This research protocol was developed a priori and prospectively registered on the Open Science Framework (registration number: osf.io/qb9gc). This SR adheres to the Preferred Reporting Items for Systematic reviews and Meta-Analyses (PRISMA) extension for acupuncture (checklists are provided in Supplementary Tables 1–3). The methodology of this SR complies with the Assessment of Multiple Systematic Reviews-2 (AMSTAR-2) tool for critical appraisal of SRs.

The study eligibility criteria were developed using the Population, Intervention, Comparison, Outcomes, and Study design (PICOS) framework (17).

Population: Studies were eligible if they included adult athletes (≥18 years) and/or child athletes (<18 years) who were practicing any sport. Studies that included individuals participating in recreational sports were excluded. No exclusions were made based on geographical coverage.

Intervention: The intervention evaluated in this SR was WMA techniques, characterized by needle insertion based on conventional anatomical and physiological considerations. Our selection criteria specifically targeted WMA techniques to assess only intervention with an evidence-based rationale. Primary studies were included if the treatment approach was grounded in biomedical principles and excluded if it referenced traditional Asian medicine concepts (e.g., Qi, meridians, Yin-Yang). Eligible studies employed WMA techniques, including manual acupuncture (MA), PNE, or DN either administered alone or in combination with other interventions (e.g., physical exercise). We excluded studies that combined acupuncture with traditional Asian therapies (e.g., acupressure, cupping, or moxibustion). While all WMA techniques share a common biomedical rationale for acupoint selection, they differ procedurally. Throughout the manuscript, we use the umbrella term ‘WMA techniques’ and specify individual techniques where relevant.

Comparator: Studies were included if WMA techniques was compared with any comparator (e.g., sham acupuncture or physical exercise).

Outcome: The primary outcome of interest analyzed in this SR was the change in pain intensity, measured by mean pain scores before and after the intervention and mean pain scores between intervention and control groups. Studies reporting pain outcomes as odds ratios were eligible. We included studies that used pain diagnostic criteria. We included studies that measured the therapeutic effect of WMA techniques (pain intensity) using standards from conventional medicine or validated tools. We excluded studies that relied on measures derived from traditional Asian medicine.

In line with PRISMA guidelines for SRs on acupuncture, we included only studies utilizing terminology from Western medicine (e.g., pain intensity). We excluded studies using terminology from traditional medicine (e.g., syndrome score for syndrome remission) (18). We included primary studies that focused on pain management related to musculoskeletal pain. We excluded primary studies using acupuncture as short-term analgesia for surgical procedures.

Study design: We included clinical trials and observational studies. We excluded animal studies. Articles in languages other than English, Arabic, Spanish, or French (the languages spoken by the research team) were excluded if their English abstracts did not provide sufficient information to address our research questions.

The detailed selection criteria are presented in Supplementary Box 1.

We conducted systematic searches across the PubMed/MEDLINE, Web of Science, SPORTDiscus, Allied and Complementary Medicine databases, and Google Scholar covering publications up to 2023. The search strategy employed the keywords athletes and acupuncture, along with relevant synonyms, across all mentioned databases. The search strategy and choice of databases were reviewed and approved by an experienced librarian. The detailed controlled vocabulary and free-text terms used to search each database are provided in Supplementary Box 2. No language restrictions were applied. Reference lists of included articles and relevant reviews identified during title and abstract screening were manually screened to supplement the search.

Removal of duplicates and multistage screening were conducted using Rayyan software (Rayyan Systems Inc., Cambridge, MA, USA¹). Title and abstract screening were performed independently by two reviewers. The full-text articles were screened independently by three reviewers. All reviewers agreed on the set of articles to include, with any disagreements resolved through discussion among team members. The final list of excluded studies is provided in Supplementary Box 3.

Data extraction was independently conducted by two reviewers using a predesigned extraction sheet developed in Microsoft Excel. The sheet included the following: (i) study characteristics (i.e., study design, sampling method, data collection time, and sample size); (ii) setting; (iii) population description, including age and sex; (iv) details of the intervention; (v) outcome (i.e., mean pain scores, mean pain differences, and pain measurement tool); and (vi) additional items described in the STandards for Reporting Interventions in Clinical Trials of Acupuncture (STRICTA) checklist (19).

