In the abstract, the sentence beginning "CMJ showed a pooled improvement ... high heterogeneity (I² = 88.5%)" incorrectly reported the pooled CMJ effect estimate and heterogeneity value. This has been corrected to read: "CMJ showed a pooled improvement of 2.70 cm; however, this estimate should be interpreted with caution due to high heterogeneity (I² = 89%)."The original version of this article has been updated.In Section 3, subsection 3.1 Study Selection, paragraph 1, the paragraph beginning "There were 2,472 records identified ... 39 studies were considered for the review" contained incorrect counts for excluded full-text reports and included studies, and did not report the exclusion of studies with participants outside the eligible age range. A correction has been made to the section 3. Results, 3.1 Study Selection, paragraph 1: "A total of 2,472 records were identified through database searching. After removal of 1,449 duplicates, 1,023 records were screened, of which 909 were excluded. Full-text reports were sought for 114 records; 6 reports could not be retrieved, leaving 108 reports assessed for eligibility. Of these, 71 reports were excluded for the following reasons: lack of an eligible control group (n = 26), absence of prespecified outcomes (n = 16), insufficient data to calculate effect sizes (n = 19), inappropriate study design (n = 8), and participants outside the eligible age range (n = 2). Ultimately, 37 studies were included in the review (see Figure 1)."The original version of this article has been updated.In Section 3, subsection 3.2 Study Characteristics, paragraph 1, the paragraph beginning "A total of 39 such RCTs were included ... about 64% of studies" contained incorrect counts of included studies, subgroup distribution, study list, age range, and percentage of RT studies. A correction has been made to the section 3. Results, 3.2 Study Characteristics, paragraph 1: "A total of 37 such RCTs were included, consisting of 24 studies in RT settings and 13 in non-RT settings (Ahmun et al., 2005;Arciero et al., 2001;Becque et al., 2000;Bemben et al., 2001;Bonilla et al., 2021;Camic et al., 2014;Cribb et al., 2007;del Favero et al., 2012;Earnest et al., 1995;Griffen et al., 2015;Havenetidis Herda et al., 2009;Hoffman et al., 2006;Izquierdo et al., 2002;Javierre et al., 2004;Kaviani et al., 2019;Kelly Kilduff et al., 2002;Law et al., 2009;Mujika et al., 2000;Noonan et al., 1998;Nunes et al., 2017;Peeters et al., 1999;Saremi et al., 2010;Stone et al., 1999;Stout et al., 1999;SYROTUIK et al., 2000;Taylor et al., 2011;Trexler et al., 2016;van Loon et al., 2003;Volek et al., 1999;Volek et al., 2004;Wang et al., 2018;Wang et al., 2016;Wilder et al., 2002;Willoughby Zuniga et al., 2012). Not all outcomes were reported in some studies, but in every study, at least one prespecified outcome was noted. Mean ages were between 18.5 and 29.5 years, and training status ranged from sedentary or untrained to well-trained, competitive, and elite athletes. RT was the most common modality (about 65% of studies), with other protocols that included team sports (rugby or American football, soccer, handball, and basketball), individual sports (canoeing or rowing and sprinting), cycling-based protocols (including Wingate testing), and those related to strength-or power-related performance assessment." The original version of this article has been updated.In Section 3, subsection 3.3 Risk of Bias Results, paragraph 1, the paragraph beginning "Risk of bias was evaluated using the Cochrane RoB 2 tool ..." contained incorrect trial counts, percentages, and outcomemeasurement counts. A correction has been made to the section 3. Results, 3.3 Risk of Bias Results, paragraph 1: "Risk of bias was evaluated using the Cochrane RoB 2 tool. Among the 37 trials, 25 (67.6%) were deemed to be at low risk of bias, 9 (24.3%) raised some concerns, and 3 (8.1%) were considered to be at high risk. Some concerns most commonly arose from deviations from intended interventions (D2), with fewer studies flagged for missing outcome data (D3) or selective reporting (D5). Outcome measurement (D4) was regarded as low risk in nearly all trials (36/37), consistent with predominantly objective performance and body-composition outcomes (see Figure 2)."The original version of this article has been updated.In Section 3, subsection 3.4 Overall effects, the sentence beginning "For countermovement jump, the pooled effect favored ... I² = 88.5%" incorrectly reported the CMJ overall pooled