Enhancing shoulder strength through blood flow restriction training: a systematic review and meta-analysis Authors: Hediyeh Hamedi, Mohammad Amin Gholipour, Tahere Seyedhoseinpoor Published: 2025 ABSTRACT AbstractBackgroundBlood flow restriction training (BFRT) has emerged as a promising way to enhance strength and build muscle with low-load exercises. This study was aimed to evaluate the effects of upper limb BFRT on shoulder strength.MethodsThe literature search encompassed PubMed, Scopus, Ovid, Web of Science, EBSCO, CENTRAL, PEDro, and Google Scholar, with no language restrictions. Included studies were clinical trials investigating the application of BFRT on the upper extremity, specifically assessing shoulder strength in flexion, abduction, internal rotation, and external rotation. The certainty of the evidence was assessed using the GRADE approach.ResultsFive studies with 158 subjects were included. The pooled analysis revealed a large effect of BFRT on enhancing flexion strength compared to the control group (Morris’ dppc = 1.79, 95% CI: 1.09 to 2.49), and moderate to high effect on internal rotation strength (Morris’ dppc = 0.67, 95% CI: 0.24 to 1.01). BFRT did not lead to a statistically significant enhancement in abduction and external rotation strength.ConclusionThis systematic review and meta-analysis suggests that BFRT has the potential to improve shoulder flexion and internal rotation strength. However, its effects on abduction and external rotation appear to be limited. While the results are encouraging, low to very low certainty of the evidence (as assessed by GRADE) indicates that more high-quality trials are needed to confirm these results.Supplementary InformationThe online version contains supplementary material available at 10.1186/s13102-025-01431-9.Keywords: Blood flow restriction, Shoulder, Restriction training, Strength Background Resistance training is a well-established intervention for enhancing muscle strength, physical performance, and overall health across diverse populations, including healthy individuals and those with various medical conditions [1, 2]. One of its primary benefits is the promotion of muscle hypertrophy and strength, which are essential for optimizing functional capacity, athletic performance, and rehabilitation outcomes [3]. Consequently, resistance exercise is a cornerstone in clinical musculoskeletal rehabilitation programs [4], particularly for counteracting age-related sarcopenia and restoring muscle mass following periods of immobilization or disuse atrophy [5, 6]. Shoulder pain and dysfunction are among the most prevalent musculoskeletal complaints and are often associated with muscle weakness and atrophy. Resistance training is commonly prescribed as a therapeutic intervention to address these impairments [7]. However, in clinical settings, especially among populations with acute or chronic injuries, the application of high mechanical loads may not be feasible due to the increased risk of exacerbating symptoms or inducing secondary musculoskeletal injury [8]. Blood flow restriction training (BFRT) has emerged as a promising adjunctive modality, allowing for strength and hypertrophy gains using low-load exercise [9]. BFRT involves the application of a pneumatic cuff to the proximal portion of a limb to partially restrict arterial inflow and fully occlude venous return. This induces a localized hypoxic and ischemic environment that enhances metabolic stress during exercise. When combined with mechanical tension, this environment promotes muscle growth [10] through various mechanisms, including increased systemic hormonal responses [11], cellular swelling [11], reactive oxygen species (ROS) production [10, 12], activation of anabolic signaling pathways [13], and increased recruitment of fast-twitch muscle fibers [13, 14]. Although BFRT has been well studied in lower limbs, its application in the upper limb—particularly the shoulder—remains less clear [15, 16]. Unlike the lower extremities, where cuff placement and vascular anatomy allow relatively consistent occlusion, the upper limb presents unique difficulties: the deltoid and rotator cuff muscles lie proximal to the cuff site, meaning the hypertrophic and strength adaptations may occur predominantly in the distal musculature (e.g., biceps and triceps) rather than in the shoulder itself. This anatomical limitation raises important concerns about whether BFRT meaningfully benefits shoulder-specific function or whether observed effects are biased by the choice of exercises and outcome measures [15, 16]. Previous systematic reviews and meta-analyses have synthesized evidence on upper-limb BFRT but did not isolate shoulder-specific outcomes. Pavlou et al. (2023) [15] examined muscles proximal to the cuff but did not focus on the shoulder, while Jing et al. (2024) [16] compared low-load BFRT to traditional resistance exercises across the entire upper extremity without differentiating shoulder-specific strength. Therefore, a gap remains regarding