Authors: Hüsnü Kocaman, Latif Aydos
Categories: Research, Blood flow restriction, Isokinetic contraction, Resistance exercise, Muscle strength, Peak Torque
Source: BMC Sports Science, Medicine and Rehabilitation
Authors: Hüsnü Kocaman, Latif Aydos
While high-load resistance training (HL-RT) is well established as an effective method for improving muscle strength, low-load blood flow restriction training (LL-BFRT) has been suggested to induce similar neuromuscular adaptations while imposing substantially lower mechanical stress. However, direct comparative evidence between LL-BFRT and HL-RT in the upper limbs remains limited.
Twenty-two healthy male participants (age: 26.3 ± 7.2 years, 18–40, 176.9 ± 6.5 cm; body 77.9 ± 13.5 kg) were randomly allocated to either LL-BFRT (n = 12) or HL-RT (n = 10). The LL-BFRT group trained at 30% of 1RM (4 sets; 30-15-15-15 repetitions) with progressive arterial occlusion pressure (30–50%), while the HL-RT group trained at 70% of 1RM (3 sets; 10–12 repetitions). Both groups, within their redesigned and equalised training programs, performed biceps curls and triceps pushdowns twice weekly for 7 weeks. Isokinetic muscle strength outcomes, including peak torque, relative strength, and total work, were assessed using isokinetic dynamometry (Isomed 2000, 60°/s). Maximal muscle strength outcomes included one-repetition maximum (1RM) strength. Statistical analyses were performed using two-way repeated measures ANOVA with Bonferroni corrections, reporting effect sizes (η²) and 95% confidence intervals (CI).
Both groups demonstrated significant pre- and post-improvements in isokinetic strength parameters (p < 0.05). No significant group × time interaction was observed for peak torque (right arm LL-BFRT + 4%, HL-RT + 9%; p = 0.478, η²=0.049). Both groups showed within-group increases in total work (LL-BFRT right arm + 16.3%, HL-RT + 8.4%) and relative strength (LL-BFRT left arm flexion + 19.0%, HL-RT + 10.6%). No adverse events were reported.
Preliminary evidence suggests that LL-BFRT can induce meaningful strength gains in healthy adults. These findings support LL-BFRT as a safe and potentially effective training alternative when high loads are not feasible, particularly in rehabilitation contexts. However, the small sample size and short duration of the study limit the generalizability of the results. Future studies with larger samples and extended follow-up periods are needed to confirm and expand upon these findings. These findings should be interpreted cautiously due to the limited statistical power to detect small-to-moderate between-group differences.
ClinicalTrials.gov, NCT07462520. Registered 5 March 2026.
Resistance exercises are widely used to increase muscle mass, develop and maintain muscle strength, and improve overall health and fitness in women and men of all ages [1]. Among the different training modalities, limited load blood flow restriction training (LL-BFRT) has attracted growing attention as an effective method for promoting hypertrophy and strength with lower mechanical loads [2–6]. Although the first scientific studies on blood flow restriction techniques were conducted in the 1990s, the method was originally developed by Yoshiaki Sato in 1966 [7].
Blood flow restriction (BFR) training induces multiple physiological mechanisms beyond a simple explanation. Specifically, the accumulation of metabolites such as lactate and hydrogen ions under restricted circulation leads to metabolic stress, which stimulates muscle protein synthesis and hypertrophy [8]. The resulting osmotic changes cause cell swelling, which activates anabolic signalling pathways and promotes muscle growth [9]. Furthermore, hypoxia during BFR accelerates fatigue of slow-twitch fibres and promotes the early recruitment of type II fibres, increasing neuromuscular activation similar to high-load training [4]. Endocrine responses are also enhanced, with significant elevations in growth hormone, IGF-1, and testosterone, along with suppression of catabolic markers, such as myostatin. Additional adaptations include HIF-1α activation and VEGF-mediated angiogenesis, which improve muscle perfusion and contribute to long-term hypertrophic responses [9]. Importantly, these effects cannot be attributed solely to placebo or motivation, since passive BFR without exercise has been shown to attenuate muscle atrophy during immobilisation, confirming a true physiological basis [8, 10].
Another well-established method to enhance muscle hypertrophy and strength is high-load resistance training (HL-RT), typically performed at 70–80% of one repetition maximum (1RM) [11]. HL-RT promotes hypertrophy mainly through high mechanical tension, activating the mTOR pathway and increasing protein synthesis [4, 9]. The recruitment of type II fibres, muscle microtrauma, satellite cell activation, and acute hormonal responses (testosterone, GH, IGF-1) collectively support rapid strength gains [9, 12–15].
Recent research demonstrates that LL-BFRT can elicit muscle strength and hypertrophy outcomes comparable to HL-RT, while imposing substantially lower mechanical stress [3–6, 16]. The method relies on metabolic stress and local hypoxia to enhance muscle activation and produce an effective hypertrophic response. It has also shown benefits in rehabilitation and recovery following injury [8, 10, 17]. Meta-analyses have reported similar strength and hypertrophy gains for both approaches [8, 10, 18–20], although some evidence suggests that HL-RT retains a small advantage in maximal strength development [21, 22].
Wang et al. (2025) and Chang et al. (2023) report in two separate meta-analyses that HL-RT is more advantageous than LL-BFRT in terms of strength gains [21, 22]. Specifically, Wang et al. (2025) reported that HL-RT had a small but significant advantage in strength gains in healthy volunteers, while both methods yielded similar results in speed and power increases. Chang et al. (2023), on the other hand, reported that LL-BFRT can provide strength gains comparable to those of HL-RT, particularly with individualised pressure and intermittent application protocols [22].
