This study was guided by the Preferred Reporting Items for Systematic Evaluation and Meta-Analysis (PRISMA) (26) and registered in the PROSPERO database under number CRD42022340727. By searching databases such as CNKI, Wanfang database, Pubmed, Web of science, and Cochrane library, the search time period is from the creation of the database to April 2022, with the last search date being April 27, 2022. The search formula in Chinese and English was (FMS OR fundamental movement skill OR basic motor skill OR gross motor skill OR physical fitness OR gross motor skill OR speed OR run OR jump OR object control OR stability OR balance) AND (children OR kid OR student) AND (football OR soccer).

In this study, two researchers conducted the quality evaluation and data extraction, a third researcher was involved in the joint work for the areas of disagreement. After extraction, the difference between the pre- and post-intervention means and the standard deviation was calculated by subtracting the base-measure value from the post-measure value, and the standard deviation was calculated by “SD difference ² = SD _{base value} ² + SD _{post value} ² − 2*R*SD _{base value} * SD _{post value}, R = 0.5″ (29). According to the Cochrane Handbook of Systematic Evaluation (30), the baseline is balanced and comparable between groups in randomized controlled trials, and theoretically there is no difference between the comparison of post-intervention measurements and the comparison of pre- and post-intervention differences. In this study, post-measure values of object control indicators were analyzed for comparison because of the relatively small number of study and missing baseline for some indicators.

Cochrane Handbook for Systematic Reviews of Interventions (30) were used to score the included studies for (1) random sequence generation; (2) allocation concealment; (3) blinding of participants and personnel; (4) blinding of outcome assessment; (5) incompleteness of outcome data; (6) selective reporting; (7) other bias. Each criterion has three options of low risk, high risk, and unclear risk, and can be classified in category A when the number of low risks in the study is ≥4, in category B when the number of low risks in the study is ≥2 and <4, and in category C when the number of low risks in the study is <2. As shown in Figures 1, 2, only 4 studies were standardized in the generation of random sequences generation (14–16, 31), and most of the studies were not assigned to concealment, because the sports interventions could not be set up double-blind, most of the experiments were mainly instrumental measurements, and the subjective influence of the evaluator on the experimental results was small, Therefore, “blinding of outcome assessment” scores are usually high. A total of 14 included studies met the criteria of category A (13–18, 20, 31–37) and 2 met the criteria of category B (12, 19), and the overall quality of the study was high.

Meta-analysis was performed using Review Manager 5.4 software (38). Heterogeneity was tested using I², and heterogeneity was considered to exist when I² > 50% (39), and a random-effects model was selected; while I² < 50% was used to study homogeneous effects and a fixed-effects model was selected. Standardized mean difference (SMD) and 95% confidence interval (95% CI) were chosen as the effect scale to combine effect sizes, and small effects were considered when the absolute value of SMD was 0–0.4, moderate effects were considered when 0.4–0.8, and high effects were considered when greater than 0.8, and differences were considered significant when P < 0.05, and subgroup analysis was performed for highly heterogeneous study results (39).

Through the search of Chinese and English study, a total of 2,036 studies were retrieved, 595 duplicate studies were excluded, 1,391 studies were excluded after reading the titles and abstracts, 37 studies were excluded by reading the full text, and 16 studies were finally screened (12–20, 31–37), including a total of 7 studies in English (12, 14–16, 31–33) and 9 studies in Chinese (13, 17–20, 34–37) (Figure 3).

In all included studies(Tables 1, 2), there were 3,121 children aged 7–13 years, including 1,958 in the experimental group and 1,163 in the control group; there were 7 studies with more than 100 participants (14–16, 20, 31, 34, 35), 8 studies with 20–100 participants (12, 13, 17–19, 33, 36, 37), and 1 study with less than 20 participants (32); 5 studies interventions were only for boys (13, 15, 17, 32, 33), 7 studies were co-educational interventions (14, 19, 20, 31, 34–36), and 4 did not specify the proportion of interventions (12, 16, 18, 37). The interventions were all based on soccer practice and 4 studies stated the intensity of the intervention (15, 19, 36, 37); a total of 10 studies (13, 14, 17, 18, 20, 31, 33–36) for less than 20 weeks and 6 studies (12, 15, 16, 19, 32, 37) for more than 20 weeks; a total of 11 studies practiced at least 3 times per week (13, 15–19, 32–36) and 5 less than 3 times (12, 14, 20, 31, 37); 6 studies with weekly intervention time less than 120 min (13–15, 18, 20, 31) and 10 with intervention time more than 120 min (12, 16, 17, 19, 32–37); a total of 15 studies with baseline testing and comparison before the experiment (13–20, 31–37); for test results, a total of 13 studies with less association with rater subjectivity, such as horizontal jump, linear sprint, and Closed-eye single-leg test (14–20, 31–37), and a total of 2 studies with rater subjectivity scoring (12, 13); all included studies were published after 2013.

