Effects of strength training on repeated sprint ability in team sports players: a systematic review

Search strategy

This study is a systematic literature review, following the Preferred Reported Items for Systematic Reviews and Meta-Analysis (PRISMA) statements (Liberati et al., 2009), in which original, peer-reviewed articles published in English without date restrictions were considered. The PubMed, Web of Science (Core Collection), and Scopus databases were considered until May 26, 2024.

Eligibility criteria

Inclusion and exclusion criteria were structured through a PICOS approach (participants, intervention, comparators, outcomes, and study design) and are presented in detail in Table 1.

Selection of articles

Search criteria were performed using the following descriptors: “ballistic training”, “resistance training” [Mesh], “strength training” [Mesh], “complex training”, “contrast training”, “weight training”, “repeated change of direction”, “repeated sprint ability”, “repeated sprint exercise”. The search equation in PubMed, Web of Science (Core Collection), and Scopus databases originated with the Boolean operators “AND” and “OR”. Table 2 shows in detail the search strategies used in the three databases.

In addition, manual searches were performed by consulting grey literature, and reference lists were analyzed from selected articles in search of possible studies.

Extraction and evaluation process

Articles were selected by title and abstract; then, the full text was reviewed to finally apply the inclusion criteria described in Table 1. Duplicate articles were removed manually using Zotero (version 6.0.37) reference management software, and the full-text articles excluded were recorded in a Microsoft Excel table with the causes for their exclusion. Two authors conducted the search independently (AO-RY); if necessary, a third reviewer acted as judge (MMA) to resolve potential discrepancies between the two authors.

Data extraction

Data extraction was performed by the lead author using a standardized form created in Microsoft Excel. The repeated sprint ability was considered the primary variable, measured with different RSA tests recorded in execution time in seconds as mean ± standard deviation (SD). From the extracted data, means, SD, first author’s last name, year of publication of the studies, sport, competition season, training intervention characteristics (frequency (days/week), duration (weeks), total sessions (frequency/duration), general characteristics of training (protocol), pause between sets and exercises (seconds), rest between sessions (hours), intervention time (minutes)) and descriptive characteristics of team sports players (number of subjects per group (size), chronological age (years), height (cm), body weight (kg), sex (M/F), performance level and training experience (years of training)) were recorded in a results table.

Study and assessment of the risk of bias

The Physiotherapy Evidence Database (PEDro) scale was used to assess the risk of bias and methodological quality of the included studies (Moseley et al., 2002). This scale was developed to quickly identify trials that tended to be internally valid and to have enough statistical information to guide decision-making. The PEDro scale is designed with eleven items, all but the first (external validity) awarding one point if present, so the final score should be between 0 and 10 points. According to the predetermined cut-off points, studies are classified as high (eight to ten), medium (four to seven), or low quality (less than four). (Table 3). The lead author assessed the risk of bias for all studies. Also, p and ES values were collected (if available) for each selected article. The percent of change (Δ%) was calculated (if not available) of each article to evaluate the magnitude of the effects using the following equation: Δ% = (Mpost –Mpre/Mpre) ×100, where Mpost represents the mean value after intervention and Mpre the baseline mean value.

Selection of the studies

Through the realization of the literature search in the selected databases, a total of 723 studies were identified in PubMed, 893 in Web of Science, and 1,760 from Scopus. After removing duplicates, meta-analyses, systematic reviews, and abstracts, a total of 2,807 results of full texts were available. After reading the titles and abstracts, 39 relevant articles remained, and 2,768 were excluded for not presenting the eligibility criteria. The full texts of the 39 articles were read using the inclusion and exclusion criteria specified for the eligibility of the studies. Finally, ten articles were selected (Fig. 1).

The methodological quality of the 10 studies included in this systematic review was quantified through the PEDro scale, yielding moderate quality in the 10 studies (five to six points). None of the studies reported whether the allocation was concealed (criterion 3) (Table 3).