While a significant portion of the literature focuses on the lower extremities [20, 23–25], comparative data on the upper extremities have been limited. In a meta-analysis conducted by Jing et al. (2024), no significant difference was reported between LL-BFRT and HL-RT in upper extremity muscle strength and hypertrophy, and both methods were found to produce similar improvements [26]. However, it has been reported that isokinetic muscle strength parameters (e.g., peak torque, relative strength, and total work) are of critical importance, particularly in rehabilitation and performance assessments [27–30]. Isokinetic tests can objectively assess muscle strength at a constant speed and with adjustable resistance. Their ability to detect small changes makes them reliable and preferred tools in clinical applications [31]. Although handheld dynamometers are considered reliable, isokinetic systems can offer greater objectivity [32]. From this perspective, there is a lack of randomised controlled trials comparing LL-BFRT and HL-RT in upper-limb isokinetic performance. The present study, therefore, aims to address this gap by examining the effects of LL-BFRT versus HL-RT on elbow flexors and extensors, contributing to the current understanding of BFR applications in upper-limb training.
The aim of this study was to compare the effects of low-load blood flow restriction resistance training (LL-BFRT) and high-load resistance training (HL-RT) on upper-extremity muscle strength and isokinetic contraction parameters of the elbow flexor and extensor muscles.
Based on the existing literature, the primary hypothesis is that LL-BFRT would elicit improvements in upper-extremity isokinetic strength parameters (peak torque, total work, and relative strength) comparable to those observed with HL-RT, despite the substantially lower mechanical loads applied.
The secondary hypothesis is that both LL-BFRT and HL-RT would result in significant within-group improvements in maximal strength (1RM) and isokinetic performance measures over the intervention period.
This study was designed as a randomised controlled trial. All participants were individually informed about the research process and provided written informed consent prior to participation. The study was conducted in accordance with the principles of the Declaration of Helsinki and was approved by the Ethics Committee of Gazi University (approval November 9, 2022; approval E-77082166-302.08.01-506408; research 2022 − 1163).
The study was retrospectively registered at ClinicalTrials.gov (NCT07462520) on 5 March 2026.
A total of 22 healthy male adults were included in the study. All participants had been performing resistance training regularly for at least six months before enrollment and were therefore classified as recreationally trained, but not professional athletes.
Healthy individuals between the ages of 18 and 40Individuals not taking any dietary supplementsIndividuals not taking a substance containing anabolic ingredientsIndividuals subjected to a resistance training program for at least 6 monthsIndividuals not performing aerobic training for more than 30 min per dayIndividuals training with a personal trainer
Individuals with a skeletal muscle injury in the last 6 monthsIndividuals who have undergone a surgical operationIndividuals with chronic high blood pressure and cardiovascular diseaseIndividuals who miss three consecutive training sessionsIndividuals with chronic cardiovascular disease
Baseline demographic and anthropometric characteristics of the participants are summarised in Table 1.
Table 1Participant CharacteristicVariablesLL-BFRT(n = 12)HL-RT(n = 10)t
p
Age (years)29.41± 8.1423.10±4.263.1250.003Height (cm)177.50±7.81176.05±1.510.6510.519Body Weight (kg)80.21±14.3475.61±14.111.0690.291BMI (kg/m^2^)25.29±3.1224.29±3.910.9470.349Systolic Blood Pressure (mmHg)118.67±8.85116.1±9.040.9740.348Diastolic Blood Pressure (mmHg)72.33±6.8969.05±8.891.3780.175Oxygen Saturation (%SpO2)96.58±1.0296.85±1.59-0.6710.506Resting Heart Rate (beats/min)73.13 ± 9.9273.25±71.08-0.0300.976Body Fat Percentage (%)19.85 ± 7.1515.16±4.712.5140.016*Significant baseline differences were observed between groups in age and body fat percentage (p < 0.05). These variables were therefore considered when interpreting the training responsesp* < 0.05, **p < 0.01 indicate statistically significant between-group differences
The study group consisted of healthy adults who regularly performed resistance exercises and voluntarily agreed to participate in the study. Twenty-six volunteers who met the inclusion criteria participated in the pre-tests. Two participants withdrew from the study, stating that they could not comply with the training programme. Two participants were excluded from the study because they failed to meet the necessary criteria, having missed three consecutive training sessions. The LL-BFRT group and the HL-RT group each consisted of eleven individuals. During the adaptation week, one person in the LL-BFRT group was transferred to the HL-RT group due to an allergic reaction. Ultimately, the study was completed with LL-BFRT (n = 12) and HL-RT (n = 10) participants. The exercise programmes regularly followed by the participants were redesigned to equalise training intensity and volume, and resistance exercises for arm muscles were added at the end of their programmes. Two days prior to the start of the study, each participant’s maximal strength measurements and anthropometric measurements were taken.
Randomisation was performed using a simple random allocation procedure in Microsoft Excel. Prior to baseline testing, a random allocation sequence was generated by assigning each participant a random number using the RAND() function. Participants were then sorted in ascending order of the generated random numbers and allocated sequentially to the LL-BFRT or HL-RT group in an intended 1 manner. Group allocation was completed before baseline testing. Due to the nature of the exercise intervention, allocation concealment was not implemented.
Participants in the LL-BFRT group performed seated dumbbell biceps curls and standing cable triceps pushdowns at 30% of 1RM, using a repetition scheme of 30-15-15-15 across four sets, as commonly applied in previous studies [20, 33, 34]. The HL-RT group performed the same exercises at 70% of 1RM for three sets of 10–12 repetitions, in line with ACSM guidelines and previous resistance training recommendations [35–37].
For the LL-BFRT group, blood flow restriction was applied bilaterally to the most proximal portion of the upper arms using 5 cm wide pneumatic cuffs (Smart Cuff Elite, Smart Tools, USA), as suggested in the literature for upper-limb application [25, 38]. The cuffs were connected to a digital manometer to ensure accurate calibration before each session. Arterial occlusion pressure (AOP) was individually determined at the beginning of each session, and training was conducted at progressive 30% AOP in sessions 1–2, 40% AOP in sessions 3–5, and 50% AOP in sessions 6–14, consistent with prior recommendations [8, 25, 39].
