Systematically searches were conducted on four electronic databases (EBSCO, PubMed, Scopus, and Web of Science) until 15 June 2022. All possible keyword combinations from the following two groups were included in the search algorithm: (“plyometric” OR “plyometric training” OR “stretch-shortening cycle”) AND (“kicking speed” OR “kicking velocity” OR “kicking distance”). Additionally, a manual search was performed to find relevant publications in the reference lists of each article that was included. Only peer-reviewed, English-language research articles were included in the search.

The meta-analysis included studies that meet all of the following criteria: 1) population: Healthy soccer players with no previous injury history; 2) intervention: Plyometric training was utilized in the experiment group; 3) comparison: Regular soccer training program was applied in the control group; 4) outcome: Kicking performance was measured, including kicking speed or kicking distance; and 5) study design: The study needed to be a randomized controlled trial.

The studies that met the eligibility requirements were selected independently by two reviewers (YZ and DL). If there were inconsistencies between the two reviewers, they discussed the rationale for their choices, and if one reviewer realized he/she had made an error, then the process was concluded (Devereaux et al., 2002). All disagreements were resolved in this manner.

One reviewer (YZ) gathered the following information from all included studies and entered it into Microsoft Excel. A second reviewer (DL) verified the accuracy of the following data: Author name(s), publication year, title, sample size, participant characteristics (age, gender, and skill level), intervention duration, study design, and kicking performance outcome(s). In the present study, the dominant leg’s kicking performance was chosen as the primary criterion (García-Ramos et al., 2017). If the dominant leg was not specified, the performance with the right leg was chosen (García-Ramos et al., 2017).

The data formats of the studies included in the meta-analysis are of two types: One data format is pretest-posttest-control design data that includes the means and SDs of the pre-test score and post-test score for the experimental group (plyometric training group) and control group, and the other one is the means and SDs of the pre-test score and the gain score (the post-test score minus the pre-test score) for the experimental group (plyometric training group) and control group.

For the data from the pretest-posttest-control design, Morris (2008) has proposed a reasonable and validated method to estimate effect sizes. The method favored an effect size based on the mean pre-post change in the treatment group minus the mean pre-post change in the control group, divided by the pooled pre-test SD (see Eqs. 1, 2) (Morris, 2008). Then, according to Borenstein et al. (2009), Cohen’s d converted to Hedge’s g (see Eqs. 3, 4).

For the data from the pre-test score and the gain score with their SDs, as the studies included in the meta-analysis provided the pre-test score and its SD, the pooled pre-test SD could be estimated by Eq. 1 (Morris, 2008). In addition, as the included studies provided the mean of the gain score (the post-test score minus the pre-test score) for the experimental group and control group, Cohen’s d could be imputed by Eq. 2 (Morris, 2008). Then, the estimations of Correction factor J and Hedge’s g are the same as the above-mentioned Eqs. 3, 4).

In cases where insufficient data were provided to assess the effect size (ES), attempts were made to contact the study’s authors. For the correlations between the pre- and post-measurements, we followed the recommendations from Follmann et al. (1992) and Higgins et al. (2022) and used an estimated correlation of 0.5. For studies that did not report SDs but provided 90% CIs of means, the 90% CIs would be transformed into SDs, according to Higgins et al. (2022). [ P o o l e d   P r e − t e s t   S D = n Exp − 1 S D P r e , Exp 2 + n C o n − 1 S D P r e , C o n 2 n Exp + n C o n − 2 ] (1)

(Morris, 2008) [ C o h e n ’ s   d = M P o s t , Exp − M P r e , Exp − M P o s t , C o n − M P r e , C o n P o o l e d   P r e − t e s t   S D ] (2)

(Morris, 2008) [ C o r r e c t i o n   f a c t o r   J = 1 − 3 4 n Exp + n C o n − 2 − 1 ] (3)

Borenstein et al., 2009 H e d g e ’ s   g = J × C o h e n ’ s   d (4)

Borenstein et al., 2009

The primary analysis was performed to estimate the overall effect of the PT intervention on kicking performance by using a meta-analysis based on the random-effects model in view of the heterogeneity within the program of PT intervention, the characteristics of participants, and the measurement of kicking performance. The ES was estimated using Hedge’s g with 95% CIs, and its magnitude was determined on the following scale: 0.2–0.5 for small, 0.5–0.8 for moderate, and >0.8 for large (Cohen, 1988).

The I ² index was used to determine the degree of study heterogeneity, and the thresholds of heterogeneity for low, moderate, and high were 25%, 50%, and 75% (Higgins et al., 2003). In order to facilitate the comprehension of between-study heterogeneity beyond the existing list of included studies, τ ² statistics were also provided (Rücker et al., 2008). To determine the impact of moderating factors (i.e., age, gender, skill level, and intervention duration) on the overall estimate, subgroup analyses using Cochran’s Q test and I ² were conducted.