the efficacy of BFRT for improving shoulder strength specifically—critical for rehabilitation populations unable to tolerate high loads. The present systematic review aims to fill this gap by evaluating the effects of upper-limb BFRT on shoulder strength outcomes. By focusing explicitly on the shoulder, considering the influence of cuff placement and upper-limb anatomy, and synthesizing evidence with clinically relevant outcome measures, this review provides novel insights for clinicians seeking low-risk, evidence-based interventions to optimize shoulder rehabilitation. Methods This systematic review was conducted in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines [17]. The review protocol was registered prospectively in the International Prospective Register of Systematic Reviews (PROSPERO; registration number: CRD42024605189), and the full protocol has been published elsewhere in detail [18]. Search strategy and study selection A comprehensive literature search was performed across multiple electronic databases, including PubMed, Scopus, Ovid, Web of Science, EBSCO, the Cochrane Central Register of Controlled Trials (CENTRAL), the Physiotherapy Evidence Database (PEDro), and Google Scholar without language restriction. Additionally, manual searches were conducted on clinical trial registries such as ClinicalTrials.gov, the WHO International Clinical Trials Registry Platform, the International Standard Randomized Controlled Trials Number (ISRCTN), the EU Clinical Trials Register, and the Iranian Registry of Clinical Trials (IRCT) to identify ongoing or unpublished studies. Reference lists of relevant systematic reviews and included articles were also screened to ensure the comprehensiveness of the search. In cases where full-text articles were not available, corresponding authors were contacted directly. If unpublished data were identified, authors were approached to verify whether the work had subsequently been published. A maximum of three contact attempts via email were made. Studies were excluded if no response was received. The electronic search strategy incorporated a combination of keywords and controlled vocabulary terms (e.g., MeSH) such as “blood flow restriction,” “KAATSU training,” “vascular occlusion,” and “shoulder.” Search syntax was tailored to each database. Detailed search strategies for each database are provided in Additional File 1. Eligibility criteria All retrieved references were imported into EndNote X21 (Clarivate Analytics, Philadelphia, PA, USA), and duplicates were removed. Three reviewers (TSH, HH, and MGH) independently screened the titles and abstracts against the predefined eligibility criteria. Full texts of potentially eligible studies were then reviewed in detail to determine final inclusion. Any disagreements between reviewers were first discussed collaboratively. If consensus could not be reached, the third reviewer (MGH) acted as an adjudicator to make the final decision. To be eligible for inclusion, studies had to meet the following conditions: Study Design: Clinical trials and randomized controlled trials (RCTs) were considered. Study Design: Clinical trials and randomized controlled trials (RCTs) were considered. Intervention: The intervention involved the use of BFRT specifically applied to the upper limb (proximal arm or shoulder region), with the cuff placed on the proximal arm to induce partial arterial inflow restriction and venous occlusion relevant to shoulder musculature. This criterion was defined to ensure that the intervention directly targeted muscles contributing to shoulder strength. Intervention: The intervention involved the use of BFRT specifically applied to the upper limb (proximal arm or shoulder region), with the cuff placed on the proximal arm to induce partial arterial inflow restriction and venous occlusion relevant to shoulder musculature. This criterion was defined to ensure that the intervention directly targeted muscles contributing to shoulder strength. Treatment Protocol: The BFRT could be implemented as either a single session or a series of sessions, with or without adjunctive therapeutic interventions. Treatment Protocol: The BFRT could be implemented as either a single session or a series of sessions, with or without adjunctive therapeutic interventions. Participants: Studies involving healthy adults, athletes, or individuals with shoulder disorders—either pre- or post-surgical—were included. Participants: Studies involving healthy adults, athletes, or individuals with shoulder disorders—either pre- or post-surgical—were included. Outcome Measures: Shoulder strength or power assessed using any type of dynamometer, in the directions of flexion, abduction, internal rotation, or external rotation. Outcome Measures: Shoulder strength or power assessed using any type of dynamometer, in the directions of flexion, abduction, internal rotation, or external rotation. Studies were excluded based on the following conditions: Study Design: Case reports, case series, and