The cadence for both groups was set to 0:2:0 (1 s concentric, no pause, 2 s eccentric, no pause), a tempo previously used in BFR training studies to standardize muscle contraction speed rest intervals were 30–45 s between sets in LL-BFRT, 60–90 s in HL-RT, and 3 min between exercises, as recommended in the BFR literature [4, 25, 40]. Certified strength and conditioning specialists supervised all training sessions to ensure adherence to the protocol [41].
To enhance safety, participants’ blood pressure (Omron M2 Basic, Netherlands), heart rate, and oxygen saturation (P2000 pulse oximeter, Germany) were monitored at every training session. No adverse events were reported, consistent with previous safety reports in BFR training [10, 25, 39].
Isokinetic muscle strength outcomes included the elbow flexors and extensors, measured by peak torque (Nm), relative strength (Nm/kg), and total work (Nm). Testing was performed using an isokinetic dynamometer (Isomed 2000, D&R Ferstl GmbH, Germany) in concentric-concentric mode at 60°/s through a 0°–125° range of motion [27, 42].
All tests were administered by the same experienced technician, and the dynamometer was calibrated prior to each testing session according to the manufacturer’s guidelines, as recommended in isokinetic assessment protocols [43].
Maximal muscle strength outcomes included one-repetition maximum (1RM) strength for biceps curls and triceps pushdowns, as well as anthropometric measures (height, body mass, BMI, arm circumference, and skinfold thickness) [8, 39, 44].
Training volume can be influenced by training frequency, training duration, and training intensity, and accordingly, the amount of strength or endurance in the training content can be shaped according to the training volume [45, 46]. In this context, participants included in the study were selected from individuals who performed regular resistance exercises under the guidance of a personal trainer and adhered to specific training programmes. Training volume can be determined by multiplying the number of repetitions performed by the number of sets. On the other hand, training volume may vary as a result of training sessions performed with different training intensities and different numbers of repetitions [47, 48]. The intensity of the training was limited to 30% of the 1 RM test in the LL-BFRT group and 70% in the HL-RT group. In order to minimise the effects of volume differences that may arise from intensity and repetition counts, three different training programmes were applied to the participants in the form of a split training programme [16, 49, 50] (Fig. 1). Training volume was equated between groups by matching the total number of sets and repetitions performed per exercise across LL-BFRT and HL-RT protocols, while load intensity differed according to the experimental condition. The LL-BFRT group performed arm exercises using the blood flow restriction method on the 1st and 3rd training days (Figs. 1 and 2).
Fig. 1Weekly Training Schedule
Fig. 2Details of the Blood Flow Restriction Applied to the Arm
Figure 3 shows the weekly training programmes followed by participants in the HL-RT group and LL-BFRT group included in the study over a period of seven weeks. Prior to each training session, a 10-minute warm-up involving dynamic movements encompassing the entire body was performed, followed by 5 min of eccentric and concentric contractions targeting the training area included in the programme to complete the warm-up. When performing the exercises in the programme, the HL-RT group rested for 60–90 s between sets and 1–3 min between movements. The restricted LL-BFRT group completed the movements in the programme by resting for 30–45 s between sets and 3 min between movements. The tempo of the movements was monitored using the digital metronome (Soundbrenner Pulse, Soundbrenner Ltd., Hong Kong, with a branch in Berlin, Germany) at a rate of 1 s concentric and 2 s eccentric, totalling 3 s. At the end of each training session, a 5-to 10-minute walk was performed to cool down and conclude the session.
Fig. 3Daily Training Programs
Blood flow restriction was applied using the Smart Tools-Smart Cuff Elite (USA). To ensure device compatibility, one week prior to testing, blood flow restriction was applied to participants. When determining AOP pressures, cuffs, i.e. restriction bands, were placed on the proximal (near the starting point) region of both arms, and arterial occlusion pressure(AOP) was applied at 50% [25]. The device automatically determines the restriction pressure corresponding to 50% of each participant’s capacity, in mmHg, on an individual basis. Ten participants in the LL-BFRT group completed the study in accordance with the research criteria. The LL-BFRT group performed the movements in 4 sets [51], with 30-15-15-15 repetitions, with 30–45 s [52], rest between sets and 3 min rest between movements. The LL-BFRT group allowed 45 s of rest between sets in the first eight training sessions, but reduced the rest period between sets to 30 s in the last six training sessions. In order to gradually increase the metabolic stress and the effect of training. AOP pressure was set at 30% for the first two training sessions, 40% for the third through fifth training sessions, and 50% for the remaining nine training sessions. For the LL-BFRT group, training intensity was performed at 30% of 1RM values. Following the eighth training session, the 1RM test was repeated to assess the participants’ potential neuromuscular adaptations and to evaluate any changes in 1RM values. The training intensity was recalculated based on the current value, and the remaining six training sessions were performed at these intensities [27, 28, 53]. The blood flow restriction pressure is reset for each training session.
The LL-BFRT group performed the seated dumbbell biceps curl exercise by completing the concentric contraction phase, during which the joint angle narrows, in 1 s, and the eccentric contraction phase, during which the joint angle widens, in 2 s. On the other hand, the standing triceps push-down exercise was performed by completing the eccentric contraction phase, during which the joint angle narrows, in 2 s, and the concentric contraction phase, during which the joint angle widens, in 1 s.
The 1RM test was conducted for the biceps curl and triceps pushdown exercises to evaluate maximal strength. The testing procedures followed standardised guidelines recommended by the American College of Sports Medicine (ACSM, 2009) and methodological descriptions by Schoenfeld (2010) and Campos et al. (2002) [12, 35, 36].
Before testing, participants performed a general warm-up (5 min of light cycling) followed by a specific warm-up consisting of 8–10 repetitions at ~ 50% of the estimated 1RM for each exercise [12, 36]. After a 2–3 min rest, participants performed 3–5 repetitions at ~ 70% of their estimated 1RM [12, 36]. Thereafter, the load was progressively increased until the participant could complete only one successful repetition with proper technique [12, 36]. Each attempt was separated by 3–5 min of rest [36].