Study characteristics

Among the main characteristics of ten studies (Chatzinikolaou et al., 2018; Durán-Custodio et al., 2023; Gonzalo-Skok et al., 2016; Hammami et al., 2019; Hammami et al., 2018; Hammami et al., 2017b; Hammami et al., 2017a; Hermassi et al., 2017; Torres-Torrelo, Rodríguez-Rosell & González-Badillo, 2017; Torres-Torrelo et al., 2018), these are composed of 10 active control groups (continued with their regular training in their respective sport) and 11 experimental groups (replaced a part of the training of their respective sport by ST or CCT), of nine studies, included only one experimental group. Only one study experimented with more than one group (Hammami et al., 2017a). All studies were randomized controlled trials except for one study (Hammami et al., 2018), which was not randomized. A total of nine studies included only men, and one included only women (Hammami et al., 2019). The age ranged from 14.1 ± 0.6 (Chatzinikolaou et al., 2018) to 25.8 ± 2.0 years (Durán-Custodio et al., 2023).

Training and strength measurements

Regarding the types of training, four studies included traditional ST with free-weight using halfback squat (Chatzinikolaou et al., 2018; Hammami et al., 2018; Hammami et al., 2017a; Hermassi et al., 2017), three included CCT also with free-weight (Chatzinikolaou et al., 2018; Hammami et al., 2017b; Hammami et al., 2017a) but one (Hammami et al., 2019) with free-weight and machine, two studies included ST with free-weight using full squat with a linear velocity transducer (T-Force System, Ergotech, Murcia, Spain) to record bar velocity (Torres-Torrelo, Rodríguez-Rosell & González-Badillo, 2017; Torres-Torrelo et al., 2018), one included ST with machine (Technogym™  and Precor™, Madrid, Spain) (Durán-Custodio et al., 2023) and another study included optimal power output load training with leg press (Gonzalo-Skok et al., 2016). Regarding movement velocity, five studies indicated “maximal voluntary effort” (Hammami et al., 2019; Hammami et al., 2018; Hammami et al., 2017b; Hammami et al., 2017a; Hermassi et al., 2017), three studies indicated “maximal concentric repetition velocity” (Gonzalo-Skok et al., 2016; Torres-Torrelo, Rodríguez-Rosell & González-Badillo, 2017; Torres-Torrelo et al., 2018), while two studies did not report it. All the studies included exercises in dynamic contractions (isotonic), multi-joint, and bilateral except the studies by Chatzinikolaou et al. (2018) and Durán-Custodio et al. (2023), which included some unilateral exercises, besides the study by Hammami et al. (2019), which included some exercises in static contractions (isometric) and single-joint.

RSA testing

Regarding RSA testing, six studies included repeated shuttle sprint ability test (6 × 15 − 20 + 15 − 20 m; 20 s passive recovery; 180° turn) (Gonzalo-Skok et al., 2016; Hammami et al., 2019; Hammami et al., 2018; Hammami et al., 2017b; Hammami et al., 2017a; Hermassi et al., 2017) and four studies included straight-line repeated sprint ability Test (5 − 9 × 20 − 30 m; 25 s active recovery) (Chatzinikolaou et al., 2018; Durán-Custodio et al., 2023; Torres-Torrelo, Rodríguez-Rosell & González-Badillo, 2017; Torres-Torrelo et al., 2018).

Also, the measures found in the RSA tests were as follows: the best time (the fastest of all sprints), the mean time (the average time of all sprints), the total time (the sum of all sprints), fatigue index (data not provided in the studies) and the percentage decrement score calculated as follows (100 × (total sprint time ÷ ideal sprint time)) –100, where total sprint time = sum of sprint times from all sprints. Ideal sprint time = the number of sprints × fastest sprint time.

