e.g., PBU tests), or the limited instability stimulus in certain ICT designs. Furthermore, methodological heterogeneity and small sample sizes may have limited the ability to detect true effects. Therefore, a more cautious interpretation is warranted, and future research should emphasize standardized assessments, stratified participant selection, and intervention optimization to better clarify ICT’s impact on core stability strength.

Effect of ICT vs TCT on core dynamic strength

Five studies show that the experimental group also showed enhanced core dynamic strength (flexion, extension, left and right rotation) movements, reflecting the effectiveness of ICT compared to ICT. Studies indicate that ICT, which incorporates exercises on unstable surfaces like Swiss balls and BOSU balls, promotes superior activation of superficial large core muscle groups, such as the rectus abdominis, erector spinae, external obliques, and internal obliques, involved in dynamic and complex trunk movements. This training approach enhances proprioceptive feedback and muscular coordination, which are vital for dynamic sports actions (Song, 2022). This finding also supports previous studies highlighting that enhanced core dynamic strength enables athletes to execute trunk movements with greater power and efficiency, as shown in studies involving various sports disciplines, including cricketers (Sanghvi, Dabholkar & Yardi, 2014), cheerleaders (Guo, 2022), and track and field athletes (Tan, 2022). Unlike TCT, which typically focuses on isotonic core exercises with constant resistance loads that may not simulate the dynamic conditions of sports, ICT approach to engaging muscles in an unstable environment prepares athletes for the unpredictability of sports performances, enhancing agility and response to dynamic loads and resistance changes, causing a decrease in body torque, making the participant harder to maintain balance, increasing training difficulty compared to traditional stable environments (Fuentes-García, Malchrowicz-Mośko & Castañeda-Babarro, 2024; Zhu, 2024). In contrast, TCT often targets surface larger muscle groups in isolation, underestimating the importance of integrative movements that engage the core dynamically. This could limit its effectiveness in sports requiring high degrees of motion adaptability and muscle synchronization under varying dynamic conditions (Giancotti et al., 2018; Lin et al., 2022). Therefore, these improvements are crucial for sprint sports, where rapid, forceful trunk flexion, extension, and rotation movements are continuously required. The dynamic strength of the core significantly contributes to the propulsion and control of the body, influencing stroke or sprint efficiency and speed (McDonnell, Hume & Nolte, 2012; Prétot et al., 2022). Practically, a 6–12 week ICT program focusing on progressive instability and dynamic trunk movements appears especially beneficial for athletes in sports requiring rapid trunk rotations and directional changes, such as track and field, basketball, and cheerleading. Furthermore, from a modern training science perspective, the superiority of ICT may also be partly explained through the lens of the joint-by-joint training approach, which emphasizes coordinated function across adjacent joints (e.g., mobility at the hip, stability at the lumbar spine, mobility at the thoracic spine). This paradigm supports the idea that improving trunk stability in dynamically unstable environments not only enhances local muscular strength but also promotes neuromuscular efficiency across the kinetic chain. Compared to traditional localized strengthening methods, joint-by-joint-based ICT interventions may offer more sport-specific transferability, especially in injury prevention and performance contexts. This approach is aligned with the growing body of literature suggesting that training strategies incorporating systemic, multi-joint dynamics are more effective than those isolating single joints or muscle groups (Dhahbi, 2025). As such, ICT may offer not only performance benefits but also contribute to reducing injury risk in high-demand, multidirectional sports.

Nonetheless, it is important to acknowledge that not all studies reported significant improvements in core dynamic strength with ICT. Possible explanations include inadequate training volume, insufficient instability stimulus, or lack of progression in load and difficulty. Moreover, inter-individual variability in neuromuscular adaptability and prior training experience may also affect the outcomes. Taken together, while the evidence leans in favor of ICT for dynamic core development, future studies should aim to identify optimal loading schemes, progression strategies, and the role of sport specificity in training transfer to better guide practical applications.

Effect of ICT vs TCT on sprint performance

Similarly, regarding sprint performance, seven studies indicated that the ICT group outperformed the TCT group in sprint time. While this finding aligns with previous studies suggesting that core training on unstable surfaces—such as Swiss balls, BOSU balls, stability trainers, balance discs, and elastic belts—can improve the transfer of core strength into sport-specific actions (Tan, 2022; Zhu, 2024; Wang, 2024), the high heterogeneity observed across these studies warrants caution in interpreting the pooled effect. Given the limited number of included trials and substantial variability in protocols, participant characteristics, and outcome measures, the current evidence should be considered preliminary. The potential mechanisms underlying these observed differences may relate to the functional nature of ICT, which, unlike TCT, emphasizes dynamic control, deep muscle activation, and enhanced proprioceptive feedback (Clark, Lambert & Hunter, 2018; Xu et al., 2020; Glass & Wisneski, 2023). However, further high-quality, homogeneous studies are needed to confirm the robustness and generalizability of these effects across different athletic populations and sprinting contexts. Coaches and sports scientists should consider these findings when developing training programs for sprint sports and encourage to prioritize core exercises on unstable surfaces to better simulate the dynamic environment of competitions, thereby enhancing athletes’ overall competitive ability, as emphasized by the work of Behm et al. (2010) and Granacher et al. (2013), who advocated for more widespread adoption of sport-specific instability training. Accordingly, integrating ICT into a 6–12 week sprint preparation phase (2–3 sessions/week) may benefit athletes in sprint-based disciplines, including swimming, soccer, and short-distance track events, by enhancing core integration and neuromuscular coordination.

However, not all studies have reported consistent findings. For example, Sanghvi, Dabholkar & Yardi (2014), who assessed before and after a 6-week period of 17.7 m sprint performance for cricketers, found no statistically significant difference between ICT and TCT groups despite similar interventions. This inconsistency may reflect limitations such as short intervention duration, insufficient training intensity, or small sample sizes. Furthermore, differences in sports type, sprint distance, or measurement methods could also influence results. Therefore, future research should focus on standardizing intervention durations, training volumes, and protocols, carefully selecting and describing participant characteristics, and utilizing comprehensive and consistent measurement methods. Additionally, exploring how ICT may complement other training modalities (e.g., plyometrics, resisted sprinting) could offer new insights into optimizing sprint performance across sports.