From Figure 7, the effect of soccer on children's object control χ² = 71.47, df = 11 (P < 0.00001), I² = 85%, therefore, Meta-analysis was performed using a random effects model with SMD = 1.32, 95% CI: [0.80, 1.85], P < 0.0001, SMD > 0.8, indicating that the soccer intervention had a better facilitation effect on children's object control, and there were significant differences between subgroups.

In stationary dribbling, χ² = 1.42, df = 1 (P = 0.23), I² = 30%, there was homogeneity between the two groups, SMD = −0.03, 95% CI: [−0.56, 0.50], P = 0.9, |SMD| < 0.4, and the diamond-shaped small square intersected with the null line, indicating that the soccer sport intervention had no improvement effect on children's stationary dribbling.

In catching ability, χ² = 18, 71, df = 1 (P = 0.08), I² = 95%, there was heterogeneity between the experimental and control groups, SMD = 1.77, 95% CI: [−0.67, 4.21], P < 0.0001, SMD > 0.8, and the diamond-shaped cubes intersected the null line, indicating that the soccer sports intervention had no effect on the children's ball catching ability.

There was heterogeneity between the two groups of overhand throwing with χ² = 3.03, df = 1 (P = 0.08), I² = 67%, SMD = 1.41, 95% CI: [0.53, 2.28], P = 0.02, SMD > 0.8, and the diamond-shaped cube was located to the right of the null line and did not intersect, indicating that the soccer sport intervention had an improving effect on children's overhand throwing.

In the underhand rolling, χ² = 2.7, df = 1 (P = 0.1), I² = 63%, there was heterogeneity between the experimental and control groups, SMD = 1.42, 95% CI: [0.59, 2.24], P = 0.0008, SMD > 0.8, and the small diamond-shaped squares were located to the right of the null line and did not intersect, indicating that the soccer sport intervention had an improving effect on children's underhand rolling.

There was heterogeneity between the two groups of two-hand striking a stationary ball χ² = 8.25, df = 1 (P = 0.004), I² = 88%, SMD = 1.91, 95% CI: [0.31, 3.51], P = 0.02, SMD > 0.8, and the small diamond-shaped square was located to the right of the null line and did not intersect, indicating that the soccer practices had an improving effect on children's two-hand striking a stationary ball.

Kicking ability is an important manifestation of lower limb object control ability, χ² = 0.84, df = 1 (P = 0.36), I² = 0%, SMD = 1.59, 95% CI: [1.08, 2.1], P < 0.0001, SMD > 0.8 between the two groups, the small diamond-shaped squares were located to the right of the null line and did not intersect, indicating that there was a height effect and that the soccer practices had an effect on children's lower limb object control ability was improved.

In this study, a total of seven included studies were conducted on children's stabilization ability. The closed-eye single-leg test and flamingo balance were used as an index. From Figure 8, χ² = 17.03, df = 8 (P = 0.03), and I² = 53% between the two groups of the closed-eye single-leg test, which can be considered as heterogeneity between the two groups, and therefore a random-effects model was performed. The results showed that the sample size was 282 cases, SMD = 0.87, 95% CI: [0.48, 1.25], P < 0.0001, and the diamond-shaped small squares were located to the right of the null line and did not intersect, indicating that there was a high effect and that the soccer practices had a significant contribution to the children's stabilization ability. Also we can see that pre- and post-intervention of the experimental group χ² = 13.83, df = 8 (P = 0.09), I² = 43%, SMD = −0.83, 95% CI: [−1.07, −0.59], P < 0.00001, |SMD| > 0.8, the small diamond-shaped square was located to the left of the null line and did not intersect, it indicates that the level of standing on one foot with eyes closed was significantly higher in the experimental group after the intervention compared to before the intervention.

In Figure 9, χ² = 1.32, df = 2 (P = 0.52), and I² = 0% between the two groups of the flamingo balance test, which can be considered homogeneous between the two groups, and the fixed effects model was performed. The results showed that the sample size was 829 cases, SMD = −0.16, 95% CI: [−0.31, −0.02], P = 0.03, and the small diamond-shaped squares were located to the left of the null line and did not intersect, indicating that there was a small effect and that the soccer practices promoted children's stabilization ability. We can see that pre- and post- intervention in the experimental group χ² = 4.56, df = 2 (P = 0.1), I² = 56%, SMD = 0.12, 95% CI: [−0.1, 0.35], P = 0.28 indicating that the difference in balance ability after the intervention in the experimental group was not statistically significant.