estimate, study count, participant count, p value, and heterogeneity. A correction has been made to the section 3. Results, 3.4 Overall effects: "For countermovement jump, the pooled effect favored creatine supplementation (11 studies, N = 273; MD = 2.70 cm, 95% CI 0.18 to 5.21; p = 0.04; I² = 89%)."The original version of this article has been updated.In Section 3, subsection 3.4 Overall effects, the sentence beginning "Significant pooled effects favoring creatine supplementation ... LBM ... I² = 36.3%" incorrectly reported the LBM overall pooled estimate, study count, participant count, p value, and heterogeneity. A correction has been made to the section 3. Results, 3.4 Overall effects: "Significant pooled effects favoring creatine supplementation were also observed for FFM (15 studies, N = 327; MD = 2.32 kg, 95% CI 0.76 to 3.89; p = 0.004; I² = 13.6%) and LBM (16 studies, N = 344; MD = 1.84 kg, 95% CI 0.53 to 3.16; p = 0.006; I² = 37%)."The original version of this article has been updated.In Section 3, subsection 3.5 Subgroup analysis by training context (RT versus non-RT), the sentence beginning "For countermovement jump, the pooled effect was not significant ... p = 0.068" incorrectly reported the non-RT study count, participant count, pooled estimate, and between-subgroup statistics. A correction has been made to the section 3. Results, 3.5 Subgroup analysis by training context (RT versus non-RT): "For countermovement jump, the pooled effect was not significant in RT studies (5 studies, N = 88; MD = 4.19, 95% CI -1.28 to 9.66; p = 0.13; I² = 95%), whereas no significant pooled effect was observed in non-RT studies (6 studies, N = 185; MD = 1.57, 95% CI -0.26 to 3.40; p = 0.09; I² = 43%); the betweensubgroup difference did not reach conventional statistical significance (p = 0.37; I² = 0%) (see Figure 4)."The original version of this article has been updated.In Section 3, subsection 3.5 Subgroup analysis by training context (RT versus non-RT), the sentence beginning "For body composition outcomes, a significant pooled effect ... LBM ... p < 0.001; I² = 93.6%" incorrectly reported the LBM subgroup study counts, participant counts, pooled estimates, and between-subgroup statistics. A correction has been made to the section 3. Results, 3."For body composition outcomes, a significant pooled effect was observed for FFM in RT studies (12 studies, N = 228; MD = 3.39 kg, 95% CI 1.77 to 5.02; p < 0.001; I² = 0%), but not in non-RT studies (3 studies, N = 99; MD = -0.89 kg, 95% CI -3.94 to 2.15; p = 0.566; I² = 24.0%), with evidence of a betweensubgroup difference (p = 0.009; I² = 85.4%). LBM showed a significant pooled effect in RT studies (12 studies, N = 279; MD = 2.70 kg, 95% CI 1.56 to 3.85; p = 0.0001; I² = 0%), whereas no significant pooled effect was observed in non-RT studies (4 studies, N = 65; MD = -0.54 kg, 95% CI -1.75 to 0.68; p = 0.39; I² = 0%); the between-subgroup difference was significant (p = 0.0001; I² = 93.1%)."The original version of this article has been updated.In Section 3, subsection 3.6 Subgroup analyses for CMJ, the sentence beginning "When stratified by intervention duration ... I² = 78.4%" incorrectly reported study counts, participant counts, pooled estimates, and subgroup statistics. A correction has been made to the section 3. Results, 3.6 Subgroup analyses for CMJ: "When stratified by intervention duration, the pooled effect was not significant in studies lasting <8 weeks (8 studies, N = 225; MD = 0.77 cm, 95% CI -0.97 to 2.50; p = 0.39; I² = 67%), whereas a significant pooled effect was observed in studies lasting ≥8 weeks (3 studies, N = 48; MD = 8.06 cm, 95% CI 3.87 to 12.25; p = 0.0002; I² = 79%); a between-subgroup difference was observed (p = 0.002; I² = 89.9%) (see Figure 5)."The original version of this article has been updated.In Section 3, subsection 3.6 Subgroup analyses for CMJ, the sentence beginning "When stratified by competitive level ... I² = 17.8%" incorrectly reported study counts, participant counts, pooled estimates, and subgroup statistics. A correction has been made to the section 3. Results, 3.6 Subgroup analyses for CMJ: "When stratified by competitive level, a significant pooled effect was observed in competitive participants (7 studies, N = 138; MD = 3.36 cm, 95% CI 0.10 to 6.63; p = 0.04; I² = 93%), whereas no significant pooled effect was observed in recreational participants (4 studies, N = 135; MD = 1.17 cm, 95% CI -1.77 to 4.12; p = 0.44; I² = 39%); no significant between-subgroup