Acute effects of prior conditioning activity on change of direction performance. A systematic review and meta-analysis

ABSTRACT We performed a systematic review and meta-analysis on the acute effects of prior conditioning activity (CA) on change of direction (COD) performance. Eligible studies, involving healthy participants undergoing acute CA with at least one measure of COD performance, were analysed across diverse databases. A total of 34 studies were included for systematic review with 19 studies included for the meta-analysis. The intervention condition resulted in significantly faster (Z = 4.39; standard mean difference [SMD] = 0.49; p < 0.05) COD performance compared with the control condition. Both unloaded and light loaded CA resulted in significantly greater (SMD = 0.58–0.59) COD performance compared to the control condition. Moreover, heavy loaded CA demonstrated a significant but small (SMD = 0.24) improvement in COD performance compared to the control condition. Age and study design had no effect on the overall meta-analysis outcomes. Both males and females exhibited similar moderate effects with CA but only males demonstrated significantly greater COD performance compared to control conditions. Our findings indicate that a range of CA protocols can acutely improve COD performance with unloaded and light-loaded CA resulting in the greatest performance enhancements. These findings will assist practitioners with the design and implementation of appropriate acute CA to improve COD performance.


Introduction
The ability to move quickly in a new direction during highintensity movements is commonly known as change of direction (COD) performance (Young et al., 2021).The COD movement is either assessed within a pre-planned condition, where athletes are instructed to run a prearranged direction or movement pattern (Brughelli et al., 2008;Young et al., 2015), or an unplanned condition -also known as agility, where athletes are required to respond to a stimulus provided during the run or movement (Young et al., 2021).An individual's COD ability is considered important for successful performance in team sports and can impart a physical and tactical advantage over an opponent (Nygaard Falch et al., 2019).Considering the importance of COD ability for sporting success, chronic training regimes such as strength, plyometric and sprint activities have consistently demonstrated long-standing improvement in COD performance (Nygaard Falch et al., 2019;Paul et al., 2019;Zouhal et al., 2019).However, acute bouts involving a priming activity have also been adopted with the aim to acutely enhance COD performance, commonly referred to as post-activation potentiation (PAP) (Boullosa et al., 2020).While PAP has been associated with improved sporting performances (Blazevich & Babault, 2019) the lack of evoked twitch verification, timeline and other potential contributors (e.g., temperature, water content) to explain improvements in subsequent sporting performances has led to the development of an alternative term for this response, postactivation performance enhancement (PAPE) (Blazevich & Babault, 2019).Thus, the term PAPE is more appropriate when considering the priming effects of acute conditioning activities (CA) on physical performance (e.g., COD) in team sports.
Many acute CA, such as plyometric and resistance exercises, have been reported to induce a short-term PAPE response for COD performance (Al Kitani et al., 2021;Dello Iacono et al., 2016;Maloney et al., 2014;Sener et al., 2021;Thapa et al., 2020).For instance, Thapa et al. (2020) reported a significant improvement in COD (Modified T-test) performance time following drop jumps as a CA compared to the condition without CA.In contrast, heavy-loaded resistance exercises have resulted in inconsistent changes for COD performance (Okuno et al., 2013;Orjalo et al., 2020;Sole et al., 2013;Zagatto et al., 2022).For example, Okuno et al. (2013) reported an improvement in COD (repeated sprint with COD) performance time following half-squat exercises performed at 50-90% of one repetition maximum (1RM) compared to a control setting in elite handball players.Conversely, Orjalo et al. (2020) reported no changes in COD (5-0-5) performance time following heavy-loaded barbell hip thrusts (85% 1RM) compared to a control condition in college-aged males and females.These inconsistent results highlight substantial variability of a PAPE response following a priming CA.The variability may be associated with several modulating factors including, but not limited to age, participant training experience, sex, performance test, and importantly the CA (Blazevich & Babault, 2019;Seitz & Haff, 2016).Further, the majority of these studies focussed on pre-planned COD tasks (Al Kitani et al., 2021;Dello Iacono et al., 2016;Maloney, Turner, & Miller, 201) whereas athletes engage in unplanned tasks during competition (Nimphius et al., 2018;Young et al., 2021) requiring further examination of a greater range of COD activities for application to real-world performances.
Given the lack of clarity for acute CA to induce a PAPE response for COD performance, a greater understanding of these acute CA is needed to help develop evidence-based recommendations for practitioners to improve COD performance.A narrative review synthesised prior work on the acute effects of strength-based CA noting that back squat exercises performed at 50-90% 1RM were able to improve COD performance (Lockie et al., 2018).Nonetheless, this review did not incorporate other established acute CA (e.g., plyometric, etc.) that have enhanced COD performance, did not assess unplanned COD tasks, and did not consider other modulating factors that impact COD performance.To our knowledge, there has been no systematic review and meta-analysis conducted to provide a comprehensive summary of the impact of CA on COD performance that considers modulating factors.Therefore, the primary aim of this study was to critically evaluate, via a systematic review, the effects of established acute CA on COD performance.A secondary aim was to identify the unique effects of these established acute CA on COD performance via a meta-analysis.Outcomes from this meta-analysis will help practitioners with the design and implementation of the most appropriate acute CA to enhance COD performance.We hypothesised that prior CA can acutely enhance COD performance.