studies employing a pre-post (before–after) design were not considered. Studies were excluded if they lacked a non-BFRT control group or only compared different BFRT protocols. Study Design: Case reports, case series, and studies employing a pre-post (before–after) design were not considered. Studies were excluded if they lacked a non-BFRT control group or only compared different BFRT protocols. Accessibility: Studies for which the full text could not be obtained were excluded. Accessibility: Studies for which the full text could not be obtained were excluded. Risk of bias assessment The risk of bias for each included study was independently assessed by two reviewers (HH and MGH) using the Cochrane Risk of Bias 2 (ROB-2) tool, which evaluates potential bias across five domains: the randomization process, deviations from intended interventions, missing outcome data, outcome measurement, and selection of the reported result. Each domain was rated as “low risk,” “some concerns,” or “high risk.” Disagreements were resolved through discussion, and if necessary, a third reviewer (TSH) was consulted to reach consensus. The Risk-of-Bias Visualization (robvis) tool was utilized to create graphical representations of the risk of bias across studies, following Cochrane and robvis guidelines [19]. To evaluate interrater reliability, Gwet’s first-order agreement coefficient (AC1) along with its 95% confidence interval was calculated [20]. Unlike Cohen’s kappa, Gwet’s AC1 provides a more stable measure of agreement and is less susceptible to the so-called “kappa paradox,” where low kappa values may occur despite high observed agreement [21]. Data extraction Two reviewers (HH and MGH) independently collected data using a standardized extraction form developed specifically for this review. The extracted data included: Study Details: Information on the first author, year of publication, study quality rating, design type, and total number of participants. Study Details: Information on the first author, year of publication, study quality rating, design type, and total number of participants. Participant Profiles: Demographic characteristics such as age and sex, as well as participant health status (e.g., healthy individuals, patients with shoulder disorders, or athletes). Participant Profiles: Demographic characteristics such as age and sex, as well as participant health status (e.g., healthy individuals, patients with shoulder disorders, or athletes). Intervention Characteristics: Details regarding the number of participants in each group, the frequency and duration of the intervention (in weeks), session length, BFRT intensity, and whether blinding was implemented for participants, therapists, or outcome assessors. Intervention Characteristics: Details regarding the number of participants in each group, the frequency and duration of the intervention (in weeks), session length, BFRT intensity, and whether blinding was implemented for participants, therapists, or outcome assessors. Outcome Measures: Data on shoulder strength in various movement directions (flexion, abduction, internal and external rotation), the devices used to assess strength or power, as well as study declarations regarding ethical approval, funding sources, and conflicts of interest. Outcome Measures: Data on shoulder strength in various movement directions (flexion, abduction, internal and external rotation), the devices used to assess strength or power, as well as study declarations regarding ethical approval, funding sources, and conflicts of interest. To ensure completeness and consistency, a third author (TSH) reviewed the extracted data. Any discrepancies between reviewers were discussed and resolved through consensus. When additional information was required, the authors of the original studies were contacted. Data synthesis When outcomes were consistently reported across eligible studies, the combined effects of continuous variables were analyzed using Morris’s delta for pretest-posttest-control group designs (dppc) with a it 95% confidence interval (CI) [18, 22], applying the effect size calculator provided by the Campbell Collaboration (http://www.campbellcollaboration.org/escalc/html/EffectSizeCalculator-SMD-main.php). This approach was chosen because it is specifically suited to pre-post-controlled trial designs, as it incorporates both baseline and follow-up measurements. Unlike standardized mean difference (SMD) or Hedges’ g, which rely primarily on post-intervention data, Morris’s dppc adjusts for baseline variability between groups and yields a more accurate estimate of the true intervention effect [23, 24]. This is particularly important in rehabilitation research, where small sample sizes and pre-post reporting are common. To compute pooled effects, outcome measures were first standardized to a 0–100 scale. A random-effects model was used for all pooled analyses to account for potential heterogeneity across studies in terms of participant characteristics, intervention