difference was detected (p = 0.33; I² = 0%)."The original version of this article has been updated.In Section 3, subsection 3.6 Subgroup analyses for CMJ, the sentence beginning "When stratified by supplementation frequency ... I² = 0%" incorrectly reported study counts, participant counts, pooled estimates, and subgroup statistics. A correction has been made to the section 3. Results, 3.6 Subgroup analyses for CMJ: "When stratified by supplementation frequency, no significant pooled effect was observed in studies with daily supplementation (8 studies, N = 157; MD = 2.81 cm, 95% CI -0.22 to 5.83; p = 0.07; I² = 92%), whereas no significant pooled effect was observed in studies with non-daily supplementation (3 studies, N = 116; MD = 2.52 cm, 95% CI -0.53 to 5.57; p = 0.11; I² = 9%); no evidence of a between-subgroup difference was observed (p = 0.90; I² = 0%)."The original version of this article has been updated.In Section 3, subsection 3.8 Assessment of publication bias, the sentence beginning "Visual inspection of funnel plots revealed no clear asymmetry ... LBM, p = 0.222" incorrectly reported the Egger's test p values for CMJ and LBM. A correction has been made to the section 3. Results, 3.8 Assessment of publication bias: "Visual inspection of funnel plots revealed no clear asymmetry, and Egger's tests were non-significant across outcomes (squat 1RM, p = 0.915; leg press 1RM, p = 0.397; CMJ, p = 0.949; Wingate peak power, p = 0.340; Wingate mean power, p = 0.724; FFM, p = 0.213; LBM, p = 0.140). These findings indicate no evidence of small-study effects or publication bias (see Figure 6)."The original version of this article has been updated.In Section 4, subsection 4.1 Main findings, the paragraph beginning "This systematic review and metaanalysis included 39 randomized controlled trials ... I² = 88.5%" contained an incorrect total number of included trials and an outdated CMJ heterogeneity value. A correction has been made to the section 4. Discussion, 4.1 Main findings: "This systematic review and meta-analysis included 37 randomized controlled trials in healthy men aged 18-30 years and examined whether training context modifies the effects of creatine supplementation. Training context (RT vs non-RT) was prespecified as the primary analytical framework to assess contextual dependence. Lean mass outcomes (FFM and LBM) increased significantly only when creatine was combined with RT, whereas no significant effects were observed in non-RT settings. In contrast, Wingate peak and mean power improved significantly in both contexts, with little evidence that training type moderated these outcomes. CMJ showed a small overall improvement, but the high heterogeneity (I² = 89%) warrants a cautious interpretation, as it limits the robustness and stability of this pooled estimate. Training context did not significantly moderate this effect. For maximal strength, squat 1RM improved in both contexts, whereas leg press 1RM showed no significant overall effect." The original version of this article has been updated.In Section 4, subsection 4.3 Interpretation and potential mechanisms, the paragraph beginning "The high variance noted in CMJ outcomes ..." contained a sentence referring to more consistent CMJ effects in longer interventions, which was removed to align the discussion with the revised subgroup results. A correction has been made to the section 4. Discussion, 4.3 Interpretation and potential mechanisms: "The high variance noted in CMJ outcomes likely reflects that CMJ is a multi-factorial, skill-dependent performance outcome. CMJ height depends not only on energy availability but also on maximal strength (Suchomel et al., 2016), neuromuscular coordination, stretch-shortening cycle efficiency, technical execution, and testing protocols (Li et al., 2025;McErlain-Naylor et al., 2014;Xu et al., 2023). Accordingly, pooled CMJ effects should be interpreted with caution, even when statistically significant."The original version of this article has been updated.There was a mistake in Figure 1 The original version of this article has been updated.There was a mistake in Notes: MD = mean difference. Certainty of evidence was assessed using GRADE for continuous outcomes. For CMJ, certainty was downgraded due to very high heterogeneity (I² = 89%) and variability in testing protocols.The original version of this article has been updated.There was a mistake in The original version of this article has been updated.
Gu et al. (Wed,) studied this question.