Methods
The current systematic review was conducted as per the guidelines of PRISMA (Page et al., 2021).

Inclusion and exclusion criteria
Only full-text, peer-reviewed, original studies were examined for the current systematic review.Studies were considered eligible and included if they met the following criteria: 1) Population consisted of healthy participants with no restrictions on fitness and playing levels, sex or age; 2) Intervention involved voluntary muscle activity or any acute CA commonly associated with a PAPE response (Boullosa et al., 2020); and 3) Outcome included at least one measure of COD performance.
For the meta-analysis, COD performance was compared between intervention conditions that included a CA commonly associated with a PAPE response (e.g., heavy squats, plyometric exercises, etc.) to control conditions without a CA commonly associated with a PAPE response (e.g., dynamic/static stretching, typical standardised sport-specific warm-ups, or rest).If a study consisted of more than one intervention condition, then the intervention group for comparison with the control group was chosen based on the authors' hypothesis.For example, if authors of a study hypothesised that horizontal drop jumps (HDJ) would induce greater improvement in COD performance than vertical drop jumps (VDJ), when compared to a control condition, then the comparison between the intervention condition with HDJ and the control condition was used for the meta-analysis.If no clear hypothesis was provided, and the intervention groups were the same but with different loading conditions, then the group with the greatest loading was chosen and compared with the control condition (e.g., if a study examined 5% and 10% of body mass as the loads for the intervention groups, then the 10% body mass group was chosen).
Studies were excluded if: 1) Participants were unhealthy (e. g., injuries, recent surgery) precluding participation in an exercise session; 2) Intervention did not involve a prior voluntary muscle activity or any CA commonly associated with a PAPE response (Boullosa et al., 2020) (e.g., typical standardised sportspecific warm-up, stretching, etc.); 3) No COD performance measure was recorded; and 4) They were published as books, book chapters, congress abstracts, review papers, case reports, special communications, letters to the editor, invited commentaries, errata, and of irresolute quality or unclear peer-review process (Grudniewicz et al., 2019).

Selection process
Two authors (US and KD) independently screened the titles, abstracts, and full-text versions of retrieved studies.During the search and review process, potential discrepancies between screening authors regarding inclusion and exclusion criteria were resolved through consensus with a third author (JC).From the selected articles to be included, reference lists were also examined to identify any additional relevant studies.

Data extraction, quality assessment and risk of bias
The information pertaining to participant characteristics (e.g., age, height, mass, sex, sport, and training/playing experience), and study characteristics (e.g., CA, outcome measure, rest interval, assessment time following CA, and performance results) were retrieved and collated into a Microsoft Excel file (Microsoft Corporation, Redmond, WA, USA).The means and standard deviation (SD) of the outcome variables for both intervention (CA intervention) and control (no CA intervention) groups were extracted from the included studies and collated into a Microsoft Excel file.For studies that reported standard errors, the SD were calculated by multiplying the standard error with the square root of the sample size.
The physiotherapy evidence database (PEDro) scale was used to assess the methodological quality of the included studies, which were rated from 0 (lowest quality) to 10 (highest quality) (Singh et al., 2022).The validity and reliability of the PEDro scale has been reported previously (de Morton, 2009;Maher et al., 2003;Yamato et al., 2017) including agreement with other scales (e.g., Cochrane risk of bias tool) (Moseley et al., 2019).The methodological quality of each study was rated using the following thresholds: poor (<4), fair (4-5), good (6-8), and excellent (9-10) (Singh et al., 2022).Item 1 of the PEDro scale was not considered in the overall study quality rating as it pertained to external validity.The methodological quality for each included study was assessed independently by two authors (US and KD), and any discrepancies between them were resolved via consensus with a third author (JC).Publication bias was investigated using Egger's test (Egger et al., 1997) and a funnel plot.The funnel plot was generated with Review Manager Software (RevMan, Version 5.4, Copenhagen: The Nordic Cochrane Centre, 2014) by plotting the standard mean difference (SMD) against standard error (Sterne et al., 2011).