Main results of strength and complex training (EG) on RSA

Effects of strength training on the RSA

Four studies (Durán-Custodio et al., 2023; Gonzalo-Skok et al., 2016; Hermassi et al., 2017; Torres-Torrelo et al., 2018) presented significant changes in their EG in specific measures in the RSA tests (best time, mean time, and total time). These results are coincident with previous studies that only performed ST with similar exercises of the lower extremity (although with different % loads) and similar RSA tests in young male soccer players, professional male futsal players, and elite male soccer players, respectively (Sanchez-Sanchez et al., 2022; Paz-Franco, Rey & Barcala-Furelos, 2017; Spineti et al., 2016).

The improvements in initial sprint performance (i.e., best time) after heavy ST and maximal power training may be attributed to specific neural adaptations (ability to activate all high threshold MUs with maximum discharge rate rapidly) and morphological adaptations (cross-sectional area, muscle architecture, and muscle–tendon stiffness) which are achieved by executing the concentric phase of lifting with the maximum intensity. Previous research indicated that both heavy ST and explosive ST are effective in improving RFD (i.e., the ability of the neuromuscular system to express the greatest amount of force in the shortest possible time) (Maffiuletti et al., 2016; Suchomel, Nimphius & Stone, 2016). Therefore, an increase in the propulsion impulse and RFD could allow a greater force to be produced over a given time, resulting in greater acceleration and shorter ground contact times (Maffiuletti et al., 2016; Suchomel, Nimphius & Stone, 2016). Another possible explanation, according to Seitz et al. (2014), is the increases in lower-body strength levels after ST, using a medium (60–84.9% 1 RM; ES = −0.97) and high (>85% 1 RM; ES = −0.52) intensity, transfer positively to sprint performance (higher peak GRF, impulse and RFD) (r = −0.77; p = 0.0001).

For example, strong relationships have been found between 1 RM strength of the half squat and 10 m (r = 0.94; p < 0.001) and 30 m sprint time (r = 0.71; p < 0.01), respectively, in elite male soccer players (Wisløff et al., 2004). Also, relationships between full squats using relative loads that maximize the mechanical power output have been found (∼60% 1 RM; mean propulsive velocity of ∼1.00 m⋅ s −1) and best time (r = −0.76; p = 0.007), mean time of the first 3 sprints (r = −0.64; p = 0.004) and mean time of 9 sprints (r = −0.52; p = 0.04) in under-19 soccer players (López-Segovia et al., 2015). Although this suggests that strength qualities are interrelated with sprint performance, this does not prove causality; however, the present results support these findings.

On the other hand, the improvements in the mean time (Gonzalo-Skok et al., 2016; Hermassi et al., 2017) and total time (Durán-Custodio et al., 2023) could be attributed to an increase in RFD and/or power through adaptations in the stiffness of the muscle–tendon unit, which result from more efficient stretch-shortening cycles (i.e., improvement storage and reuse of elastic energy during the braking and propulsion phases) (Li et al., 2019). In particular, mean time is related to the initial sprint performance due to a perfect correlation (r = 0.93; p = 0.000) between the best time and mean time of the first 3 sprints, whereas this relationship decreases for the subsequent sprints (López-Segovia et al., 2015). In contrast, the lack of improvements in mean and total time in the other studies could be explained because despite maximal strength increases (1 RM) in six studies (Table S1), the importance of 1RM decreases when the number of sprints increases. Here, the aerobic capacity could be more relevant for improving recovery between sprints and/or avoiding the drop performance in the subsequent sprints (López-Segovia et al., 2015). Therefore, other, more specific training methods are needed to improve RSA. Previous research found a positive relationship between 1 RM strength of the half squat and best time (r = 0.811; p < 0.001) and total time (r = 0.625; p = 0.002) (Hermassi et al., 2019), also in the mean time (r = 0.50; p < 0.05) and total time (r = 0.78; p = 0.01) in male handball players respectively (Hermassi et al., 2014). In other words, the subjects with greater 1 RM performed the worst results on the RSA test.