protocols, and outcome measurement methods. This model assumes that the true effect size may vary between studies, providing more conservative and generalizable estimates than a fixed-effects model, particularly in meta-analyses with small numbers of studies and diverse populations. Further analyses were performed using STATA software (Version 14; StataCorp, College Station, TX, USA). According to conventional thresholds, a dppc less than 0.40 indicates a small effect, values between 0.40 and 0.70 represent a moderate effect, and values greater than 0.70 are considered large. An effect size of moderate magnitude or greater was interpreted as a meaningful impact on the outcome [23]. Heterogeneity assessment To evaluate variability across the included studies, statistical heterogeneity was assessed using the I² statistic and Cochran’s Q test (χ²), in accordance with the Cochrane Handbook for Systematic Reviews of Interventions [25]. The I² values were interpreted using the following thresholds: 0–40% indicated low or negligible heterogeneity, 30–60% reflected moderate heterogeneity, 50–90% signified substantial heterogeneity, and values between 75 and 100% represented considerable heterogeneity [26]. Prior to performing the meta-analysis, the extent of heterogeneity was examined. When I² values exceeded 50% and the forest plot demonstrated overlapping confidence intervals, a random-effects model was applied to accommodate between-study differences [27]. A p-value of less than 0.05 was considered statistically significant. Publication bias assessment Begg’s and Egger’s regression methods were employed to evaluate the presence of publication bias [27]. Additionally, the ‘trim and fill’ technique was applied to estimate the potential impact of such bias on the results [28]. All analyses related to publication bias were conducted using STATA software (Version 14; StataCorp, College Station, TX, USA). Sensitivity analysis Sensitivity analyses were conducted using the leave-one-out (Jackknife) method to evaluate the influence of individual studies on the overall results [29]. These analyses were performed using STATA software (Version 14; StataCorp, College Station, TX, USA). Summary of evidence The GRADE (Grading of Recommendations Assessment, Development and Evaluation) approach was employed to evaluate the certainty of evidence and the strength of recommendations for the primary outcomes. This framework considers five critical domains: study limitations (e.g., risk of bias), inconsistency (e.g., variability in results across studies), indirectness (e.g., differences in populations, interventions, or outcomes), imprecision (e.g., wide confidence intervals or small sample sizes), and publication bias. Based on these factors, the overall quality of evidence is rated as high, moderate, low, or very low [30]. Results of the systematic review and meta-analysis Study selection As depicted in the PRISMA flow diagram (Fig. 1), a total of 1,697 records were initially identified through the database search. After removing duplicates, 1,181 unique studies remained. Following title and abstract screening, the full texts of 23 articles were assessed for eligibility. Ultimately, 5 studies met the inclusion criteria and were incorporated into the meta-analysis. Of the included studies, one focused on individuals with rotator cuff tendinopathy [31], another involved athletes [32], and the remaining three examined healthy participants [33–35]. A total of 158 subjects participated in the five included trials. Table 1 summarizes the characteristics of the intervention and control groups, including study design, treatment protocols, measured outcomes, and the assessment tools employed. Flow diagram of the study identification and inclusion procedure Table 1The studies are arranged in chronological order, starting with the most recent and progressing to the earliest based on year of publicationAuthor -yearStudy designTreatment group characteristicsTreatment duration & TypeControl group characteristicsControl duration & TypeBlindingOutcome measureMeasurement toolPatientAssessor1 Kara 2024RCTN = 14, Rotator cuff tendinopathyAge: 31.1 ± 2.9 yGender: Male = 7Female = 7• 8 weeks• 16 sessions• resisted training + BFRT (50%)• BFRT applied by manually Standard KAATSU Unite)• Cuff placement: Details not availableN = 14, Rotator cuff tendinopathyAge: 34.9 ± 3.2 yGender: Male = 7Female = 7• 8 weeks• 16 sessions• resisted trainingnoyes• Int Rot strength (60°, 180°)• Ext Rot strength (60°, 180°)Dynamometer2 Lambert 2023RCTN = 15, AthletesAge: 19.7 ± 1.2 yGender: N/A*• 8 weeks• 16 sessions• low load resisted exs + BFRT (50%)• BFRT applied by automated tourniquet system (Delfi Medical Innovations• Cuff placement: Distally to the shoulderN = 13, AthletesAge: 19.8 ± 1.5 yGender: N/A• 8 weeks• 16 sessions• low load resisted exsnoyes• Flex strength• Scaption strength• Int Rot strength (0°, 90°)• Ext Rot strength (0°, 90°)Dynamometer3 Lambert 2021RCTN = 16, HealthyAge: 27.6 ± 2.2 yGender: Male = 13Female = 3• 8 weeks• 16 sessions• low load resisted exs + BFRT (50%)• BFRT applied by automated tourniquet system (Delfi Medical Innovations)• Cuff placement: Distally to the shoulder on the brachial region of the armN = 16, HealthyAge: 25.8 ± 2.2 yGender: Male = 10Female = 6• 8 weeks• 16 sessions• low load resisted exsnoyes• Flex strength• Int Rot strength (0°, 90°)• Ext Rot strength (0°, 90°)Dynamometer4 Brumitt 2020RCTN = 24, HealthyAge: 25 yGender: N/A• 8 weeks• 16 sessions• external rot exs + BFRT (50%)• BFRT applied by automated tourniquet system (Delfi Medical Innovations• Cuff placement: Proximal portion of the upper extremityN = 22, HealthyAge: 25 yGender: N/A• 8 weeks• 16 sessions• external rot exsnoyes• Abd strength• Int Rot strength (0°, 90°)• Ext Rot strength (0°, 90°)Dynamometer5 Bowman 2020RCTN = 14, HealthyAge: 26.2 ± 3.4 yGender: Male = 6Female = 8• 6 weeks• 12 sessions• low load resisted exs + BFRT (60%)• BFRT applied by automated tourniquet system (Delfi Medical Innovations• Cuff placement: Upper armN = 10, HealthyAge: 27 yGender: Male = 4Female = 6• 6 weeks• 12 sessions• low load resisted exsnoyes• Flex torque• Abd/Scaption torque• Int Rot torque• Ext Rot torqueDynamometerRCT randomized control trial, N/A not available, BFRT blood flow restriction training, Int Rot internal rot, Ext Rot external rotation, Flex flexion, Abd abduction, exs exercise*Data is not reported The studies are arranged in chronological order, starting with the most recent and progressing to the earliest based on year of publication N = 14, Rotator cuff tendinopathy Age: 31.1 ± 2.9 y Gender: Male = 7 Female = 7 • 8 weeks • 16 sessions • resisted training + BFRT (50%) • BFRT applied by manually Standard KAATSU Unite) • Cuff placement: Details not available N = 14, Rotator cuff tendinopathy Age: 34.9 ± 3.2 y Gender: Male = 7 Female = 7 • 8 weeks • 16 sessions • resisted training • Int Rot strength (60°, 180°) • Ext Rot strength (60°, 180°) N = 15, Athletes Age: 19.7 ± 1.2 y Gender: N/A* • 8 weeks • 16 sessions • low load resisted exs + BFRT (50%) • BFRT applied by automated tourniquet system (Delfi Medical Innovations • Cuff placement: Distally to the shoulder N = 13, Athletes Age: 19.8 ± 1.5 y Gender: N/A • 8 weeks • 16 sessions • low load resisted exs • Flex strength • Scaption strength • Int Rot strength (0°, 90°) • Ext Rot strength (0°, 90°) N = 16, Healthy Age: 27.6 ± 2.2 y Gender: Male = 13 Female = 3 • 8 weeks • 16 sessions • low load resisted exs + BFRT (50%) • BFRT applied by automated tourniquet system (Delfi Medical Innovations) • Cuff placement: Distally to the shoulder on the brachial region of the arm N = 16, Healthy Age: 25.8 ± 2.2 y Gender: Male = 10 Female = 6 • 8 weeks • 16 sessions • low load resisted exs • Flex strength • Int Rot strength (0°, 90°) • Ext Rot strength (0°, 90°) N = 24, Healthy Age: 25 y Gender: N/A • 8 weeks • 16 sessions • external rot exs + BFRT (50%) • BFRT applied by automated tourniquet system (Delfi Medical Innovations • Cuff placement: Proximal portion of the upper extremity N = 22, Healthy Age: 25 y Gender: N/A • 8 weeks • 16 sessions • external rot exs • Abd strength • Int Rot strength (0°, 90°) • Ext Rot strength (0°, 90°) N = 14, Healthy Age: 26.2 ± 3.4 y Gender: Male = 6 Female = 8 • 6 weeks • 12 sessions • low load resisted exs + BFRT (60%) • BFRT applied by automated tourniquet system (Delfi Medical Innovations • Cuff placement: Upper arm N = 10, Healthy Age: 27 y Gender: Male = 4 Female = 6 • 6 weeks • 12 sessions • low load resisted exs • Flex torque • Abd/Scaption torque • Int Rot torque • Ext Rot torque RCT randomized control trial, N/A not available, BFRT blood flow restriction training, Int Rot internal rot, Ext Rot external rotation, Flex flexion, Abd abduction, exs exercise *Data is not reported Three of the included studies reported shoulder flexion strength [32, 33, 35], three reported abduction strength [32–34], four assessed internal rotation strength [31–33, 35], and five evaluated external rotation strength [31–35]. All trials implemented moderate-intensity BFRT (50% [31, 32, 34, 35] or 60% [33] occlusion) as part of their intervention. Additionally, all studies employed assessor blinding in their measurement protocols. Quality assessment and level of evidence The quality of the included studies was assessed using the Cochrane Risk of Bias 2 (RoB 2) tool. Most assessments indicated a moderate risk of bias, primarily due to concerns in Domain 2. These concerns arose because neither the participants nor the intervention providers were blinded, potentially introducing bias into the outcomes (Fig. 2). Assessment of potential bias across the included studies Effects of interventions Three of the included studies [32, 33, 35] evaluated the impact of the intervention on shoulder flexion strength. The pooled analysis revealed a statistically significant and large effect of BFRT on enhancing shoulder flexion strength compared to the control group (Morris’ dppc = 1.79, 95% CI: 1.09 to 2.49). However, there