Statistical analysis
RevMan (version 5.4) statistical software (2020; The Nordic Cochrane Centre, Copenhagen, Denmark) was used to conduct the meta-analysis that compared outcome measures between intervention and control conditions.The forest plot was generated as a random effects model to control for inter-study heterogeneity.The heterogeneity of each study was determined using the I 2 statistic with values of <25%, 25-75%, and >75% representing low, moderate, and high levels of heterogeneity, respectively (Higgins and Thompson, 2002).The SMD and its associated standard error for each study was quantified using an equation appropriate for studies with a repeated measures design (Uanhoro, 2017).Therefore, magnitude of effect (effect size, ES) between conditions was reported as SMD and 95% confidence intervals with mean SMD values of 0.2, 0.5 and 0.8 classified as small, moderate and large ES, respectively (Cohen, 1988).
The typical timeframe in which CA enhanced anaerobic performance was reported to range from 3 to 10 minutes (Blazevich & Babault, 2019;Boullosa et al., 2020;Vargas-Molina et al., 2021).Therefore, when studies reported performance at multiple time points after the CA, the time points that were within the 3-10-minute range were chosen for the meta-analysis.For example, if a study reported COD performance at 4, 8 and 12 minutes, then the performance measures reported at 4 minutes and 8 minutes were averaged for the meta-analysis as these time-points were within the 3-10-minute range.This conservative approach was followed to eliminate any possible bias that may be caused by choosing one-time point.Furthermore, modulating factor analyses (i.e., subgroup and sensitivity) were conducted to investigate the effects of CA strenuousness (i.e., unloaded, light-loaded and moderate-heavy-loaded), performance test, sex, age, study design, training experience and bias.For age, comparisons were made between studies where the average age of participants was ≥18 years old (adults) and <18 years old (youth).The lightloaded CA was defined as a load of 10-30% of body mass while moderate-heavy loaded CA represented loads of 50-90% of 1RM or 75% of body mass (Liguori, 2020).The level of statistical significance was set at p < 0.05 for all analyses.

Certainty of evidence
The certainty of evidence for each outcome was evaluated by two authors (US and KD) using the Grading of Recommendations Assessment, Development and Evaluation (GRADE) protocol with ratings ranging from very low to high levels of certainty (Guyatt et al., 2011;Zhang et al., 2019Zhang et al., , 2019)).Initially, the evidence was considered as a high level of certainty, but was downgraded based on the following criteria: (1) Risk of bias in studies: downgraded by one level if the average PEDro scores were moderate (<6) or by two levels if PEDro scores were poor (<4); (2) Indirectness: downgraded by one level if there was indirectness (e.g., use of inconsistent populations, interventions, comparators and outcomes) or by two levels if there was indirectness from various sources; (3) Risk of publication bias: downgraded by one level if publication bias was suspected (Egger's test); (4) Inconsistency: downgraded by one level when the inter-study heterogeneity (I 2 ) was high (>75%); and (5) Imprecision: downgraded by one level when <800 participants were available for a comparison (G.H. Guyatt et al., 2011) or if there was no obvious direction of the effects (SMD <0.2; p > 0.05).If both imprecision criteria were observed, then certainty was downgraded by two levels.Finally, the certainty of evidence for each outcome measure was evaluated as very low when the outcome was not included in the meta-analysis (Ramirez-Campillo et al., 2022).

Study selection
A total of 6247 articles were identified from the database searches (Figure 1).Following the removal of duplicates and the elimination of articles based on title and abstract screening, 55 studies were evaluated that resulted in 34 studies satisfying the inclusion criteria.From the 34 studies, 19 studies were included for the meta-analysis (Figure 1).