Publication bias was evaluated using a funnel plot (Sedgwick and Marston, 2015) and Egger’s regression tests (Egger et al., 1997). Additionally, a sensitivity analysis was conducted by sequentially deleting each study to determine whether the pooled estimate was influenced excessively by a single study (Tobias, 1999).

Meta-analysis and subgroup analyses were carried out with the Comprehensive Meta-Analysis (CMA) 3.0 software (Borenstein, 2022). Two-sided tests were used in all analyses, with a p-value of less than 0.05 considered statistically significant.

The methodological quality of the included studies was measured using the Physiotherapy Evidence Database (PEDro) scale (Blobaum, 2006). The PEDro scale has 11 items that are used to rate the methodological quality. Each item that is achieved adds one point to the total PEDro score, which ranges from 0 to 10, and item 1 was not included in the study’s methodological quality evaluation (Maher et al., 2003). The quality assessment of the included studies was interpreted as follows: 1–3 points of the total PEDro score equaled poor quality, 4–5 points equaled moderate quality, and 6–10 points equaled high quality (Ramírez-Campillo et al., 2021).

The flow chart for the literature selection is shown in Figure 1. The initial search yielded a total of 1,319 records, including 129 from EBSCO, 343 from PubMed, 384 from Scopus, and 463 from Web of Science. After deleting duplicates with 625 records being eliminated, the left 694 records were screened for title and abstract. After screening the title and abstract, 670 records were deleted, and the full text of the remaining 24 articles was examined for compliance with the study selection criteria. Of these, eight articles were excluded. In total, 16 studies were finally included in the meta-analysis.

The essential characteristics of the 16 studies (17 ES) considered in the meta-analysis are summarized in Table 1. These are all randomized controlled trials. The sample size spans from 15 to 76, totaling n = 553 participants. The average age of the participants ranges from 10 to 23 years old. Furthermore, 11 studies focused solely on male athletes, while five studies focused solely on female athletes. In addition, 12 studies recruited non-elite players, while the remaining four used elite players. The intervention duration of the included studies ranges from 6 to 14 weeks. The most common training frequency was two sessions per week (n = 12). Regarding kicking performance measures, eight studies have utilized the kicking distance test, and eight have adopted the kicking speed test.

Figure 2 depicts the overall estimated ES for the efficacy of PT on kicking performance. There was a large-sized significant enhancement in kicking performance after PT implementation (Hedge’s g = 0.979, 95% CI [0.606, 1.353], p < 0.001, τ ² = 0.449). In addition, a large proportion of the between-study variance was demonstrated according to an I ² value of 76.180%.

The findings of the subgroup analysis are presented in Table 2. No significant between-group difference was found when subgroup analyses were conducted based on participants’ age (Q = 0.534, p = 0.465), gender (Q = 0.611, p = 0.434), skill level (Q = 0.321, p = 0.571), and the intervention duration (Q = 0.167, p = 0.683), suggesting that these variables did not moderate the effect of PT on kicking performance.

A visual examination of the funnel plot (see Figure 3) revealed a slightly asymmetrical distribution of the included studies; nevertheless, Egger’s regression test (t = 0.332, p = 0.745) indicated no significant publication bias. Furthermore, the results of the sensitivity analysis showed no individual study considerably altered the overall effect of PT on kicking performance because when one study was eliminated at a time, the pooled estimate ranged from g = 0.880 (95% CI [0.539, 1.221]) to g = 1.051 (95% CI [0.688, 1.413]).

The total PEDro score of each included study is displayed in Table 1. According to the interpretation of the total PEDro score (Ramírez-Campillo et al., 2021), four studies receiving 4–5 points were deemed to have moderate methodological quality, while the remaining twelve studies obtaining 6–10 points were assessed as being of high methodological quality.

The improvements in kicking performance after PT could attribute to the following hypotheses.

First, PT potentiation is first explained mechanically as the use of elastic energy that has been stored. Our legs, for instance, behave like springs when hopping, jumping, and running; they compress when they make contact with the ground and store energy before rebounding when we push off and release it (Hobara et al., 2008). It is now understood that the tendon serves as the main location for storing elastic energy. (Kubo et al., 1999; Lichtwark and Wilson, 2007). It is theorized that the magnitude of elastic energy stored is proportional to the applied force and the generated deformation (Zatsiorsky et al., 2020). According to previous studies, elasticity significantly improves motor output in sports motions. (Kubo et al., 1999; Lichtwark and Wilson, 2007; McBride et al., 2008).