The extracted data were compared to identify discrepancies. All extracted data were independently checked for accuracy by a third reviewer. Any disagreements were resolved through discussion among team members.

The risk of bias (RoB) and declared conflicts of interest of the included studies were appraised independently by two reviewers. Each RCTs included in this review was assessed for RoB using the Cochrane Collaboration Risk of Bias 1 (RoB-1) tool (20) (Supplementary Table 4). Each RCT was assessed for six sources of bias: (i) sequence generation; (ii) allocation concealment; (iii) blinding of participants and personnel; (iv) blinding of the outcome assessment; (v) incomplete outcome data; and (vi) selective outcome reporting. The RoB for each study was categorized as low, some concerns, high, or unclear.

For observational studies, we used the National Heart, Lung, and Blood Institute (NHLBI) study quality assessment tool for observational cohort and cross-sectional studies (21) (Supplementary Table 5).

When differences in opinion emerged during the RoB assessment, they were discussed to reach consensus. In cases where the articles did not provide sufficient information for a clear determination, we attributed an unclear RoB to the specific source of bias. To ensure a comprehensive evaluation of each study’s methodological quality, we used information from the publications and their supplementary materials and communicated with the study authors when necessary. The final decision for each RoB assessment is reported along with relevant quotes from the study publications to provide transparency and support the rationale behind our assessments.

Weighted inverse-variance, random-effects meta-analyses were conducted to estimate pooled differences in mean pain scores: within groups (pre- vs. post-intervention) and between groups (intervention vs. control). If the standard deviation (SD) was not reported, we calculated it from the 95% confidence interval, as recommended by the Cochrane Handbook (22). The inclusion of studies in the meta-analysis is described in Supplementary Table 6 and Supplementary Box 4.

To explore heterogeneity, subgroup meta-analyses were conducted according to intervention type (i.e., WMA techniques alone and with exercise/physiotherapy), pain etiology [i.e., delayed-onset muscle soreness (DOMS), shoulder, elbow, hip, and knee], body section studied (i.e., upper and lower body), needle insertion location (i.e., in muscle and in tendon), and pain measurement tool (i.e., visual analog scale [VAS], numerical pain rating scale [NPRS], and patient-rated elbow evaluation [PREE]). We assessed heterogeneity between studies via the I² statistics, with a cutoff point of ≥50% and a p value <0.10 per the chi-squared test, which was defined as a significant level of heterogeneity. Statistical significance was set at α = 0.05. Differences in mean scores was considered significant when the 95% confidence interval did not include the null value (zero) (23). Meta-analyses were conducted with R software (version 4.0.0, 64 bit).

Potential small study effects or publication bias were explored via contour-enhanced funnel plots (24). The commonly used Egger’s test was not conducted because each meta-analysis included fewer than 10 studies.

The certainty of evidence was assessed using the Grading of Recommendations, Assessment, Development, and Evaluation (GRADE) approach (25) in accordance with the Cochrane Handbook.

A total of 4,548 records were identified through database and Google Scholar searches (Figure 1). After duplicate removal and title/abstract screening, 473 full-text articles were assessed for eligibility. Of these, 6 RCTs (26–31) and 2 longitudinal studies (32, 33) met the inclusion criteria, for a total of 8 studies included in this SR (26–33). A detailed description of the included studies, according to the STRICTA checklist (19), is provided in Supplementary Tables 6, 7.

The included studies examined adult athletes with elbow (29), shoulder (28, 30), hip (31), or knee (26, 27) pain, and healthy adolescent athletes with DOMS in the muscle groups of the lower limbs (32, 33). The studies evaluated various medical procedures falling within the broader framework of WMA, including DN (27–29, 31), MA (32, 33), ultrasound-guided DN (26, 30), or ultrasound-guided percutaneous needle electrolysis (PNE) (26). WMA techniques were applied either to muscles (27, 28, 30–33) or to tendons (26, 29). Interventions included WMA techniques alone (28, 30, 32, 33), WMA techniques combined with exercise (26, 27), or WMA techniques combined with both exercise and physiotherapy (29, 31). The control interventions varied substantially across studies and included DN applied at acupoints different from those selected for the intervention group (28), exercise (27), exercise combined with physiotherapy (29, 31), no intervention (30), or sham needling (26).