A repetition was considered valid when the participant achieved the full range of motion with controlled tempo (1 s concentric, 2 s eccentric) and without compensatory movements [35, 36]. The maximum load lifted successfully was recorded as the 1RM [12, 36]. If a repetition was unsuccessful, the load was adjusted accordingly, and the test was repeated after sufficient recovery [36]. The 1RM test was re-administered to participants in both the LL-BFRT and HL-RT groups after the eighth training session to detect potential improvements in maximal strength and, if necessary, adjust training intensities accordingly. Although an eight-session interval may be relatively short for substantial strength gains, this re-assessment ensured accurate load prescription throughout the intervention.
The same experienced technician supervised all tests, and the dynamometer was calibrated prior to each testing session in accordance with the manufacturer’s guidelines, as recommended in standardised isokinetic assessment protocols. Spotters were present during all attempts to ensure participant safety. The pulley system for triceps pushdowns and dumbbells for biceps curls (Technogym, Cesena, Italy) were calibrated before testing.
Participants had at least 6 months of resistance training experience, ensuring adequate familiarity with the test procedures.
The isokinetic muscle strength of the right and left arm biceps brachii and triceps brachii muscles of the individuals participating in the research process was measured using an isokinetic dynamometer device manufactured by Germany-based Isomed2000 (D.&R. Ferstl GmbH). Participants were not familiar with the isokinetic device beforehand. Therefore, the week before the study began, participants were informed about the procedures and invited to the laboratory in pairs for measurements. The device was calibrated before each measurement [27]. Prior to measurement, the researcher applied physical warm-up exercises for 10 min in accordance with the warm-up protocol. Subsequently, the participant sat on the device and was secured at the upper body, shoulder and waist regions. After the participant’s details were entered into the device, the axis of the elbow joint was adjusted, and the upper arm was secured at the midpoint of the humerus bone. Measurements were performed in concentric-concentric mode at an angular velocity of 60^0^/s, with 0^0^ extension and 125^0^ flexion joint range of motion (ROM). The right arm was tested first, followed by a 5-minute rest period; then, the same protocol was applied to the left arm. The participant was verbally motivated by the researcher during the application. Before starting the measurement, participants performed a 5-repetition trial with submaximal strength. They then performed the movement concentrically-concentrically at an angular velocity of 60^0^/s, 0^0^ extension and 125^0^ flexion joint range of motion in 1 set of 5 repetitions, and the highest peak torque strength was recorded. Peak torque strengths for both arm biceps brachii and triceps brachii muscles were measured in ‘Nm’, peak torque strength angle in “0”, and relative strengths in ‘Nm/kg’.
An a priori sample size calculation power analysis was performed using G*Power (version 3.1), based on a two-way repeated-measures ANOVA design [39, 54, 55], with an assumed medium effect size (f = 0.25), an alpha level of 0.05, and a power of 0.80. This analysis indicated that a minimum of 18 participants would be required to detect a medium-sized group × time interaction effect.
The overall experimental design is illustrated in Fig. 4, and the participant flow diagram throughout the study is provided in Fig. 5, in accordance with CONSORT guidelines.
Fig. 4Experimental Design. Overview of the experimental design and study timeline showing the study procedures.
Fig. 5CONSORT flow diagram showing participant recruitment, allocation, follow-up, and analysis
Each participant in the groups continued to exercise twice a week for seven weeks. Measurements were taken at the Biomechanics Laboratory of the Faculty of Sports Sciences at Gazi University in Ankara.” The group allocation of the participants was randomly assigned by an independent researcher and subsequently communicated to them.” Both groups performed dumbbell curls for the biceps and triceps push-downs for the triceps, using the sets and repetitions specified in the training program. All measurements and exercises were conducted under the supervision of the researcher throughout the study. At the end of the study, the initial measurements were retaken and statistically evaluated.
Participants’ body weight and height were measured using a Seca 803 scale (precision of 0.1 kg/0.01 mm, Germany), while blood pressure and heart rate were measured with an Omron M2 Basic (HEM-7120-E) device (Netherlands). Saturation measurements were performed with a P2000 oxygen saturation device from pulse oximeter version 1.3 (Germany), and skinfold measurements were performed with a Holtan (0.2 mm precision, London) skinfold. Upper arm circumference measurements were taken with a Seca 201 ergonomic tape measure. Body fat percentage measurements were performed with a Tanita tape measure. The measurements were performed using a BC418 device (Japan). Isokinetic muscle strength was measured using a Germany-based “Isomed2000” (D.&R. Ferstl GmbH) isokinetic dynamometer. A “Smart Tools” digitally inflatable blood flow restriction device, model “Smart Cuff Elite (USA), was used to restrict blood flow.
All statistical analyses were performed using STATA 18 (StataCorp LLC, College Station, TX, USA). The assumptions of normality and homogeneity of variance were tested using the Shapiro–Wilk and Levene’s test (Brown–Forsythe, median), respectively [29, 56, 57]. All primary outcome variables met the assumptions required for parametric analysis (p > 0.05). Sphericity was assessed using Mauchly’s test where applicable; however, as all repeated-measures analyses involved only two time points (pre- and post), violations of sphericity were inherently limited. When necessary, Greenhouse–Geisser corrections were applied.
A two-way repeated measures ANOVA was applied to examine the main effects of time (pre- vs. post-test), group (LL-BFRT vs. HL-RT), and their interaction. When significant effects were detected, post hoc pairwise comparisons with Bonferroni correction were conducted [8, 58]. Effect sizes were calculated using partial eta squared (η²) for the ANOVA results and Cohen’s d for within-group (pre vs. post) and between-group (Δ change) comparisons. Effect sizes were interpreted as small (d = 0.2), moderate (d = 0.5), and large (d ≥ 0.8) [29, 55]. In addition, 95% confidence intervals (CI) and percentage changes from pre- to post-test were presentedPercentage changes from pre-to post-test were calculated using the [(post-test - pre-test) / pre-test] x 100 [59].