was considerable heterogeneity across these studies (χ² = 8.61, I² = 76.8%, p = 0.00) (Fig. 3A). Forest plot of the effect of BFRT compared to the control group on: (A) shoulder flexion strength, (B) shoulder abduction strength, (C) shoulder internal rotation strength, and (D) shoulder external rotation strength Among the studies included in this meta-analysis, three [32–34] examined the effect of the intervention on shoulder abduction strength. The pooled results indicated that BFRT did not produce a statistically significant improvement in abduction strength compared to controls (Morris’ dppc = 0.17, 95% CI: −0.24 to 0.59). Notably, there was substantial heterogeneity among these studies (χ² = 5.02, I² = 60.2%, p = 0.08) (Fig. 3B). Four studies included in this meta-analysis [31–33, 35] assessed the impact of the intervention on shoulder internal rotation strength. The pooled analysis showed a moderate to large positive effect of BFRT compared to control groups (Morris’ dppc = 0.67, 95% CI: 0.24 to 1.01), with low heterogeneity across studies (χ² = 4.06, I² = 26.0%, p = 0.26) (Fig. 3C). All studies included in this meta-analysis evaluated the effect of the intervention on shoulder external rotation strength [31–35]. The combined findings showed that BFRT did not lead to a statistically significant enhancement in external rotation strength compared to the control group (Morris’ dppc = 0.15, 95% CI: −0.13 to 0.44). Additionally, variability across studies was minimal, indicating consistent results (χ² = 4.66, I² = 14.2%, p = 0.32) (Fig. 3D). Sensitivity analysis Given the limited number of included studies, conducting a subgroup analysis for the outcomes was not feasible. Sensitivity analysis based on study quality, which is useful for addressing heterogeneity, could not be performed, as all included studies were rated as having “some concerns.” Moreover, Jackknife Method analysis revealed that no single study had a significant impact on the overall pooled results across the four outcome measures (Fig. 4). Influence analysis of individual studies on the aggregated effect estimates for: (A) shoulder flexion strength, (B) shoulder abduction strength, (C) shoulder internal rotation strength, and (D) shoulder external rotation strength Assessment of publication bias Egger’s regression test indicated no significant evidence of publication bias for outcomes related to shoulder abduction, internal rotation, and external rotation strength. However, a significant bias was detected for shoulder flexion strength (Fig. 5A–D). In contrast, Begg’s test did not reveal any significant publication bias across all four outcomes. Further analysis using the trim-and-fill method did not identify any potentially missing studies for abduction or external rotation strength. However, the method imputed two missing studies for flexion strength and one for internal rotation strength, suggesting potential underreporting in these domains. Egger’s graphs illustrating the assessment of publication bias in the meta-analysis for: (A) shoulder flexion strength, (B) shoulder abduction strength, (C) shoulder internal rotation strength, and (D) shoulder external rotation strength Quality of evidence assessment The quality of evidence for each outcome was evaluated using the GRADE framework. For shoulder flexion and abduction strength, the evidence was rated as very low, indicating that current findings are based on studies with substantial limitations and provide only limited support for these outcomes (Table 2). Table 2Grades of Recommendation, Assessment, Development and Evaluation quality of evidence for the primary outcomesOutcomeLimitationsInconsistencyIndirectnessImprecisionPublication BiasEffect sizeGRADE QualityFlexion-1*-1†0-2-2ǂ+2£ Very lowAbduction-1*-1†0-200 Very lowInternal Rotation-1*00-1-1ǂ+1 LowExternal Rotation-1*00-100 LowThe symbols indicate how many of the items were met (one item was met for each level of evidence)* Downgraded by one level as the majority of the studies have some concerns in risk of bias† One level downgrade because I2 value was >50%ǂ Downgraded by two or one level due to suspected publication bias£ Upgraded two or one level due to large or medium effect size Grades of Recommendation, Assessment, Development and Evaluation quality of evidence for the primary outcomes The symbols indicate how many of the items were met (one item was met for each level of evidence) * Downgraded by one level as the majority of the studies have some concerns in risk of bias † One level downgrade because I2 value was >50% ǂ Downgraded by two or one level due to suspected publication bias £ Upgraded two or one level due to large or medium effect size The quality of evidence for internal and external rotation strength was rated as low. The rating for internal rotation was downgraded due to study limitations, imprecision, and potential publication bias, but partially offset by an upgrade for demonstrating a large effect size. External rotation was similarly