Pain intensity was assessed using different instruments across studies, including the Kujala anterior knee pain scale (27), NPRS (27, 30), PREE (29), and VAS (26, 28, 31–34).

All studies reported mean score differences, and none reported odds ratios. The number of acupuncture sessions ranged from one to nine, with needle sizes ranging from 0.20 × 25 mm to 0.30 × 65 mm. Outcome assessments were conducted between 5 days and 8 weeks post-intervention or averaged across five consecutive daily sessions. Treatment duration ranged from 1 to 4 weeks. Needle insertion depth, reported in only two studies, ranged from 3 mm to 25 mm (32, 33) (Supplementary Tables 7, 8).

RoB was assessed to evaluate the internal validity of the included studies and to inform confidence in the meta-analytic findings. All RCTs (6/6, 100%) (26–31) had a low RoB for sequence generation, as they all employed randomization to produce comparable intervention and control groups. All RCTs had a low RoB for selective outcome reporting, as they consistently reported findings on their prespecified outcomes (26–31) (Table 1).

The RoB for allocation concealment was unclear in most RCTs (4/6, 66.7%) (28–31) because the reports lacked sufficient detail to assess whether intervention allocations could have been anticipated before or during enrollment. Blinding of the researchers performing the intervention was not feasible due to the inherent features of the intervention and was therefore either not applied (26–28, 30) or not reported (29, 31) in any RCT. Blinding of participants and blinding of the outcome assessment were applied in most RCTs (4/6, 66.7%) (26–28, 30). Most RCTs (4/6, 66.7%) had a low risk of attrition bias (26–28, 30).

Both longitudinal studies (2/2, 100%) (32, 33) provided positive answers to most of the signaling questions in the quality assessment tool (8/14, 57.1%) (21), reflecting overall good quality (Table 2). Blinding of participants was applied in all longitudinal studies (2/2, 100%) (32, 33). However, neither blinding of the researchers involved in providing the treatment nor blinding of the assessors was performed in any study. No study reported the participation rate or adjusted for the impact of potential confounding variables on the relationship between WMA techniques and pain reduction.

Overall, the included studies showed low RoB in randomization, outcome reporting, and attrition. However, limitations were noted in allocation concealment, blinding, and the absence of confounding control in observational studies.

For shoulder pain, we identified two good-quality RCTs demonstrating that DN significantly reduced pain in overhead athletes (Supplementary Table 6). The first RCT (30), conducted in volleyball and handball players reported short-term shoulder pain relief after a single session of ultrasound-guided DN applied to the teres major (MD = −3.3, p value ≤ 0.001), compared with a minimal change in the control group (MD = −0.4, p value = 0.07), resulting in significant between-group difference (p < 0.001). The second RCT (28), conducted in semi-elite throwers, swimmers, volleyball, and basketball players found significant pre–post pain reductions after three DN sessions targeting the upper trapezius (intervention) or infraspinatus (control) (MD = −4.7 vs. − 4.8, both p value≤ 0.001). Despite differences in target muscles and procedures, both studies suggest that DN reduces shoulder pain in overhead athletes.

For elbow pain, one RCT (29) demonstrated that DN combined with exercise and physiotherapy resulted in significantly greater pain reduction than physiotherapy alone. Between-group differences were significant at the seventh session (p value< 0.0001), ninth session (p-value = 0.006), and 1 week post-treatment (p-value < 0.001). To date, no RCT has isolated the effect of DN alone on elbow pain.

For hip pain, one RCT (31) demonstrated that DN combined with physiotherapy and stretching significantly reduced pain compared to physiotherapy alone (31). Between-group differences were significant after the fifth session (p-value < 0.001). To date, no RCT has isolated the effect of DN alone on hip pain.

For knee pain, one RCT (26) conducted in athletes with patellar tendinopathy, compared DN plus exercise and PNE plus exercise to sham needling plus exercise, and found no significant differences between intervention and control groups. However, all three showed a significant improvement in pain after a 22-week intervention (p-value ≤ 0.05). The second RCT (26), conducted in female athletes with unilateral prepatellar or retro-patellar pain, reported that that DN combined with exercise produced greater reduction in unilateral prepatellar or retro-patellar pain than exercise alone at weeks 4 and 6 postintervention (p-values < 0.001) (27). Methodological heterogeneity in controls and pain measures prevents firm conclusions about the independent effectiveness of DN for knee pain.

