Abstract
The overhead throw is a fundamental technical-tactical ability for overhead sports. The purposes of this review are to assess the effect of resistance training (RT) in enhancing throwing velocity in athletes and to investigate the relationships between age or gender in this effect. Control group trials were identified through looking up electronic databases with a search span of 10th December 2020. Only studies which have control groups within research design, subjects randomly assigned to groups, healthy athletes with experience in the sport, an intervention consisting of a supervised RT program of a minimum duration of 4 weeks, and assessment of sport-specific throwing velocity were taken into account for this meta-analysis. A total of 16 studies with 424 subjects were deemed eligible per the inclusion criteria. The overall pooled analysis demonstrated that a large effect was observed for throwing velocity outcomes (ES 1.10; 95% CI 0.64–1.57; p < 0.00001). Differences were due to gender, with male (ES 1.12; 95% CI 0.55–1.78; p < 0.0001) and female athletes (ES 1.22; 95% CI 0.25–2.20; p < 0.00001). And due to age, with teenager athletes (ES 0.49; 95% CI -0.18–1.17; p = 0.04) and adult athletes (ES 1.34; 95% CI 0.64–1.92; p < 0.00001). Throwing velocity enhancement after RT was greater for women than for men, and for adults more than underage subjects. RT should last at least four weeks, with 2–3 sessions each week, and with any available implement.
Keywords
Introduction
Overhead throw is a fundamental motor pattern which consist in a transfer of momentum through the pelvis and trunk to the throwing arm. 1 A proper progression of movements occurs in a proximal-to-distal order will enable higher velocities in throwing movements. 2 This technical-tactical ability is essential for athletes in multiple sports such as baseball, softball, handball, javelin throw, volleyball, and tennis among others.1,3 In this movement, the throwing velocity is a definite performance indicator, having a direct influence on the athlete's success or failure. 4
The influence of throwing velocity on these sports were analyzed elsewhere.5,6 Many studies have analyzed the influence of throwing velocity on these sports from different perspectives.7--9 Some of these studies had as objective to improve and maintain throwing velocity throughout a sporting event (short term)10,11 or a complete season (long term).12--14 The strength training methods used for the improvement of this capacity have been very diverse.12,15,16 In the reviewed literature, studies have used methodologies varying from more generalized work of strength, in which exercises such as bench press or squats are employed,17--19 to more specific methods that imitate the technical action in competition and whose aim is to immediately transfer the strength gain to the play.20,21 However, the problem appears when reviewing the existing literature because of not solely the methodological diversity for training these qualities, but also the necessity of examining other aspects of strength training to find out what kind of relationship they have with other research variables, such as the age of the athletes,22,23 differences between genders,24,25 or expertise in strength training.26,27 These variables are known to affect strength, power and the gains of strength after specific training.28,29 It is essential to understand when and under which circumstances they exist as well as what is the appropriate age to propose specific training for strength gain 30 and throwing velocity. Also, it is paramount to understand the expected improvement ratio for different athletes' populations given the same strength training will yield a different improvement percentage. 29 For example, while untrained players can improve throwing velocity using a generic workout with average weights this might not have the same effect when used with highly trained athletes.7,31
In recent years, the number of papers on this topic has increased.32--36 A certain consensus has been reached on the variables that affect the speed of the throw such as sex, age, sports experience or strength training among others.3,37,38 Several attempts have been made to analyze how these variables affect athletes' performance.3,32,33,38,39 Two meta-analysis32,38 and other systematic or narrative reviews had been published3,33,39 but there are some limitations on these studies which need to be taken into account in order to be able to state with certainty what effect each of these variables has on an athlete's performance. Some of these studies only analyzed a partial group of overhead sports such as baseball, softball, and tennis and did not include all the relevant scientific evidence on the subject. 38 Other studies did not take into account variables which brings heterogeneity to their analysis 32 or were too distant from the reality of current training.3,40,41
For these reasons, a meta-analysis is necessary. Although other meta-analyses regarding the effects of RT on throwing velocity have been published32,38 none have limited their analyses to measures of throwing velocity by gender, age, or all overhead throw sports. The purposes of this study were: (I) to assess the level of evidence for the effect of RT on throwing velocity in athletes; (II) to establish whether there are any differences in throwing velocity due to gender, and (III) to discern the effect of age on a specific training program in overhead athletes. Based on previous meta-analytic data, 25 we hypothesized that there would be a significant post-training improvement of throwing velocity in all populations but with a more considerable increase in adult athletes.