When analyses were repeated using ANCOVA adjusted for baseline age and body fat percentage, no statistically significant group effects were observed for any of the primary outcomes (all p > 0.05). The overall pattern of results was consistent with the repeated-measures ANOVA findings, indicating that baseline differences did not materially alter the observed outcomes.
Group × time interaction effects were consistently reported using F-values, exact p-values, and partial eta squared (η²) to ensure standardised statistical reporting throughout the manuscript.
Participant flow through the study is presented in Fig. 5. Of the 26 individuals assessed for eligibility, 22 met the inclusion criteria and were randomly allocated to the LL-BFRT (n = 12) or HL-RT (n = 10) groups. No participants were lost to follow-up, and all allocated participants were included in the final analysis.
This study was conducted with a total of 22 healthy participants who performed regular resistance exercise. Table 1 shows the descriptive characteristics of the participants. Despite randomisation, statistically significant baseline differences were found between the groups in age (LL-BFRT: 29.4 ± 8.1 years vs. HL-RT: 23.1 ± 4.3 years, p = 0.003) and body fat percentage (LL-BFRT: 19.9 ± 7.1% vs. HL-RT: 15.2 ± 4.7%, p = 0.016). These differences were considered potential confounding factors when interpreting the training outcomes.
Total work (Tables 2 and 3; Fig. 6). Both groups improved significantly over time (p < 0.01 for most time effects), with no group × time interactions (right F = 0.01, p = 0.935; left F = 1.85, p = 0.189). Notably, the magnitude of improvement favoured LL-BFRT in several parameters (e.g. left-arm total +16.3% vs. + 8.4%), with effect sizes in the small-to-moderate range, indicating potential practical relevance. Detailed descriptive data (Mean ± SD), percentage changes (Δ%), and Cohen’s d are provided in Table 3; see Fig. 6 for visualisation. Although no statistically significant group × time interactions were observed, the consistent within-group improvements and associated effect sizes suggest potentially meaningful training-related adaptations.
Table 2Two-way repeated measures ANOVA of elbow flexion and extension total work at 60°/sWorkExtremityTestLL-BFRTHL-RT95% CI Lower95% CI UpperTimeGroupTimex Group\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$ :\stackrel{-}{x}
\usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$ \:\stackrel{-}{x} $$\end{document}±ss **LL-BFRT** **HL-RT** **LL-BFRT** **HL-RT** **F(sd)** **Sig** Extension(N.m) **Right 60** ^**0**^ **/s** **Pre Test** 416.08 ± 139.68395.50 ± 102.32316.16330.49516.0460.51 **F(1.20)** 16.210.070.47 **P** < 0.01**0.7990.499 **Post Test** 447.16 ± 152.64439.40 ± 111.46337.97368.58556.35510.22 **Partial** \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$ \:{\boldsymbol{\eta\:}}^{2} $$\end{document} 0.4470.0030.023 **Left 60** ^**0**^ **/s** **Pre Test** 461.42 ± 149.31465.40 ± 111.44334.61394.59568.23536.21 **F(1.20)** 2.890.092.01 **P** 0.1040.7730.172 **Post Test** 508.83 ± 176.13469.70 ± 124.91382.33390.34634.83549.06 **Partial** \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$ \:{\boldsymbol{\eta\:}}^{2} $$\end{document} 0.1260.0040.091Flexion(N.m) **Right 60** ^**0**^ **/s** **Pre Test** 268.58 ± 145.08316.50 ± 134.13164.8231.28372.36401.72 **F(1.20)** 25.030.440.90 **P** < 0.01**0.5130.352 **Post Test** 315.16 ± 146.65346.20 ± 112.73210.25274.57420.07417.83 **Partial** \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$ \:{\boldsymbol{\eta\:}}^{2} $$\end{document} 0.5550.0210.043 **Left 60** ^**0**^ **/s** **Pre Test** 255.41 ± 136.63298.30 ± 128.15157.67216.88353.15379.72 **F(1.20)** 19.320.340.11 **P** < 0.01**0.5650.741 **Post Test** 325.08 ± 180.52358.10 ± 167.95195.94251.39454.22464.81 **Partial** \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$ \:{\boldsymbol{\eta\:}}^{2} $$\end{document} 0.4910.0160.005Total **Right 60** ^**0**^ **/s** **Pre Test** 684.67 ± 272.83710.00 ± 233.19489.5561.84879.84858.16 **F(1.20)** 37.760.050.01 **P** < 0.01**0.8270.935 **Post Test** 762.33 ± 292.32785.60 ± 215.98553.22648.37971.44922.83 **Partial** \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$ \:{\boldsymbol{\eta\:}}^{2} $$\end{document} 0.6530.0020.000 **Left 60** ^**0**^ **/s** **Pre Test** 716.83 ± 280.31763.70 ± 234.98516.31614.4917.35913.0 **F(1.20)** 21.590.031.85 **P** < 0.01**0.8680.189 **Post Test** 833.92 ± 351.81827.80 ± 254.72582.25665.961085.59989.64 **Partial** \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$ \:{\boldsymbol{\eta\:}}^{2} $$\end{document} 0.5190.0010.084Effect sizes are reported as partial eta squared (η²) for ANOVA main and interaction effects and interpreted as small (η² = 0.01), medium (η² = 0.06), and large (η² = 0.14). Cohen’s d values for within- and between-group comparisons are interpreted as small (d = 0.2), moderate (d = 0.5), and large (d ≥ 