downgraded for study limitations and imprecision, with no upgrading factors identified (Table 2). In summary, although the included studies offer some support for the effectiveness of the interventions, the overall confidence in the findings remains limited due to methodological concerns and inconsistency in the evidence base. Discussion The findings suggest that BFRT yields a substantial improvement in shoulder flexion and internal rotation strength, while its impact on abduction and external rotation appears negligible. However, given the very low to low quality of the available evidence, these results should be interpreted with caution. Further high-quality randomized controlled trials are necessary to confirm these effects and strengthen the reliability of the conclusions. The specific exercises incorporated in BFRT may inherently favor shoulder flexion and internal rotation, which could explain the observed greater strength gains in these movements. For example, several included protocols emphasized pressing-based movements such as the bench press and shoulder press [32], push-up variations [31, 32], and resisted internal rotation tasks using elastic bands or cables [33, 35]. These exercises predominantly recruit the pectoralis major, anterior deltoid, and subscapularis muscles—key contributors to shoulder flexion and internal rotation. Consequently, when these movements are combined with BFRT, the involved muscle groups are more likely to undergo hypertrophic and neuromuscular adaptations [15]. BFRT typically involves placing a cuff on the proximal part of the limb, which exerts direct pressure on muscles that contribute to these motions. This proximal occlusion enhances muscle activation and metabolic stress, promoting strength improvements particularly in muscles situated closer to the trunk. Additionally, BFRT has been shown to induce hypertrophy in muscles proximal to the occlusion site. As a result, muscles involved in flexion and internal rotation receive a greater training stimulus due to their anatomical proximity to the cuff [15]. In contrast, the lack of significant improvements in shoulder abduction and external rotation strength may be attributed to the anatomical location of the muscles responsible for these movements. Muscles such as the supraspinatus and middle deltoid (abduction), and the infraspinatus and teres minor (external rotation), are located more laterally and posteriorly. Since the majority of included protocols did not directly train these actions (i.e., isolated abduction or external rotation exercises were rarely included), and because these muscles lie further from the cuff site, their adaptive response to BFRT may be attenuated. This combination of anatomical positioning and exercise selection likely accounts for the comparatively limited strength adaptations observed [36]. Numerous studies have explored the impact of BFRT on muscle strength and hypertrophy across both upper and lower extremities [37]. Similar to Pavlou et al. [15], our findings support the effectiveness of BFRT in enhancing upper-limb muscle strength. However, while Pavlou et al. [15] primarily examined muscles proximal to the cuff without isolating shoulder-specific outcomes, our review focuses specifically on shoulder strength, providing a more targeted synthesis of clinically relevant outcomes. A systematic review and meta-analysis by Jing et al. [37] concluded that low-load BFRT is as effective as high-load resistance training in improving muscle strength and hypertrophy in the upper limbs. Consistent with Jing et al. [37], we also observed that low-load BFRT demonstrated superior outcomes compared to low-load training without BFRT, reinforcing its effectiveness in enhancing muscle performance [37]. Additionally, other studies have suggested that low-load BFRT may induce greater strength and hypertrophy in muscles located proximal to the occlusion site when compared to equivalent training without BFRT. However, the overall quality of evidence remains low, and results across studies are somewhat inconsistent [15]. In athlete populations, 58% of trials reported significantly greater strength gains in BFRT groups relative to non-BFRT controls. Notably, low-intensity exercise combined with BFRT produced comparable benefits to high-intensity training without BFRT [38]. A separate meta-analysis focusing on lower extremities found that BFRT significantly enhances muscle strength and functional performance, particularly when implemented three times per week [39]. For outcomes with substantial heterogeneity (I² >75%, e.g., flexion), subgroup analysis could not be performed due to the limited number of available studies. As the included trials generally employed comparable BFRT protocols, intervention-related factors are unlikely to fully account for the observed variability. A more plausible explanation may lie in differences in participant characteristics, since although most studies involved young adults, they encompassed heterogeneous populations