Methods
A systematic revision was carried out 10th December to search the published scientific evidence and understand how the different RT programs affect throwing velocity in sports where overhead throw is performed. The reporting flow diagram of this systematic review was based on the Preferred Reporting Items for Systematic reviews and Meta-Analyses (PRISMA) guidelines (Figure 1).

Flow of journal articles through the systematic review process.
Search strategy
Control group trials were identified through searching the five principal electronic databases: Web of Science, SCOPUS, SportDiscus, MedLine, and PubMed. Prior to the bibliographic search, a scoping search was carried out to identify possible keywords. In this search the first systematic reviews and meta-analyses on the subject were found. 3,19,32,38 Identifying both Medical Subject Headings (MeSH) terms and the following keywords: Resistance Training, Strength Train*, Resistance Conditioning, Weight* Train*, Athletic Performance, Throw* Performance, Throw* Velocity, Overhand Throw*, Overhead Throw*, Baseball, Volleyball, Racquet Sport, Tennis, Softball, Handball, Cricket, Waterpolo, and Football. These search terms were combined with two Boolean operators AND; OR. Also, the bibliographies of other previous related reviews and the studies finally selected were examined to search for new studies. Other possible scientific evidence related to the subject was identified by contacting authors of the published articles through email.
Two independent reviewers examined the title/summary of the articles found in the databases. After the initial selection, they analyzed each study with the inclusion criteria. Each criterion was evaluated as yes/no. If discrepancies existed between the authors, the ratings of the articles were shared and discussed until a consensus was reached. The authors were familiar with the existing literature and did not have a different bias with any of the studies selected for inclusion in the review.
Inclusion criteria
The eligibility of each investigation was measured according to the following inclusion criteria: a) preferred if control groups were included within research design, with subjects randomly assigned to groups; b) athletes must have had more than 2 years of experience in sport; c) athletes must not have experienced injuries either 6 months before the study or during the intervention; d) RT program supervised with the RT intervention of similar order and standardized periods; e) minimum duration of 4 weeks in the training intervention; f) throwing velocity assessment implemented within a specific sport, and g) sufficient data to calculate effect sizes (ESs). The articles that met the inclusion criteria were identified, and their full-text versions were obtained. Randomized controlled trials (RCTs) and randomly assigned trials (RANs) using stratified allocation were evaluated for this analysis.
Articles with one or more of these criteria were excluded: a) legal or illegal ergogenic aids or supplementations were used during interventions; b) cohort or narrative reviews; and c) mixed-gender studies.
Data collection process
All calculations were conducted using Microsoft Excel (Microsoft, Redmond, WA, USA) spreadsheet containing data extracted from each publication. For studies that only reported their data in charts, WebPlotDigitizer version 4.4 was used to obtain the throwing velocity values. Review Manager (RevMan) version 5.3.5 was used for all statistical analyses’ forest plots. The Cochran Q statistic 42 was used to assess heterogeneity between studies. Heterogeneity is a measure of the differences in main effects between studies. Also, I 2 statistics were used to assess heterogeneity (I 2 > 50%) to indicate heterogeneity.
RT program effects on throwing velocity were calculated for each included study following coding of pre-to-post changes and standard deviations (SDs) of both groups. The standardized mean difference (SMD) was calculated by subtracting baseline and post-intervention values of throwing velocity measures. Data were required to take these forms: a) mean and SDs (pre- and post-intervention); b) 95% confidence interval (CI) data for pre- to post-intervention change for each group; or when this was unavailable, c) actual p values for pre- to post-intervention change for each group; or if only the level of statistical significance was available, d) default p values (e.g., p B 0.05 becomes p B 0.49, p B 0.01 becomes p B 0.0099, and p B not significant becomes p B 0.05). A random-effects inverse variance (IV) was used with the measure of the effect of SMD.