For DOMS pain, two longitudinal studies (32, 33) used the same MA protocol and reported statistically significant post-intervention pain reductions among adolescents who received at least one of five treatment days (p-values<0.05). However, the clinical relevance of these reductions remains uncertain, as no minimal detectable change has been established for the VAS in DOMS (32).

To examine within-group changes in pain, we conducted meta-analyses comparing mean pain scores pre- and post-intervention. WMA techniques alone was associated with a significant reduction in the mean pain score [meta-analysis, number of studies (n) = 5, p-value = 0.002]. In contrast, WMA techniques combined with exercise and/or physiotherapy resulted in a nonsignificant reduction in the mean pain score (meta-analysis, n = 3, p-value = 0.206, forest plot in Figure 2). Both WMA techniques and WMA techniques combined with exercise and/or physiotherapy showed high heterogeneity in treatment effects, with I² values >50%.

Pre-post intervention differences in pooled mean scores varied significantly according to pain etiology, pain measurement instrument, and the number of sessions (p values<0.0001, forest plots in Supplementary Figures 1–3). No significant difference in the reduction of mean pain scores was observed based on the body section treated or the needle insertion location (forest plots in Supplementary Figures 4, 5).

Forest plot shows pooled mean differences (MD) with 95% confidence intervals for studies using WMA techniques alone or combined with exercise/physiotherapy (Ex/PT). The control group in Kamali, 2019, received dry needling in a shoulder muscle, and this was different from the method selected for the intervention group (Supplementary Table 6 and Supplementary Box 4). Intervention and control groups from Kamali, 2019 were included in the pre- to post meta-analysis.

We next examined between-group differences in pain by comparing WMA techniques interventions with control conditions. WMA techniques alone (n = 1, p-value = 0.0003) (30) and WMA techniques combined with exercise and/or physiotherapy (meta-analysis, n = 3, p value = 0.011) were associated with significant reductions in mean pain scores compared with control groups (forest plot in Figure 3). The meta-analysis of WMA techniques combined with exercise and/or physiotherapy revealed high heterogeneity in treatment effects, with an I² value of 86%. Differences in pooled mean scores between the intervention and control groups varied significantly according to the type of control intervention (p-value = 0.001), pain measurement instrument (p-value = 0.0003), and number of sessions (p-value < 0.0001; forest plots in Supplementary Figures 6–8). No significant decreases in mean pain scores were observed based on the intervention type, pain etiology, body section treated, or needle insertion location (forest plots in Supplementary Figures 9, 11).

Forest plot shows pooled mean differences (MD) with 95% confidence intervals comparing WMA techniques alone with control groups. Results are stratified by intervention type: WMA techniques alone versus WMA techniques combined with exercise and/or physiotherapy (Ex/PT). The standardized mean difference was not computed because variability between RCTs was not solely attributable to the measurement tool used. Standardization assumes that differences in SDs across studies are due to differences in measurement scales rather than actual variability among study populations (22).

Publication bias was assessed using contour-enhanced funnel plots. In the pre–post intervention meta-analyses, the distribution of effect sizes showed asymmetry consistent with small-study effects (Supplementary Figure 12). In the intervention versus control analyses, asymmetry was also observed, suggesting that smaller or nonsignificant studies may be underrepresented (Supplementary Figure 13).

The certainty of evidence was rated based on changes in pain intensity, the primary outcome of interest. Across the included studies, we identified five types of musculoskeletal pain: elbow pain, shoulder pain, hip pain, knee pain, and DOMS. While, the meta-analysis of five pre–post comparisons showed a statistically significant reduction in mean pain scores, the effectiveness of WMA techniques compared with a control group was evaluated in only one RCT, which showed a significant between-group difference (30). Consequently, the overall certainty of evidence was downgraded due to limited data, clinical and statistical heterogeneity, potential publication bias, and the risk of performance bias—an inherent issue when performing acupuncture interventions. Based on these factors, the certainty of evidence was rated as low for WMA techniques alone, and moderate for WMA techniques combined with physiotherapy and/or exercise. Overall, the available evidence suggests that WMA techniques may offer therapeutic benefit for managing musculoskeletal pain in athletes but confidence in the findings varies depending on the intervention type and study limitations.