0.8)Additional ANCOVA analyses adjusting for baseline age and body fat percentage did not reveal any significant group effects (all *p* > 0.05)**p* < 0.05, ***p* < 0.01 Table 3Total work (Nm) for elbow flexion and extension at 60°/s in both groupsParameterGroupPre-test (Mean ± SD)Post-test (Mean ± SD)Δ Change (%)Effect size (Cohen’s d)Right ArmLL-BFRT684.67 ± 272.83762.33 ± 292.32+ 77,66 (+ 11.3%)0.65 (moderate)Right ArmHL-RT710.00 ± 233.19785.60 ± 215.98+ 75.6 (+ 10.7%)0.39 (small)Right Arm ExtensionLL-BFRT416.08 ± 139.68447.16 ± 152.64+ 31.08 (+ 7.5%)0.91 (large)Right Arm ExtensionHL-RT395.50 ± 102.32439.40 ± 111.46+ 43.90 (+ 11.1%)0.76 (moderate)Right Arm FlexionLL-BFRT268.58 ± 145.08315.16 ± 146.65+ 46.58 (+ 17.3%)-0.02 (trivial)Right Arm FlexionHL-RT316.50 ± 134.13346.20 ± 112.73+ 29.70 (+ 9.4%)0.81 (large)Left ArmLL-BFRT716.83 ± 280.31833.92 ± 351.81+ 117.09 (+ 16.3%)0.68 (moderate)Left ArmHL-RT763.70 ± 234.98827.80 ± 254.72+ 64.1 (+ 8.4%)0.36 (small)Left Arm ExtensionLL-BFRT461.42 ± 149.31508.83 ± 176.13+ 47.41 (+ 10.3%)0.26 (small)Left Arm ExtensionHL-RT465.40 ± 111.44469.70 ± 124.91+ 4.30 (+ 0.9%)0.90 (large)Left Arm FlexionLL-BFRT255.41 ± 136.63325.08 ± 180.52+ 69.67 (+ 27.3%)-0.16 (trivial)Left Arm FlexionHL-RT298.30 ± 128.15358.10 ± 167.95+ 59.80 (+ 20.0%)0.012(trivial) Fig. 6Bar graphs of total work for elbow flexion and extension at 60°/s. Bar graphs illustrating total work during elbow flexion and extension at 60°/s for the right and left arms in the LL-BFRT and HL-RT groups. Pre-test and post-test values are presented as mean ± standard deviation (SD). Error bars represent ± 1 SD. Group (LL-BFRT vs. HL-RT) and time (pre vs. post) conditions are indicated within each panel Peak torque (Tables 4 and 5; Fig. 7). No group × time interactions were observed (*p* > 0.05). Time effects indicated improvements in selected outcomes (e.g. right-arm F = 4.78, *p* = 0.041; left-arm F = 13.90, *p* = 0.001). Specifically, no statistically significant time × group interactions were observed for left-arm extension (HL-RT: *p* = 0.057; LL-BFRT: *p* = 0.078). Magnitude indices from Table 5 (Cohen’s d) suggest small-to-moderate practical changes across conditions; see Fig. 7 for pre-/post-distributions. Table 4Two-way repeated measures ANOVA of elbow flexion and extension peak torque at 60°/sPeak TorqueExtremityTestLL-BFRTHL-RT95% CI Lower95% CI UpperTimeGroupTime x Group\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$ \:\stackrel{-}{x} $$\end{document}± ss\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$ \:\stackrel{-}{x} $$\end{document}±ss **LL-BFRT** **HL-RT** **LL-BFRT** **HL-RT** **F(sd)** **Sig** Extension(N.m) **Right 60** ^**0**^ **/s** **Pre Test** 59.25 ± 22.3954.90 ± 15.6243.2344.9875.2764.82 **F(1.20)** 4.780.130.52 **P** 0.041*0.7210.478 **Post Test** 61.91 ± 21.9660.20 ± 18.4646.248.4777.6271.93 **Partial** \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$ \:{\boldsymbol{\eta\:}}^{2} $$\end{document} 0.1920.0060.025 **Left 60** ^**0**^ **/s** **Pre Test** 65.41 ± 20.6659.20 ± 14.6850.6349.8780.1968.53 **F(1.20)** 4.050.102.77 **P** 0.0570.7580.112 **Post Test** 66.16 ± 29.9567.10 ± 20.3144.7454.287.5880.0 **Partial** \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$ \:{\boldsymbol{\eta\:}}^{2} $$\end{document} 0.1680.0040.121Flexion(N.m) **Right 60** ^**0**^ **/s** **Pre Test** 43.41 ± 21.5242.80 ± 15.8128.0232.7558.852.85 **F(1.20)** 1.040.041.23 **P** 0.3190.8490.280 **Post Test** 43.25 ± 18.9946.8 ± 15.4129.6737.0156.8356.59 **Partial** \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$ \:{\boldsymbol{\eta\:}}^{2} $$\end{document} 0.0490.0010.057 **Left 60** ^**0**^ **/s** **Pre Test** 37.91 ± 17.0441.60 ± 16.4925.7231.1250.0152.08 **F(1.20)** 13.900.062.20 **P** 0.001**0.8130.154 **Post Test** 44.41 ± 21.3844.40 ± 16.6229.1233.8459.754.96 **Partial** \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$ \:{\boldsymbol{\eta\:}}^{2} $$\end{document} 0.4100.0020.099Effect sizes are reported as partial eta squared (η²) for ANOVA main and interaction effects and interpreted as small (η² = 0.01), medium (η² = 0.06), and large (η² = 0.14). Cohen’s d values for within- and between-group comparisons are interpreted as small (d = 0.2), moderate (d = 0.5), and large (d ≥ 0.8)Additional ANCOVA analyses adjusting for baseline age and body fat percentage did not reveal any significant group effects (all *p* > 0.05)**p* < 0.05, ***p* < 0.01 Table 5Peak torque (Nm) for elbow flexion and extension at 60°/s in both groupsParameterGroupPre-test (Mean ± SD)Post-test (Mean ± SD)Δ Change (%)Effect size (Cohen’s d)Right ExtensionLL-BFRT59.25 ± 22.3961.91 ± 21.96+ 4.5%1.24 (large)Right ExtensionHL-RT54.90 ± 15.6260.20 ± 18.46+ 9.6%0.73 (moderate)Left ExtensionLL-BFRT65.41 ± 20.6666.16 ± 29.95+ 1.1%1.39 (large)Left ExtensionHL-RT59.20 ± 14.6867.10 ± 20.31+ 13.4%0.80 (moderate)Right FlexionLL-BFRT43.41 ± 21.5243.25 ± 18.99–0.4%0.02 (trivial)Right FlexionHL-RT42.80 ± 15.8146.80 ± 15.41+ 9.3%0.25 (small)Left FlexionLL-BFRT37.91 ± 17.0444.41 ± 21.38+ 17.2%0.33 (small)Left FlexionHL-RT41.60 ± 16.4944.40 ± 16.62+ 6.7%0.18 (small)Both groups improved peak torque values. LL-BFRT showed large within-group effects for extension (d = 1.24–1.39), while HL-RT showed moderate effects (d = 0.73–0.80). Flexion improvements were smaller, with effects ranging from trivial to small Fig. 7Bar graphs of peak torque for elbow flexion and extension at 60°/s. Bar graphs illustrating peak torque during elbow flexion and extension at 60°/s for the right and left arms in the LL-BFRT and HL-RT groups. Pre-test and post-test values are presented as mean ± standard deviation (SD). Error bars represent ± 1 SD. Group (LL-BFRT vs. HL-RT) and time (pre vs. post) conditions are indicated within each panel Relative strength (Tables 6 and 7; Fig. 8) increased over time in both groups, with one statistically significant group × time interaction observed for left-arm extension (F = 6.72, *p* = 0.017). For all other relative strength outcomes, no statistically significant between-group or interaction effects were detected. Effect sizes ranged from small to moderate (Table 7). Figure 8 displays the pre- and post-test profiles. One-repetition maximum (Tables 8 and 9; Fig. 9). Both groups demonstrated significant within-group increases in 1RM from pre- to post-testing for biceps and triceps (time effects, *p* < 0.01), with no between-group or interaction effects. Effect sizes were moderate-to-large across exercises (Table 9). Figure 9 illustrates the magnitude of change. Table 6Two-way repeated measures ANOVA of elbow flexion and extension relative strength at 60°Relative StrengthExtremityTestLL-BFRTHL-RT95% CI Lower95% CI UpperTimeGroupTime x Group\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$ \:\stackrel{-}{x} $$\end{document}± ss\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$ \:\stackrel{-}{x} $$\end{document}±ss **LL-BFRT** **HL-RT** **LL-BFRT** **HL-RT** **F(sd)** **Sig** Extension (N.m/kg) **Right 60** ^**0**^ **/s** **Pre Test** 0.72 ± 0.240.69 ± 0.120.5460.6140.8920.766 **F(1.20)** 3.320.001.03 **P** 0.0830.9850.323 **Post Test** 0.73 ± 0.280.76 ± 0.130.530.6770.930.843 **Partial** \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$ \:{\boldsymbol{\eta\:}}^{2} $$\end{document} 0.1420.0000.048 **Left 60** ^**0**^ **/s** **Pre Test** 0.76 ± 0.210.72 ± 0.160.610.6180.910.822 **F(1.20)** 10.680.026.72 **P** 0.003*0.8970.017* **Post Test** 0.78 ± 0.240.85 ± 0.160.6080.7480.9520.952 **Partial** \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$ \:{\boldsymbol{\eta\:}}^{2} $$\end{document} 0.3480.0010.251Flexion (N.m/kg) **Right 60** ^**0**^ **/s** **Pre Test** 0.43 ± 0.190.53 ± 0.150.2940.4350.5660.625 **F(1.20)** 14.112.040.06 **P** 0.001*0.1680.804 **Post Test** 0.48 ± 0.180.59 ± 0.130.3510.5070.6090.673 **Partial** \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$ \:{\boldsymbol{\eta\:}}^{2} $$\end{document} 0.4130.0920.003 **Left 60** ^**0**^ **/s** **Pre Test** 0.42 ± 0.160.47 ± 0.180.3060.3560.5340.584 **F(1.20)** 12.570.191.09 **p** 0.002**0.6710.310 **Post Test** 0.50 ± 0.210.52 ± 0.160.350.4180.650.622 **Partial** \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$ \:{\boldsymbol{\eta\:}}^{2} $$\end{document} 0.3850.0090.051Effect sizes are reported as partial eta squared (η²) for ANOVA main and interaction effects and interpreted as small (η² = 0.01), medium (η² = 0.06), and large (η² = 0.14). Cohen’s d values for within- and between-group comparisons are interpreted as small (d = 0.2), moderate (d = 0.5), and large (d ≥ 0.8)Additional ANCOVA analyses adjusting for baseline age and body fat percentage did not reveal any significant group effects (all *p* > 0.05)**p* < 0.05, ***p* < 0.01 Table 7Relative strength (Nm/kg) for elbow flexion and extension at 60°/s in both groupsParameterGroupPre-test (Mean ± SD)Post-test (Mean ± SD)Δ Change (%)Effect size (Cohen’s d)Right ExtensionLL-BFRT0.72 ± 0.240.73 ± 0.28+ 1.4%0.22 (small)Right ExtensionHL-RT0.69 ± 0.120.76 ± 0.13+ 10.1%0.44 (small)Left ExtensionLL-BFRT0.76 ± 0.210.78 ± 0.24+ 2.6%0.20 (small)Left ExtensionHL-RT0.72 ± 0.160.85 ± 0.16+ 18.1%0.47 (moderate)Right FlexionLL-BFRT0.43 ± 0.190.48 ± 0.18+ 11.6%0.29 (small)Right FlexionHL-RT0.53 ± 0.150.59 ± 0.13+ 11.3%0.41 (small)Left FlexionLL-BFRT0.42 ± 0.160.50 ± 0.21+ 19.0%0.33 (small)Left FlexionHL-RT0.47 ± 0.180.52 ± 0.16+ 10.6%0.25 (small) Fig. 8Bar Graphs of relative strength for elbow flexion and extension at 60°/s. Bar graphs illustrating relative strength during elbow flexion and extension at 60°/s for the right and left arms in the LL-BFRT and HL-RT groups. Pre-test and post-test values are presented as mean ± standard deviation (SD). Error bars represent ± 1 SD. Group (LL-BFRT vs. HL-RT) and time (pre vs. post) conditions are indicated within each panel Table 8Two-way repeated measures ANOVA of 1RM (kg) for biceps and triceps1 