The analysis of ES was conducted with a random-effects model estimated using the DerSimonian and Laird method. 43 A random-effects model is incorporated when the assumption is that the effect across studies is randomly situated about a central value. Forest plots were generated to demonstrate the study-specific pre- to post-training velocity differences and ESs, within the respective 95% CIs. Combining estimates then allowed for assessment of a pooled effect, in which the reciprocal sums of two variances were accounted for, including the estimated variance associated with the study and the estimated component of variance due to the variation between studies. A sensitivity analysis was conducted to identify the presence of highly influential studies that might have biased the analysis.
The study-specific weight was derived as the inverse of the square of the respective standard errors. The ESs of ≤0.2, ≤0.5, ≤0.8, and ≥0.8 were considered trivial, small, moderate, and large, respectively. 44
Results
The flow diagram of this article search and selection is depicted in Figure 1, from “potentially relevant” to “finally included”.
Study selection
The preliminary search yielded 3.265 relevant abstracts and citations. The full text of 63 articles were deemed to meet inclusion criteria. From these 47 studies were rejected for this meta-analysis due to the reasons that can be seen in figure 1. Finally, a total of 16 studies with 424 subjects met inclusion criteria. 17,21,24--26,45,48,52,54,56--61 The experimental design of included studies was either a RAN or RCT. Of the 16 studies, gender and age representations were male (n = 10); female (n = 5) and mixed (n = 1), and teenagers (n = 3) and adults (n = 13).
Study characteristics
The age range of the samples varied widely between studies. Where 3 investigations used a sample composed of teenagers (14–18 years old), 13 studies used adults (≥18 years old). Contrarily, the samples of athletes included experienced players in all studies. However, the experience of sample’s populations with strength training was specified in 10 out of 16 studies. Aside from these, in only 6 investigations the sample had previous experience in strength training, while in the rest, the sample had minimal or no experience.
The duration of the programs ranged between 4 and 18 weeks, being the most frequent program duration 6–8 weeks with 2–3 sessions per week. Regarding the training sessions, the duration of each session was only specified in 12 articles; in these, it ranged from 20 to 120 minutes. Finally, 9 studies were carried out in a period of competition or a rest period (off-season), while this timing was not specified in the other cases.
Sensitivity analysis
Galbraith plots were used to identify any potential outliers. The examination of Galbraith plot did not reveal any influential study (Figure 2). The shape of the funnel plot did not reveal any evidence of apparent asymmetry (Figure 3).

Galbraith plot of all studies included.

Funnel plot of all studies included.
Assessment of bias
Authors did not detect any publication’ bias or heterogeneity in this meta-analysis. A funnel plot reveals most data points within the plot are within the funnel, indicating bias and between-study heterogeneity do not exist (Figure 2). If bias did exist, the data points would be congregated outside of the reverse funnel, denoting asymmetry, and bias by unpublished or inaccessible studies.
Effects of a resistance training program on throwing velocity
Between groups strength differences were assessed via meta-analysis for all included studies. Male and female subjects were combined into different subgroup analysis, and teenagers and adult subjects were combined into a separate subgroup. To conclude, subjects were separated into different subgroups by gender and age. Due to potential heterogeneity, a random-effects model was incorporated with I 2 used to assess throwing velocity measures. Forest plot of figure 4 shows the obtained results of this parameter. This forest plot contains the SMD and corresponding CIs for throwing velocity gain, as well as, the overall effect test and heterogeneity analysis. The pooled mean ES estimated of throwing velocity comprised 20 treatment groups from 16 studies.17,21,24–26,45,48,52,54,56–61 (Table 1). There was a significant heterogeneity detected in all the 12 studies included in the meta-analysis (I 2 = 78%). When a random-effects analysis was applied a large effect was observed for throwing velocity outcomes (ES 1.10; 95% CI 0.64–1.57; p < 0.00001).