Methodological appraisal of the included studies
Overall, the included studies were of fair (n = 12) to good (n = 22) quality with a mean PEDro score of 5.6 (Table 1).
None of the included studies scored more than 7 points out of 10.Most of the studies addressed the following PEDro items: eligibility criteria, random allocation of subjects, outcome measures were reported from more than 85% of the participants, subjects received the desired treatment as per the allocation, between condition statistical comparison and reporting of point measure and/or measure of variability (either SD, standard error or confidence interval).The least A detailed explanation for each PEDro scale item can be accessed at https://www.pedro.org.au/english/downloads/pedro-scale.NA -not assessed; Note: Studies with * symbol were included for meta-analysis.
From a possible maximal score of 10.
reported PEDro items included the following: allocation of concealment, and blinding of therapist, subjects and assessor.

Study characteristics
All of the included studies utilised a crossover study design with a total of 584 participants (male n = 433; female n = 115; unidentifiable for sex n = 36, Table 2).For the studies that were included in the meta-analysis, there were 323 (male n = 209; female n = 94; unidentified sex n = 20) participants (Table 2).The mean ± SD for age, height and body mass for all participants within the included studies were 20.5 ± 2.2 years, 1.78 ± 0.29 m and 73.9 ± 9.3 kg, respectively.Furthermore, six studies included participants with a mean age <18 years old while 28 studies included participants with a mean age >18 years old (Table 2).For the meta-analysis, four studies (Al Kitani et al., 2021;Ouergui et al., 2022;Rumeau et al., 2023;Zagatto et al., 2022) included participants with a mean age <18 years old and 15 studies included participants with a mean age >18 years old (Table 2).The COD protocols consisted of either pre-planned movements (i.e., athletes move in accordance with a predetermined movement pattern) or unplanned movements (i.e., agility) (Young et al., 2021).From the included 34 studies, two studies (Cochrane, 2013;Zois et al., 2011) reported outcomes from unplanned COD protocols with the remainder (32 studies) examining performance from pre-planned COD protocols (Suplementary Table S1).