The main mechanical factor determining RSA performance is the technical ability to apply a large forward-oriented ground reaction force vector effectively (i.e., ground reaction impulse in the forward direction of the movement) during the acceleration of each sprint instead of the amount (magnitude) of total force applied, minimizing the decrease in the stride frequency (Morin et al., 2011; Brocherie et al., 2016; Jiménez-Reyes et al., 2019). Exercises of high force to low speed with a horizontally directed force vector would be more specific for improving forward acceleration from a static start or reacceleration from each sprint (Hicks et al., 2019). Therefore, the lack of improvements in the best and mean time (Gonzalo-Skok et al., 2016; Hammami et al., 2018; Hammami et al., 2017a; Torres-Torrelo, Rodríguez-Rosell & González-Badillo, 2017) could be due in part to the lack of a specific stimulus (force vector theory) during the acceleration of each sprint (Loturco et al., 2018). However, Junge et al. (2021) did not find significant associations between hip thrust exercise and acceleration in 30 m (r = 0.18; p = 0.36). In contrast, Loturco et al. (2018) found a strong association (r = 0.93) between hip thrust with maximal acceleration (zero to 10 m) and jump squats with sprints above 40 m in both sprinters and high-level jumpers. However, studies that showed improvement in RSA used similar exercises of back squats in the vertical direction as those that did not show improvement. Therefore, the force vector theory is inconclusive because it has only been measured in single-sprint tests, not RSA tests. On the other hand, there is no relation between straight-line and shuttle sprint performance (Wisløff et al., 2004); we speculate that the lack of improvements in the studies (Gonzalo-Skok et al., 2016; Hammami et al., 2018; Hammami et al., 2017a) could be because the strength exercises do not accentuate the eccentric phase, since the straight-line sprint with one 180° change of direction requires the inclusion of a deceleration movement before a reacceleration.

Effects of complex training on the RSA

Regarding the effects of CCT on RSA, one of three studies (Hammami et al., 2019) reported a trivial significant change in their EG in specific measures (best time, mean time, and total time) in the RSA test. This result coincides with the meta-analysis of Thapa et al. (2022), where three studies on youth, junior, and professional male soccer players found a small non-significant effect (p = 0.156; ES = 0.32) in the RSA tests.

After the CCT, improvements in the RSA test may be specifically related to optimizing the entire spectrum of the force-velocity curve in the same session (Thapa et al., 2022). Heavy load training and low speed (e.g., squat 90% of 1RM) increase the excitability of the MUs (i.e., recruitment and synchronization of MU higher order/high threshold and the potentiation of reflexes), creating optimal and favorable conditions for plyometric or power training (e.g., CMJ) after maximal muscle contractile activity (Freitas et al., 2017; Cormier et al., 2022). These neural adaptations may ultimately lead to a higher RFD (Maffiuletti et al., 2016) and improved efficiency of the stretch-shortening cycle. However, the time course of the potentiation effect also depends on the time course of fatigue (the net balance determining if there is an increased or decreased output).

Previous studies compared CCT vs. ST. Spineti et al. (2016) observed that CCT induced a significant improvement in percentage decrement (moderate ES; −0.91) and mean time (moderate ES; −0.85) when compared with ST in under-20 elite male soccer players. Hammami et al. (2017a) observed that CCT presented more significant changes for the mean (p ≤ 0.001) and total time (p ≤ 0.01) compared to ST in junior male soccer players. This suggests that CCT could be an effective strategy to optimize and improve performance during the competitive season compared to ST alone. However, the methodology used in the present systematic review and the lack of studies do not allow a comparison of the results to conclude that CCT is superior to ST. The lack of improvements in the studies by Hammami et al. (2017b); Hammami et al. (2017a), could be due to the lack of a specific stimulus, as aerobic capacity is an important factor in recovery between sprints; CCT does not develop this quality. Also, it could be due to a nonspecific recovery post-CA (i.e., the relation between the CA on subsequent exercises) (Cormier et al., 2022); however, none of the studies reported the pause intra-contrast rest to affirm it.