Forest plot of all studies included. The vertical line indicates the overall estimate of combined throwing velocity studies standardized mean effect size. The horizontal line indicates 95% CI, squares indicate estimates, square size is proportional to sample size, and rhombus indicates meta-analytically pooled estimates' 95% CI. IV = inverse variance.
Study and subject characteristics.
EG: experimental group; CG: Control group; M: male; F: female; - : Not Stated.
Effects of a resistance training program on throwing velocity due to gender
Outcomes for throwing velocity due to gender are shown in the forest plot in Figure 5. The forest plot contains the SMD and corresponding CIs for throwing velocity gain as well as the overall effect test and heterogeneity analysis in total and separated by gender. The pooled mean ES estimated of throwing velocity (Table 1) comprised 14 male treatment-groups from 10 studies17,24,26,45,48,54,56,57,60,61 and 5 female treatment-groups from 5 studies.21,25,27,52,58 There was significant heterogeneity detected in male subgroups (I 2 = 78%), also significant heterogeneity was found for women subgroup (I 2 = 77%). When a random-effects analysis was applied, a large effect was observed in both subgroups for throwing velocity outcomes. For males (ES 1.12; 95% CI 0.56–1.78; p < 0.0001), and almost double ES was observed in women for throwing velocity outcome (ES 1.22; 95% CI 0.25–2.20; p < 0.00001).

Forest plot of population separated by gender. The vertical line indicates the overall estimate of combined throwing velocity studies standardized mean effect size. The horizontal line indicates 95% CI, squares indicate estimates, square size is proportional to sample size, and rhombus indicates meta-analytically pooled estimates' 95% CI. IV = inverse variance.
Effects of a resistance training program on throwing velocity due to age
Outcomes for throwing velocity due to age are shown in the forest plot in Figure 6. The forest plot contains the SMD and corresponding CIs for throwing velocity gain as well as total and by age overall effect test and heterogeneity analysis (Figure 6). The pooled mean ES estimates of throwing velocity (Table 1) comprised 5 teenager treatment groups from 3 studies26,48,54 and 15 adults treatment groups from 13 studies.17,21,24,25,27,45,52,56–58,60,61 There was moderate heterogeneity detected for teenagers (I 2 = 59%) and significant heterogeneity for adults (I 2 = 78%) in the meta-analysis. When a random-effects analysis was applied, a small effect was observed in teenager subgroup (ES 0.49; 95% CI -0.18 - 1.17; p = 0.04), and a larger effect was observed in adult subjects for throwing velocity outcomes (ES 1.34; 95% CI 0.64 - 1.92; p < 0.00001).

Forest plot of population separated by age. The vertical line indicates the overall estimate of combined throwing velocity studies standardized mean effect size. The horizontal line indicates 95% CI, squares indicate estimates, square size is proportional to sample size, and rhombus indicates meta-analytically pooled estimates' 95% CI. IV = inverse variance.
Effects of a resistance training program on throwing velocity due to gender and age
Outcomes for throwing velocity due to age and gender are shown in the forest plot in Figure 7. The forest plot contains the SMD and corresponding CIs for throwing velocity gain, as well as, the overall effect test and heterogeneity analysis in total and separated by gender and age. The pooled mean ES estimates of throwing velocity (Table 1) comprised 5 male teenager treatment groups from 3 studies,26,48,54 9 male adults treatment groups from 7 studies, 17,24,45,56,57,60,61 and 5 female adult treatment groups from 5 studies. 21,25,27,52,58 There was moderate heterogeneity detected in male underage subjects (I 2 = 59%), significant heterogeneity for male adults (I 2 = 84%), and significant heterogeneity for female adults (I 2 = 77%). When a random-effects analysis was applied, a small effect was observed in male teenagers subjects for throwing velocity outcomes (ES 0.49; 95% CI -0.18–1.17; p = 0.04), a large effect was observed in male adults subjects for throwing velocity outcomes (ES 1.59; 95% CI 0.74–2.44; p = 0.001), and a large effect was observed in female adults subjects for throwing velocity outcomes (ES 1.22; 95% CI 0.25–2.20; p < 0.00001).