In this study, the total intervention period (weeks), weekly intervention frequency, and length of each lesson(min) were considered to be factors influencing the study results, and subgroup analysis of one of them alone would result in the study outcomes being confounded by the other two factors. Therefore the “total intervention period (weeks)” × “weekly intervention frequency” × “length of each lesson(min)” was used to obtain the total intervention duration, which was used to explore the relationship between intervention time and effect. In terms of the degree of distinction as well as the rationality of the intervention, “15 weeks * 3 times/week * 40 min/time  =  1,800 min” was used as the dividing line. In Table 3, the|SMD_{1,800 min below}| < |SMD_{1,800 min more}| of linear sprint ability and horizontal jump, the intervention above 1,800 min was a moderate and high effect, while below 1,800 min was a small effect, so the training effect above 1,800 min promoted the development of linear sprint ability and horizontal jump more. Flamingo balance had no statistically significant difference between groups below 1,800 min, while there was a large effect above 1,800 min. The duration of the closed-eye single-leg test intervention was below 1,800 min, and there was a high effect. There was no statistically significant difference between groups above or below 1,800 min for the countermovement jump.

In this study, the studies were divided into two subgroups based on age: 7–9 years old and 10–13 years old. In Table 4, there was no statistically significant difference between the groups of linear sprint ability and horizontal jump in children aged 10–13 years old, but there were medium and small effects in 7–9 years old, and it can be concluded that the effect of soccer practices on linear sprint ability and horizontal jump in children aged 7–9 years old was better than that in 10–13 years old, and there was no statistically significant difference between the groups of countermovement jump and flamingo balance in both 7–9 years old and 10–13 years old, while the eyes closed Closed-eye single-leg test SMD_7–9years < SMD_10–13years, indicating that children aged 10–13 years had better intervention effects on the ability to stand on one leg with eyes closed.

The study was divided into three subgroups: “male”, “female”, and “mixed” according to the gender of the intervention subjects. In Table 5, linear sprint ability and closed-eye single-leg test |SMD _female| >|SMD _male| indicated that the soccer intervention was more effective for girls. There was a high effect of horizontal jump in boys, but no statistically significant effect in girls. There was no statistically significant effect for both subgroups of countermovement jump and flamingo balance.

Locomotion is an indispensable ability in soccer, and linear sprint ability and jumping ability are the two main indicators to judge locomotion. The results of study have identified the facilitative effect of soccer practices on linear sprint ability, and there are relevant studies that have demonstrated this (17, 35, 36). It is related to the soccer itself including running, kicking, and shifting movements, but the findings of M.N. Larsen (15) in 2017 diverge from the findings of this study, whose intervention period was 43 weeks long and with an exercise intensity of 70%–90% HRmax, but the weekly intervention duration was only 60 min, which may indicate that long-period, short-per-intervention methods are not effective in improving linear sprint ability. Jianing He (36) showed a more significant improvement in linear sprint ability after 12 weeks of soccer technique and ball handling intervention at 270 min per week, which may indicate that a longer weekly intervention duration can better promote linear sprint ability improvement. A subgroup analysis showed that interventions longer than 1,800 min were more effective than those below 1,800 min, that interventions were more effective in children aged 7–9 years than in those aged 10–13 years, and that interventions were more effective in girls than in boys, suggesting that soccer interventions should be conducted at ages 7–9 years to improve linear sprint ability. Meanwhile, it is difficult to compare the effects of factors such as the ratio of physical fitness training to soccer technique training, the ratio of different soccer technique training, and the ratio of aerobic to anaerobic training on linear sprint ability in the intervention because subgroup analysis of the intervention content was not conducted in this study.

Secondly, the results of the study showed that the soccer practices had a facilitating effect on horizontal jump, and Yan Zheng (37) showed a significant improvement in children's jumping ability after a 32-week * 1 time/week * 120 min/time intervention, which might indicate that long-term low-frequency high-intensity prolonged soccer sports have a better effect on jumping ability, while C. Ørntoft (14) showed that after a 12-week * 2 times * 45 min intervention, the intervention effect in the experimental group did not differ from the control group, considering its similarity to the intervention content of Yan Zheng (37), we therefore hypothesized that a low frequency, short duration intervention would be ineffective. The subgroup analysis of the study also proved the important role of intervention duration and age on horizontal jump ability, as shown by the fact that the effect of intervention duration >1,800 min was better than that of <1,800 min, which also reminds us that the intervention duration should be extended in future interventions to improve the effect, while there was no significant difference between the intervention effect of boys and girls. Meanwhile, the effect of practice intensity on horizontal jump was studied (40), and in the medium- and high-intensity soccer practice, soccer players who received the high-intensity intervention performed significantly better in the horizontal jump than those with moderate practice intensity, this is consistent with Fei Xin's conclusion. For the countermovement jump, the results showed no difference between the experimental and control groups after the intervention, perhaps caused by the same participation of the control group in the physical education program, which is consistent with the findings of Xulong Li (19) and N. Cvetković (33), in addition to the findings of some of the studies (15, 41) that could not be combined due to different data presentation, which supports the conclusion. Whereas there were differences before and after the intervention in the experimental group, indicating that the soccer practices were effective for children, but the effect did not differ from participation in a general physical education program, the results of the countermovement jump should be treated with caution due to the small number of included studies.