Qualitative analysis
Significant improvements in COD performance were reported following plyometric exercises comprising of drop jumps, and ankle and hurdle hops (Mh et al., 2021).Both HDJ and VDJ significantly improved COD performance at 8 minutes post the CA with HDJ producing greater improvements than VDJ (Al Kitani et al., 2021;Dello Iacono et al., 2016).Likewise, alternate leg bounding performed on a hard or natural grass surface significantly improved COD performance compared to the control condition at 8 minutes following the CA (Dann et al., 2023).
Drop jumps performed at different heights significantly improved COD performance compared to the condition without drop jumps with the fastest COD performance occurring at 5 minutes post-CA (Sener et al., 2021).Similarly, a box jump followed immediately by a drop jump significantly improved COD performance compared to the control condition at 3 minutes post the CA (Thapa et al., 2020).
Apart from plyometric exercises, eccentric (Beato et al., 2019(Beato et al., , 2021)), isometric (Marshall et al., 2019;Pojskić et al., 2015) and dynamic (Coutinho et al., 2022;Escobar Hincapié et al., 2021;Ishak et al., 2022;Moreno-Pérez et al., 2021;Okuno et al., 2013;Orjalo et al., 2020;Rumeau et al., 2023;Sole et al., 2013) lower body strength exercises were used to improve COD performance.For example, eccentric overload exercises (squats), which included performing eccentric contractions on a flywheel, resulted in an improvement in COD performance at 3-6 minutes following the CA (Beato et al., 2019(Beato et al., , 2021)).Isometric squats significantly improved COD performance compared to the control condition at 5 minutes after the CA (Pojskić et al., 2015).Conversely, maximal isometric squats resulted in slower COD performance at 1 minute following the CA, with no significant differences occurring at 3, 5, or 7 minutes post-CA (Marshall et al., 2019).In regard to dynamic lower body exercises (e.g., squats, leg press, barbell hip thrust), significant improvements (5 minutes post-CA) in COD performance were noted after unilateral (large ES) and bilateral (moderate ES) squats (Escobar Hincapié et al., 2021).Ishak et al. (2022) assessed COD performance after 85% 1RM back squats and reported that COD performance was significantly quicker at 12 minutes after the CA compared to earlier recovery times (i.e., 4 and 8 minutes).Likewise, Rumeau et al. (Rumeau et al., 2023) reported improvements in COD performance at 7 minutes following a set of leg presses at 90% 1RM compared to the control condition.Half-squat (Okuno et al., 2013) and full-squat (Sole et al., 2013) exercises performed at 50-90% 1RM also significantly improved COD performance within 4-12 minutes following the CA compared to the control condition.In contrast, Coutinho et al. (2022) reported no improvement in COD performance (2 minutes post-CA) following regular half-squat movements at maximal concentric velocity.Likewise, a barbell hip-thrust exercise failed to improve the COD performance beyond that of the control condition (Orjalo et al., 2020).Only one study used variable resistance and reported an improvement in COD performance at 4 minutes after maximal (3-and 5-RM) squats with elastic bands (Peng et al., 2021) compared to the baseline.
Several studies have incorporated multiple CA in combination or isolation, noting variable outcomes in COD performance (Kalinowski et al., 2022;Toprak et al., 2022;Zagatto et al., 2022).For example, Toprak et al. (2022) incorporated squats at 85% 1RM and maximal isometric efforts with a knee angle of 120°-130°, and reported faster COD performance at 4 and 6 minutes, with peak performance occurring at 8 minutes post-CA compared to the control condition.Similarly, Alessandro M. Zagatto et al. (2022) incorporated two CA (drop jumps and heavy sled towing with 75% of body mass) and reported that drop jumps improved COD performance compared to the control (rest) condition however, heavy sled towing did not change the COD performance post-CA.Likewise, isometric half back squats      For each post-warm-up tested, recovery-times (i.e., 15-s, 4-min, and 8-min), both total, and peak times were faster following condition 1, 2 and 3 compared to the condition 4.There were no significant differences in-between recovery times in both total and peak times following condition 1, 2 and 3.
WBV 1-min warm-up exercise on a vibration platform at a frequency of 30 Hz and peak-to-peak displacement of 2 mm (120° knee angle) The participants' BAT performance improved significantly 1 min after the vibration stimulation (p < 0.01).
- followed by bilateral drop jumps resulted in no change in COD performance at 6 minutes following the CA (Kalinowski et al., 2022).
While most studies have employed plyometrics or traditional resistance exercises as the priming CA protocols prior to COD assessments, several studies have also employed loaded warm-ups (Chua et al., 2021;Maloney et al., 2014;Turki et al., 2020;Yeung et al., 2021) or whole-body vibration (WBV) training as part of the CA (Munshi et al., 2022;Wu et al., 2021;Yang et al., 2017).Faster COD performances were observed for up to 8 minutes following loaded warm-up (5-15% of body mass) compared to the unloaded condition (Maloney et al., 2014;Turki et al., 2020).However, Yeung et al. (Yeung et al., 2021) observed no significant differences between loaded (10% body mass) and unloaded conditions, although small effect sizes were noted suggesting faster COD performance following the loaded condition.For WBV, Wu et al. (Wu et al., 2021) reported improvements immediately after 1 minute of a WBV squat exercise performed on a vibration platform at a frequency of 30 Hz. Yang et al. (Yang et al., 2017) reported improvements (~3 minutes post-WBV exercise) in COD performance after 4 minutes of partial squats (60° knee angle) and dual frequency WBV (35−45 Hz applied individually to each leg) but showed no improvement after 4 minutes of partial squats (60° knee angle) and single-frequency WBV (35-45 Hz applied to both legs).A significant improvement in COD performance (at 4-and 12-minute intervals) was also reported following separate plyometric and WBV protocols (with unilateral and bilateral squats) with COD performance better for the plyometric condition (Munshi et al., 2022).
There were only two studies (Cochrane, 2013;Zois et al., 2011) that examined performance from unplanned COD as an outcome measure.Cochrane (2013) reported no change in unplanned COD performance after a CA of side-alternating vibration exercise during a static squat position.On the contrary, Zois et al. (2011) reported an improvement in unplanned COD performance at ~4 minutes following a CA of leg-press exercises (5RM).

Effect of CA on COD performance
Overall, there was a significant, but small, effect of CA protocols on COD performance with moderate, inter-study heterogeneity (Figure 2).The results from the Egger's test suggested absence of publication bias (p = 0.131).Through visual inspection of the funnel plot, three studies (Orjalo et al., 2020;Pojskić et al., 2015;Sener et al., 2021) were identified as outliers, falling outside the 95% CI (Figure 3).Removing these outliers from the metaanalysis yielded a similar small effect (SMD = 0.40, p < 0.05) of CA protocols on COD performance with elimination of the observed interstudy heterogeneity (I 2 = 0%; p = 0.49).