Forest plot of population separated by gender and age. The vertical line indicates the overall estimate of combined throwing velocity studies standardized mean effect size. The horizontal line indicates 95% CI, squares indicate estimates, square size is proportional to sample size, and rhombus indicates meta-analytically pooled estimates' 95% CI. IV = inverse variance.
Discussion
The purposes of this study were: (I) to assess the level of evidence for the effect of RT on throwing velocity in athletes; (II) to establish whether there are any differences in throwing velocity due to gender, and (III) to discern the effect of age on a specific training program in overhead athletes.
Throwing velocity is an important research topic in sport sciences due to the amount of published studies in the last ten years and because it is a performance variable for an overhead athlete that could predict success or failure in overhead sports. 32,38 Our findings suggest that specific RT for the enhancement of throwing velocity has a significant effect on all populations (ES 1.10; 95% CI 0.64–1.57; p < 0.00001). These results agree with previous investigations, which have correlated strength gains with throwing performance. 62,63
Adult men were the most studied population with 182 subjects in 8 studies with a large ES (ES 1.59; 95% CI 0.74–2.44; p = 0.001) between the control group and experimental group. Adult women only had 5 studies included with 93 subjects in this meta-analysis. While this population had large ES of all populations (ES 2.03; 95% CI 1.38–2.68; p < 0.00001), not enough studies were found and included to ensure these results firmly. More research is needed to fulfill that gap of periodized resistance training in women. While men threw faster in absolute terms, it may be observed that women could accomplish the same improvements than men. These results show that gender is less relevant variable than age for performance improvements after resistance training. Despite all this evidence, women population is underrepresented in this research field. In this particular case, adult women represented only 22% of all subjects included, and no data of teenager women were included because of the lower volume of studies found in bibliography and their non-compliance of inclusion criteria. Women's resistance training is being highlighted as a topic for further study.64,65 This meta-analysis is in line with giving the necessary importance to strength training in women and to fill this gap in the topic.
Adult population showed a great enhancement in throwing performance (ES 1.34; 95% CI 0.64–1.92; p = 0.001). When results were examined separately by gender, both genders still evidenced large effects in their improvements (ES 1.59; 95% CI 0.74–2.44; p = 0.001 for adult men and (ES 1.22; 95% CI 0.25–2.20; p < 0.001) for adult women. Even though both effect sizes were large and significant, it is clear that men yield greater improvements than women. Women usually perform lower body workouts to prevent a higher incidence of injuries, 50 so when they perform whole-body RT, they obtain equal improvements in all their outcomes. The lower improvements in women may be due to the heterogeneity of RT method design, with no consensus on the best intensity or adequate volume for the improvement of the throwing velocity. Nevertheless, studies with higher intensities obtained a greater percentage of improvement. 45 This lack of consensus may be due to the different levels of resistance training experience in the included studies. These did not specify the resistance training experience and population could range from elite players level with improvement ceiling 49 or professional players with high trainability. 45
Adults seem to be the best population for a specific resistance program for throwing velocity (ES 1.34; 95% CI 0.64--1.92; p < 0.00001) when it comes to age. This population is widely studied in literature, with 322 subjects in 13 studies included. Therefore, these results on this point could be used as reference values and expected results in future research. However, the teenage population only had 84 male subjects in 3 studies included mostly due to our inclusion criteria to provide previous sport or resistance training experience. Teenagers had a moderate ES (ES 0.49; 95% CI -0.18–1.17; p = 0.04) with a more widely CI possibly because of the maturation stage of subjects or resistance training methodologies. Teenagers female athletes have not been included, and this is something which must be taken into account by researchers to fulfill the gap in future studies on topic or address recommendations to that population. This investigation encourages to erase stereotypes related to strength and women or youth, both aesthetically and physiologically. It can also highlight the need to increase strength training in overhead female athletes not only in adulthood but also during adolescence as other studies have appointed before. 67,68