Second, the stretch reflex and other involuntary nervous system functions have been linked to SSC potentiation (Dietz et al., 1979; Bosco et al., 1981). In a nutshell, the muscle spindles trigger a reflex action upon the musculotendinous unit’s pre-stretching. Type Ia afferent fibers provide a neural impulse to the spinal cord when the muscle spindles notice an abrupt increase in muscle length. The alpha motor neuron and type Ia afferent fibers synapse after that, causing a reflexive muscular movement (Kandel et al., 2000). It is thought that when voluntary and involuntary (reflexive) motions are coordinated at the right time, the agonist’s muscle is activated to its supramaximal concentric level.

Thirdly, according to additional hypotheses, the eccentric action of the prestretch during PT increases the muscle’s active state (Bobbert and Casius, 2005), reducing the electromechanical delay, or the time interval between stimulation and mechanical output, in order to produce force in less time. (Cavanagh and Komi, 1979). As a result, the muscle’s functioning range is expanded, enabling it to produce more force and impulse during the movement’s concentric phase. Moreover, it has been hypothesized that a prestretch and subsequent lengthening would position the muscle in a more advantageous area of the length-tension relationship (Gordon et al., 1966a; Gordon et al., 1966b), increasing force production at the start of and throughout the concentric action as a result (Ettema et al., 1992).

Fourthly, PT could enhance explosive leg power. Kicking is a “whip-like” sequence in which the foot is the quickest link in the open kinetic chain (Carr and Carr, 2004). The velocity of the foot upon impact with the ball is a significant predictor of ball speed and kicking distance (Young and Rath, 2011). As the identifying feature of PT, SSC can assist neuronal and musculotendinous systems in producing maximum force in a minimum time (Markovic and Mikulic, 2010). Bober et al. (1987) have found that maximum knee extension velocity can be 20% higher during a rapid SSC compared to a concentric-only movement in a seated position. Therefore, PT improves kicking performance, possibly by increasing the explosive power of the legs.

Finally, Another possible explanation for PT’s benefits on kicking performance is the enhancement of players’ balance ability after PT intervention. Soccer is a sport that requires bipedal movement, which means standing on one leg while kicking the ball with the other leg (Bressel et al., 2007). That means the player must keep the balance at high speed while kicking the ball powerfully. A growing body of research demonstrated the significance of balance on kicking performance. The research of Cerrah et al. (2016) showed that the kicking velocity of the dominant leg correlated with the dynamic balance ability of the dominant leg and both legs. Furthermore, PT training has shown its advantages in improving players’ balance performance (Ramírez-Campillo et al., 2015a; Alikhani et al., 2019). Therefore, the PT intervention may improve soccer players’ kicking performance by increasing players’ balance ability.

The subgroup analyses show that potential factors (i.e., age, gender, skill level, and intervention duration) do not play a moderating role in the PT effect. Those results may be because of the uneven covariate distribution (e.g., five ES related to participants ≥18 years old vs. 12 ES related to participants <18), implying that the subgroup analyses may not be able to provide accurate results (Richardson et al., 2019). In addition, the unequal number of participants presenting data to these subgroups (e.g., n = 117 participants in the subgroup of age ≥18 years old vs. n = 436 participants in the subgroup of age <18) suggests that the analyses may not be capable of uncovering subgroup differences (Richardson et al., 2019). It would be prudent for more research to be conducted in the research field in order to investigate the subgroup effect.

PT is a recommended method for enhancing kicking performance. To reach a high level of kicking performance, strength, and conditioning professionals may incorporate PT into soccer players’ overall training program in addition to well-known training approaches such as strength and power training in the weight room. Based on different objectives, coaches can select different forms of PT, such as the drop jump, the countermovement jump, the squat jump, alternate-leg bounding, or hopping. In the practical application process, strength and conditioning professionals can use PT alone or in conjunction with other training methods (Sáez de Villarreal et al., 2015; Bouteraa et al., 2020).

Some potential limitations of this study must be outlined. First, because the majority of the studies examined had small sample sizes, the impact of PT may have been overestimated compared to research with larger sample sizes. The effectiveness of PT on kicking performance should thus be examined and replicated using larger sample sizes to boost statistical power. In addition, the binary classification of the continuous data (e.g., the intervention duration > 7 weeks vs. ≤ 7 weeks) may lead to residual confounding and decreased statistical power (Altman and Royston, 2006). Finally, although the included studies did not specify any side effects associated with the PT intervention, it is unclear whether the researchers attempted to document all possible adverse events comprehensively. Therefore, researchers should provide more details on the possible injury, pain, and potential adverse impacts of PT training in future studies to enhance our understanding of the safety of PT.