Modulating factors' analysis
There was a significant moderate effect of unloaded and lightloaded CA on COD performance with moderate-high, interstudy heterogeneity (Figure 4).There was a significant small effect of moderate-heavy loaded CA on COD performance with no interstudy heterogeneity (Figure 4).Further, when differentiated based on type of assessment (i.e., pre-planned and unplanned COD protocols), a slightly greater PAPE response was observed for pre-planned COD with only the pre-planned COD performance significantly enhanced, and high inter-study heterogeneity, compared to the control condition (Figure 5).When examining the effects of sex on COD performance following CA, both males and females exhibited similar moderate effects with only males exhibiting significantly greater COD performance compared to control conditions (Figure 6).In contrast, age had no effect on meta-analysis outcomes with exclusion of studies involving participants aged <18 years old (Al Kitani et al., 2021;Ouergui et al., 2022;Rumeau et al., 2023;Zagatto et al., 2022) still resulting in significantly greater COD performance following the CA compared to the control condition (SMD = 0.49; p < 0.05).
In regard to study design, the timing of post-CA assessment had minimal impact on the meta-analysis results.Specifically, exclusion of studies that reported COD performance at > 10 minutes after the CA (Yeung et al., 2021), immediately after the CA (Cochrane, 2013) and the best COD performance time at 4-12 minutes after the CA (Sole et al., 2013), resulted in a marginal increase in ES from small (SMD = 0.49; p < 0.05) to moderate (SMD = 0.54; p < 0.05)    With regards to training experience, three studies reported that participants had >2 years of training experience (Al Kitani et al., 2021;Maloney et al., 2014;Sener et al., 2021).Five studies reported that participants had ≥3 years of playing experience (Cochrane, 2013;Moreno-Pérez et al., 2021;Ouergui et al., 2022;Yeung et al., 2021;Zagatto et al., 2022).One study reported at least 1 year of plyometric training experience for participants (Dann et al., 2023).Three studies reported that participants had both ≥2 years playing experience and ≥1-year resistance training experience (Orjalo et al., 2020(Orjalo et al., , 2020;;Turki et al., 2020) while two studies reported that participants had at least 1 year of resistance training experience (Okuno et al., 2013;Sole et al., 2013).The degree of training experience for participants was not reported in the remaining five studies (Pojskić et al., 2015;Rumeau et al., 2023;Thapa et al., 2020;Zagatto et al., 2022;Zois et al., 2011).Due to the inconsistent and variable levels/types of training experience reported, a sensitivity or subgroup analysis could not be performed.

Certainty of evidence
The certainty of evidence for the COD outcomes was low based primarily on limited sample size for comparison (n < 800) and PEDro scores of <6.