ranging from healthy individuals and athletes to patients with musculoskeletal conditions. These differences in baseline health status are therefore potential contributors to the heterogeneity, although their precise impact cannot be determined from the current evidence. The assessment of publication bias suggests that the available evidence should be interpreted with caution. While Egger’s test detected potential bias for shoulder flexion strength and the trim-and-fill method indicated a few possibly missing studies for flexion and internal rotation, no significant bias was observed for abduction or external rotation. Clinically, this highlights that the reported effects, particularly for flexion and internal rotation, may be slightly overestimated due to underreporting in the literature, emphasizing the need for further high-quality trials to confirm these findings. Strengths and limitations of the study One of the major strengths of this systematic review and meta-analysis lies in its clearly defined inclusion criteria, which facilitated the selection of more homogenous and relevant studies. The methodological rigor adhered to PRISMA guidelines, and an extensive search strategy was employed across multiple electronic databases and gray literature sources to reduce the risk of publication bias and ensure comprehensive coverage. The use of standardized assessment tools, such as the Cochrane Risk of Bias 2 (RoB 2) tool and the GRADE approach, further strengthened the evaluation of methodological quality and evidence certainty. Despite these strengths, the review has notable limitations. The most prominent is the limited number of studies included in the meta-analysis, which restricts statistical power and limits the generalizability of the findings. Moreover, the small number of trials and generally small sample sizes raise the possibility of small-study effects, which may lead to an overestimation of BFRT’s effects in the literature. Additionally, the majority of included studies were conducted on healthy individuals or athletic populations, rather than clinical cohorts. As a result, the applicability of these findings to patients with musculoskeletal disorders—such as shoulder injuries or chronic pain—remains limited. Future research should prioritize well-designed trials involving clinical populations and larger sample sizes to enhance the external validity and clinical relevance of the findings. Conclusions This systematic review and meta-analysis suggests the potential of BFRT to enhance shoulder flexion and internal rotation strength, though its effects on abduction and external rotation appear limited. The observed specificity of strength gains may be attributed to the anatomical positioning of the occlusion cuff and the selection of exercises that preferentially activate muscles involved in flexion and internal rotation. However, these findings are preliminary and should not be interpreted as definitive evidences for clinical practice. the overall quality of evidence remains low to very low, and most of the included studies focus on healthy or athletic populations, limiting clinical applicability. To establish more definitive conclusions and inform clinical practice, future research should focus on high-quality randomized controlled trials with larger sample sizes and inclusion of individuals with musculoskeletal conditions. Consequently, firm clinical recommendations cannot be made at this stage. These efforts will help clarify the broader therapeutic potential of BFRT in shoulder rehabilitation and performance enhancement. Supplementary Information Supplementary Material 1. Supplementary Material 2. Acknowledgements The authors would like to thank the Babol University of Medical Sciences for all the support they have provided. Abbreviations Blood Flow Restriction Training Preferred Reporting Items for Systematic Reviews and Meta-Analyses Confidence Interval Grading of Recommendations Assessment, Development and Evaluation Authors’ contributions TSH was responsible for the conceptualization, development of methodology, formal analysis, data interpretation, and manuscript editing. HH contributed to data screening, data extraction, quality assessment, and drafting of the manuscript. MGH was involved in data screening, data extraction, and quality assessment. All authors have read and approved the final version of the manuscript and agree with the stated order of authorship. Funding No funds, grants, or other support was received. Data availability The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request. Declarations Ethics approval and consent to participate Not applicable. Consent for publication Not applicable. Competing interests The authors declare no competing interests. Footnotes Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. References Associated Data Supplementary Materials Supplementary Material 1. Supplementary Material 2. Data Availability Statement The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.