As mentioned before, age could be a more decisive variable than gender when it comes to systematic strength training. Improvements in throwing velocity can have significant differences due to that factor. 30,69 Another variable to consider is previous resistance training experience, which is not specified in most of the included studies for neither male nor female athletes. Future research should investigate the effect of previous resistance training experience and maturation age of the overhead, throwing athletes on throwing velocity gains after resistance training. Thus, coaches could adjust in a better way physical training in their teams, depending on the target population. A better and systematized perspective is necessary to improve the performance and prevention of injuries associated with overhead sports. 3,19,38
Researchers have overcomplicated their RT designs and have inadvertently produced data that provide unreliable and confusing guidance regarding intensity or rest between exercises.49,70–72 The use of different training methods has resulted in a moderately large body of evidence that cannot fully answers the question at hand individually or collectively.7,32,73
Nevertheless, we can make some general recommendations to be taken into consideration such as the equipment used for resistance training is not as important as the fact of implementing these sessions in the daily life of overhead athletes. 74–76 Free weights are the implements with the most scientific support and seem to achieve the greatest improvement in throwing velocity.57,60 Elastic bands and/or medicine balls, are also an effective method to improve this variable, being the own launch, or exercises with self-weight those that less improvements have been able to obtain in the analyzed studies.22,55,70 We cannot be sure if this is due to the equipment itself, or to the individualization of the load (≥60 -- 70% RM) that has been done in a more exhaustive way in the first studies.24,77
For a complete analysis of this research, it is necessary to consider some limitations present in this paper. Only studies published in English or Spanish were considered for inclusion. No search was made in grey literature as could be theses or conference abstracts. Finally, due to strict inclusion criteria such as 5 years of experience in sport, it was not possible to obtain robust results with respect to certain populations as adolescent athletes. In future systematic reviews and meta-analyses on the issue without this type of inclusion criteria, it would be possible to include a greater evidence.73,77–80
Guidelines on the appropriate number of sets, load, and durations per resistance program required to establish the specific data on throw velocity gains as a divisive issue for researchers, clinicians, and fitness trainers are not stablished. 3,19,38 Within each population, some have the false belief that throwing is best for creating a better throw performance. The current literature does little to settle the issues of when and how the optimal way to achieve throwing velocity gains. The current meta-analysis presents additional evidence regarding a relationship between gender or age and throwing velocity, concluding that specific RT improves throwing velocity in all athlete populations. Women achieve greater throwing velocity enhancement than men, and the adult population more than underage, following RT.
Conclusions
In recent years, overhead athletes' throwing velocity has been widely studied. There have been several attempts to synthesize the variables that affect this performance factor and how to improve it5,32,39 but no attempt has been made to evaluate the influence of age and gender. After a systematic search and subsequent statistical analysis of the studies included, this research encourages the use of a specific training program to increase throwing velocity. RT should last four weeks at least, with a 2–3-hour session each week with any implement at one's disposal e.g. free weights, elastic bands, or medicine ball. For young male subjects, the use of strength training may be an appropriate evaluation of maturation development.30,55 For male adult subjects, specific RT may show benefits from additional time and training volume to reap the smaller increases in performance generally seen at this level of training progression.
Athletes of any condition should include RT to improve athletic performance in overhead sports. Coaches, clinicians, and strength and conditioning professionals who use one or both above training protocols should see not only muscular improvements but functional performance improvements too. This meta-analysis suggests that an individualized and progressive methodology for both the upper and the lower extremities according to maturation and level of training should be done in terms of both volume and intensity for each player. Future studies in this area should focus on improving the methodology and controlling the intensity in all RT methods that are depicted in detail to achieve the best results. However, the generalization of our findings is limited to the specific protocol applied in the study. When the population does not have as much experience in the practice of the specific sport it is possible that other methods could be required for the improvement of the throwing velocity. Another limitation of the current study is that only studies that randomized a sample with more than 5 years of specific sport experience were included, so a small group of studies has been eligible for the final analyses.
Footnotes
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) received no financial support for the research, authorship, and/or publication of this article.