Discussion
The primary aim of this systematic review and meta-analysis was to critically examine the effects of established CA on COD performance.Thirty-four fair-to-good quality studies, involving 584 participants, were identified with the qualitative and metaanalysis confirming that most CA can significantly improve COD performance compared to a control condition.From the modulating factor analyses, unloaded and light-loaded CA showed significant moderate effects, whereas moderate-heavy loaded CA demonstrated significant, but small effects on COD performance compared to the control condition.When examining sex, males exhibited slightly greater and significant COD performance improvements compared to females.Likewise, the improvements were slightly and significantly greater for preplanned COD tasks compared to unplanned COD tasks.In contrast, age and timing of CA had no substantial effect on COD performance.The findings from our study suggest unloaded CA (i.e., plyometric exercises) and light-loaded (10-30% body mass) warm-up can acutely enhance COD performance.These findings can assist practitioners in selecting appropriate CA protocols to enhance COD performance when considering a range of modulating factors.
Our meta-analysis identified that CA protocols, known to induce a PAPE response, significantly improved COD performance when compared to a control condition.These results extend upon findings from a previous narrative review (Lockie et al., 2018) that qualitatively reported the beneficial effects of strength-based CA protocols on COD performance.The current review included a variety of CA protocols including, but not limited to, plyometric exercises, strength-based (i.e., lightloaded and moderate-heavy loaded) and a range of "other" CA protocols.The modulating factors analyses characterised that unloaded (plyometric exercises) and light loaded CA exhibited significant moderate and greater effects on COD performance compared to moderate-heavy loaded CA.This result was in line with a previous meta-analysis where plyometric exercises produced greater PAPE effects (ES = 0.47) on a variety of performance measures (i.e., jump, sprint, throw and upperbody ballistic-bench press throw, etc.) compared to traditional high (≥85% 1RM) and moderate intensity (30-85% 1RM) resistance exercises (Seitz & Haff, 2016).Potentially, plyometric exercises and light-loaded CA improve stretch shortening cycle mechanics via greater muscle recruitment (Seitz & Haff, 2016) and leg stiffness (Maloney et al., 2014;Pearson & McMahon, 2012) compared to heavy loaded resistance exercises, resulting in greater performance.It has been purported that by utilising high velocity and power movements, there is greater preferential recruitment of Type II motor units, resulting in a greater PAPE response (Brink et al., 2022;Seitz & Haff, 2016).In contrast, the moderate-heavy loaded CA protocols may have resulted in more fatigue compared to plyometric and lightloaded CA, impairing the PAPE response.The lower PAPE response with moderate-heavy loading may also be influenced by participant training experience (Wilson et al., 2013).Previous studies reported that individuals with prior resistance training experience exhibited a considerably larger PAPE effect compared to those with no prior experience (Seitz & Haff, 2016;Wilson et al., 2013).In the current review, training experience was not clearly reported in the included studies and therefore it was difficult to evaluate the effect of training experience on COD performance within our meta-analysis.Future research may clarify the role, if any, of training experience for a COD PAPE response.Additionally, the current review provides guidance for coaches who may utilize plyometric exercises or light-loaded warm-ups as a CA to acutely enhance pre-planned COD performance.Employing plyometric and light-loaded approaches may be advantageous for athletes and coaches, given their minimal equipment demands and the ease with which they can be implemented on fields or during athletes' travels for competitive events.For instance, exercises such as vertical jumps, drop jumps, horizontal drop jumps, and similar activities can be effectively utilized by athletes for the acute enhancement of COD performance.Coaches/athlete's use of these exercises should also consider loading levels when implementing CA protocols for COD performance and the range of factors that influence PAPE highlighted in this and other studies.
Previously, sex and age were reported to be important factors that influence the magnitude of a PAPE effect (Arabatzi et al., 2014;Hancock et al., 2015).Superior PAPE effects were reported to be associated with greater recruitment of Type II muscle fibres (Fontanetti et al., 2023) with males exhibiting a greater proportion of Type II muscle fibres compared to females, and adults exhibiting a higher percentage of Type II muscle fibres (Arabatzi et al., 2014).In the current review, moderate-effect PAPE responses were observed for both males and females with slightly greater and significant PAPE effects for males only.This potential sex effect may reflect the difference in fibre-type composition (Fournier et al., 2022).However, there were limited studies on females in the current review, which might have biased the sex outcomes with further work needed to clearly identify the effects of sex on CA to enhance COD performance.Further, our modulating factor analysis identified similar SMD values when studies were excluded/included based on age (i.e., <18 and >18 years) with the disparity possibly due to the limited number of studies in youth within the current review and/or the influence of other modulating factors such as sex (Hancock et al., 2015).
With respect to unplanned COD tasks, our systematic review identified two studies (~6%) that examined the acute effects of CA on unplanned COD performance (Cochrane, 2013;Zois et al., 2011), a movement pattern that is common within team sport settings (Young et al., 2021).Specifically, heavy lower body resistance exercise (5RM leg press) enhanced unplanned COD performance (Zois et al., 2011).However, WBV exercise as a CA did not result in any acute benefits for unplanned COD performance (Cochrane, 2013), suggesting that heavy lower body resistance exercise is more beneficial to induce a PAPE response for unplanned COD tasks.These findings were in contrast to the overall findings in our systematic review for pre-planned COD performance, whereby unloaded exercises (i.e., plyometrics), or light-loaded exercises as a CA exhibited a greater PAPE response than moderate-heavy resistance exercises.In fact, none of the studies to date have examined unloaded, or light-loaded CA on unplanned COD performance as far as we are aware.Given that unplanned COD tasks are far more complex than pre-planned COD tasks, and of greater application to conditioning practitioners who work in team sport settings, more research is required to confirm of the effect of CA on physical and/or cognitive skills during unplanned COD tasks (Sheppard & Young, 2006).Nonetheless, practitioners could still consider implementing heavy loaded lower body resistance exercises as a CA to improve unplanned COD performance.
Our systematic review with meta-analysis makes a novel and significant addition to practitioners' existing knowledge of CA to enhance COD performance.Specifically, we collated the acute effects of various types of CA on COD performance via meta-analysis, extending prior work (Lockie et al., 2018).Based on our work, we recommend that coaches implement plyometrics or light-loaded warm-ups as a CA for preplanned COD tasks.Further, these findings highlight the need for further research on modulating factors such as sex, age, training experience, and type of assessment (i.e., preplanned and unplanned COD).Despite these critical findings, our systematic review has some limitations that should be considered.First, a sensitivity/subgroup analysis could not be performed based on training experience as this information was not clearly reported.Second, we could not identify the optimal time point for the best COD performance following the CA due to the limited number of studies with 3-10 minutes being the typical timeframe examined (Vargas-Molina et al., 2021).Third, the variation in the type of COD task and range of other factors such as population, age, CA, etc., might have contributed to the moderate-to-high interstudy heterogeneity.Lastly, the certainty of evidence was low due to low PEDro scores and insufficient number of participants.Future studies are encouraged to employ an a priori sample size power analysis (Abt et al., 2020) and recruit an appropriate number of participants to improve the overall quality of the study.