Effects of strength and complex training on fatigue

Fatigue was assessed through two measures (fatigue index or percentage decrement); only one study (Durán-Custodio et al., 2023) presented a large significant change (p = 0.003; ES = 1.25), except for the study by Hermassi et al. (2017) that did not report this measure (does not allow us to know its meaning). However, this unique positive result does not coincide with other studies (Paz-Franco, Rey & Barcala-Furelos, 2017; Spineti et al., 2016). These results should be interpreted with caution because a better initial performance of the first sprint will commonly lead to a high fatigue index (an indicator of the loss of performance between the best and the worst sprint), given that there is a positive correlation with the decrease performance in the following sprints; therefore, the subjects with higher anaerobic power reserve could be impaired in subsequent bouts (Girard, Mendez-Villanueva & Bishop, 2011). Probably, this decrease in performance would be a combination of neural factors (i.e., lower recruitment and discharge frequency fast twitch MUs), as well as alterations in intermuscular coordination patterns (agonist-antagonist ratio) (Girard, Mendez-Villanueva & Bishop, 2011) and muscular (i.e., accumulation of hydrogen ions (H⁺), inorganic phosphate and extracellular potassium secondary to decreased sodium-potassium pump ATPase activity), also limitations in the energy supply of phosphocreatine availability, anaerobic glycolysis and oxidative metabolism (Girard, Mendez-Villanueva & Bishop, 2011), that may alter the sprint mechanics, such as reductions in forward-oriented ground reaction force (total force), increase in the contact time (i.e., decreased efficiency of the stretch-shortening cycle), increased vertical displacement of the center of mass during the braking phase and increased swing time, reductions in vertical stiffness and stride frequency (Morin et al., 2011; Brocherie et al., 2016; Jiménez-Reyes et al., 2019). In summary, ST or CCT would influence the neuro-mechanical properties of initial sprint performance; however, other training methods targeting oxidative capacity, phosphocreatine recovery, and H⁺ buffering would be the most suitable for improving recovery between sprints and/or fatigue resistance (Bishop, Girard & Mendez-Villanueva, 2011; Gaitanos et al., 1993). Nevertheless, Edge et al. (2006) and Hill-Haas et al. (2007) showed in females recreationally active in various team sports that high-repetition resistance training that includes a high metabolic load (because of short rest periods 20 s) improved not only single-sprint performance but also the performance of subsequent sprints (peak and mean power). Although those studies were evaluated on a cycloergometer and did not comply with the principle of specificity, high-repetition resistance training is an effective method to improve the H⁺ regulation and avoid the drop performance in subsequent bouts.

Some limitations can be identified. Firstly, since allocation was concealed, the moderate quality of the studies could produce systematic biases in random allocation. Also, it is difficult to mask the application of treatments or programs. The small number of studies on the effectiveness of ST on RSA does not allow us to compare the results regarding the characteristics of the subjects and intervention of the training. Furthermore, the heterogeneity of the characteristics of the subjects and intervention are factors to consider, mainly because subjects with lower levels of strength and experience could have a more considerable margin of adaptations. Future studies should include exercises in all the spectrum of the force-velocity curve, including exercises in the horizontal direction, ballistic where higher RFDs can be achieved, and high repetition with short rest periods. Finally, the diversity of RSA tests and the lack of measurements in 5 studies do not precisely allow us to know the interventions’ efficacy.

Practical applications

According to the findings of the articles reviewed in the present study, coaches should consider ST and CCT to improve RSA in team sports. Specifically, training maximal strength, RFD, and power using a combination of heavy/low load strength exercises while executing the concentric phase with maximum intensity may improve the propulsion impulse over the ground during the acceleration or reacceleration of each single sprint. Including exercises specific in a horizontal direction should also be considered to minimize the reduction in forward-oriented GRF during repeated sprinting, which could lead to important changes in RSA.