The present study found that soccer had a good promotion effect on children's object control in general, but there was no statistical significance between groups on stationary dribbling and catching. This is consistent with the conclusion reached by Yangyang Guo (13), in which there was a significant improvement in two-hand striking a stationary ball and throwing ability, and slightly higher scores on stationary dribbling and catching than the control group but no significant difference, he concluded that kicking movements involve precision control of the limbs on the target, and long-term practice tends to improve proprioception to help develop object control in the upper limbs. In contrast, the results of the study by P.F. Nazario (12) yielded the same results in terms of the promotion of stationary dribbling, but with a significant increase in catching ability, and it concluded that the characteristics of the experimental subjects, the environment they were in, and the time of the intervention were all influential factors. Considering the small difference between the two in terms of age and that the intervention by Yangyang Guo (13) was not traditional soccer instruction but ball practice, this study hypothesized that when the intervention included passing and catching football, upper limb throwing and catching accuracy would also be enhanced, but when the intervention was only ball practice and lacked passing and catching football, the enhancement of all upper limb object control abilities would be limited. Since this study included less study on upper limb object control, the specific effect still needs to be verified in the future. And since soccer itself is a sport involving lower limb control, most FMS assessment tools use soccer as a test of lower limb object control, so the effect of soccer on lower limb object control is direct, and this study used kicking in TGMD-2 as a test item, and the results showed that soccer had a facilitating effect on the effect of lower limb intervention, and not only that, Xulong Li (19) found that children's ability to change direction and dribble was improved after the soccer intervention, which can also represent the improvement of children's lower limb object control ability, although the specific intervention effect can be affected by various aspects such as training content, time, and characteristics of the experimental subjects, but the appropriate intervention has a positive effect.

Balance is an important prerequisite for the development of motor skills, and examining children's balance will help future research and interventions that will lead to better overall movement skills (42). The results of the study point to differences in the results of the two different testing methods. In the four studies tested by Closed-eye single-leg test, the experimental group significantly improved compared to the control group, which is consistent with the study of Zhongqiu Ji (20), which concluded that movements such as running, paddling and dribbling have a helpful effect on stability, while the fatigue resistance of the calf muscles is also a major factor affecting balance, and soccer contains both aerobic and anaerobic exercises that promote fatigue resistance in the calves. And Xulong Li (19) and Shuxiang Wang (17) although there was no significant difference between the experimental and control groups, there was a significant difference between the posttest of the experimental group compared to the base test, as the balance ability can be effectively improved by practice can be improved (43), which may be related to the participation of the control group in the normal physical education program. The intervention effect was better in girls and 10–13 years old than in boys and 7–9 years old in the subgroup analysis, indicating that there are gender differences in FMS (44, 45) and it is more appropriate to intervene in soccer at the age of 10–13 years old in children, but the present findings do not coincide with the developmentally sensitive period of balance ability, and considering that the present study did not analyze factors such as practice content, ratio, and intensity, more research in this area is necessary in the future. However, the analysis results of the flamingo balance test showed that the balance ability was not improved significantly after the intervention, which can have an impact on the accuracy of the results due to the small amount of study, but it is worth noting that C. Ørntoft (32) in the present study performed an intervention with a total duration of 1,080 min, balance ability did not improve, while subgroup analysis, for >1,800 min study showed that the soccer practices had a better effect on the flamingo balance test, indicating that the increase in the total duration of the intervention was a major factor in improving the effect of the intervention. Secondly, related scholars (40) conducted flamingo balance tests for 11-year-old soccer players after medium- and high-intensity soccer interventions with regular children, and the results showed that the balance ability of soccer players was significantly better than that of regular children, and the balance ability of soccer players with high exercise intensity was better than that of athletes with medium intensity.

The search terms in this study only included Chinese and English, and some of the studies could not be viewed in full text, resulting in fewer included study for the countermovement jump, object control ability, and flamingo balance tests, which had some impact on the reliability of the results. In the quality evaluation of the included study, most of the study failed to accurately state the principle of random assignment, and in some experiments where the subjectivity of the raters was relatively high, the scores were not blinded, and some of the literature did not report the withdrawal, resulting in a lower level of quality of the study. In the subgroup analysis, the small number of studies for some indicators and the difficulty in unifying the intervention intensity and proportion of experimental subjects across the study resulted in the inability to further analyze the interventions.