Conclusion
We found significant, but small effects of various CA protocols on COD performance, which support practitioner's use of CA protocols to enhance COD performance.Modulating factor analysis, considering CA difficulty, suggested that plyometric exercises or warm-up exercises with lighter loads (10-30% body mass) may be preferentially used by practitioners to induce significantly greater COD performance effects compared to moderate-heavy intensity exercises.Other modulating factors (e.g., age, sex, task, etc.) had variable effects on PAPE responses and COD performance and should be considered by coaches/practitioners when implementing pertinent CA protocols to acutely enhance COD performance.
Computerised literature searches were conducted across the electronic databases of PubMed, CINAHL, SportsDiscus, Web of Science and SCOPUS on 24.05.2022 and updated on 27.05.2023.The search strategy was conducted using (in different combinations) the Boolean operators AND/OR with the following MeSH terms (all database fields used): (Young adult[MeSH Terms] OR Adolescent[MeSH Terms] OR Athletes [MeSH Terms]) AND (Athletic Performance*[MeSH Terms] OR Athletic Performance/physiology*[MeSH Terms] OR Muscle strength[MeSH Terms] OR (Muscle Strength/physiology* [MeSH Terms]) AND (Muscle contraction*[MeSH Terms] OR Warm-Up Exercise*[MeSH Terms] OR Resistance Training* [MeSH Terms] OR Physical Education and Training/methods* [MeSH Terms] OR Plyometric Exercise*[MeSH Terms]) AND (Soccer*[MeSH Terms] OR Sports/physiology[MeSH Terms] OR Football/physiology*[MeSH Terms] OR (Basketball*/physiology[MeSH Terms] OR (Tennis/physiology*[MeSH Terms] OR Racquet Sports/physiology*[MeSH Terms]).Furthermore, the following keyword terms were also used: (postactivation potentiation OR post activation potentiation OR Warm-up) AND (COD OR Change of direction OR Agility OR unplan* OR plan* OR anticipat* OR unanticipat* OR side* OR cutt*).These keywords were truncated using the quotation marks.

Figure 1 .
Figure 1.Flow chart illustrating the study selection process.

Figure 2 .
Figure 2. Forest plot for COD performance.

Figure 3 .
Figure 3. Funnel plot of standard mean difference (SMD) vs. standard error (SE) denoting outliers for the COD outcome.

Figure 5 .
Figure 5. Forest plot on COD performance based on pre-planned and unplanned COD.

Figure 4 .
Figure 4. Forest plot on COD performance based on loading.

Figure 6 .
Figure 6.Forest plot on COD performance based on gender.

Table 1 .
Methodological quality of the included studies using the PEDro rating scale.

Table 2 .
Participant's characteristics of the included studies.

Table 3 .
Characteristics of the included studies.

Table 3 .
(Continued).Time period post: time period between the conditioning activity and performance assessment; CA: conditioning activity; RM: repetition maximum; rep: repetition; min: minute; COD: change of direction; RCOD: repeated sprint with change of direction; WBV: whole body vibration; Note: Studies with * symbol were included for metaanalysis.