RepetitionMaximum (kg)TestLL-BFRTHL-RT95% CI Lower95% CI UpperTimeGroupTime x Group\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$ \:\stackrel{-}{x} $$\end{document}± ss\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$ \:\stackrel{-}{x} $$\end{document}±ss **LL-BFRT** **HL-RT** **LL-BFRT** **HL-RT** **F(sd)** **Sig** Biceps **Pre Test** 18.29 ± 6.6918.22 ± 5.5513.514.6923.0821.75 **F(1.20)** 78.250.010.13 **P** < 0.01**0.9340.723 **Post Test** 22.23 ± 7.1821.85 ± 5.8817.0918.1127.3725.59 **Partial** \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$ \:{\boldsymbol{\eta\:}}^{2} $$\end{document} 0.7960.0000.006Triceps **Pre Test** 76.38 ± 27.2278.04 ± 25.3656.9161.9395.8594.15 **F(1.20)** 46.330.030.03 **p** < 0.01**0.8710.859 **Post Test** 98.16 ± 36.14101.01 ± 41.0572.3174.93124.01127.09 **Partial** \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$ \:{\boldsymbol{\eta\:}}^{2} $$\end{document} 0.6980.0130.001Effect sizes are reported as partial eta squared (η²) for ANOVA main and interaction effects and interpreted as small (η² = 0.01), medium (η² = 0.06), and large (η² = 0.14). Cohen’s d values for within- and between-group comparisons are interpreted as small (d = 0.2), moderate (d = 0.5), and large (d ≥ 0.8)Additional ANCOVA analyses adjusting for baseline age and body fat percentage did not reveal any significant group effects (all *p* > 0.05)**p* < 0.05, ***p* < 0.01 Table 91RM (kg) results for biceps curl and triceps pushdown in LL-BFRT and HL-RT groupsParameterGroupPre-test (Mean±SD)Post-test (Mean±SD)Δ Change (%)Effect size (Cohen’s d)1RM – Biceps (kg)LL-BFRT18.29 ± 6.6922.23 ± 7.18+ 21.5%0.58 (moderate)1RM – Biceps (kg)HL-RT18.22 ± 5.5521.85 ± 5.88+ 19.9%0.53 (moderate)1RM – Triceps (kg)LL-BFRT76.38 ± 27.2298.16 ± 36.14+ 28.5%0.81 (large)1RM – Triceps (kg)HL-RT78.04 ± 25.36101.01 ± 41.05+ 29.5%0.80 (large) Fig. 9Bar graphs of 1RM for elbow flexion and extension at 60°/s. Bar graphs illustrating one-repetition maximum (1RM) strength for biceps (elbow flexion) and triceps (elbow extension) exercises in the LL-BFRT and HL-RT groups. Pre-test and post-test values are presented as mean ± standard deviation (SD). Error bars represent ± 1 SD In Table 2, group differences were not statistically significant; nevertheless, both groups showed increases in total work output. Total work increased across both groups, with LL-BFRT showing greater relative improvements in flexion, whereas HL-RT demonstrated consistent gains in extension. Effect sizes ranged from small to moderate. The magnitude and variability of these changes are further illustrated by the corresponding 95% confidence intervals reported in Table 2. In Table 3, total work increased in both arms. For the right arm, LL-BFRT improved from 684.67 ± 272.83 to 762.33 ± 292.32 (d = 0.65, moderate) and HL-RT from 710.00 ± 233.19 to 785.60 ± 215.98 (d = 0.39, small). For the left arm, LL-BFRT rose from 716.83 ± 280.31 to 833.92 ± 351.81 (d = 0.68, moderate) and HL-RT from 763.70 ± 234.98 to 827.80 ± 254.72 (d = 0.36, small). In Table 4, group differences were generally not statistically significant; however, a significant improvement was observed in left elbow flexion according to time (*p* = 0.001), indicating that LL-BFRT elicited greater gains in this specific parameter. Both groups demonstrated notable improvements in peak torque values, with small to moderate effect sizes. LL-BFRT demonstrated consistent gains in both flexion and extension, while HL-RT produced slightly greater increases in extension tasks (Table 5). “A noteworthy improvement was observed in left arm extension peak torque in the HL-RT group, with post-test values showing a marked increase compared to pre-test (67.10 ± 20.31 vs. 59.20 ± 14.68, *p* = 0.057). Right arm extension increased from 59.25 ± 22.39 to 61.91 ± 21.96 in LL-BFRT (d = 1.24, large) and from 54.90 ± 15.62 to 60.20 ± 18.46 in HL-RT (d = 0.73, moderate). Left extension improved from 65.41 ± 20.66 to 66.16 ± 29.95 in LL-BFRT (d = 1.39, large) and from 59.20 ± 14.68 to 67.10 ± 20.31 in HL-RT (d = 0.80, moderate). Flexion changes were right flexion showed trivial-to-small effects (LL-BFRT d = 0.02; HL-RT d = 0.25), while left flexion improved modestly (LL-BFRT d = 0.33; HL-RT d = 0.18). Between-group effects were consistently small (d < 0.20). In Table 6, relative strength increased in both groups, with HL-RT showing more pronounced improvements in left arm extension. LL-BFRT also resulted in meaningful gains, particularly in flexion parameters. No statistically significant group differences were found; however, both LL-BFRT and HL-RT demonstrated improvements of similar magnitude, as reflected by the reported effect sizes. The observed changes and their associated uncertainty are reflected in the 95% confidence intervals presented in Table 6. In Table 7, both groups demonstrated small to moderate improvements in relative strength. For example, right extension increased by + 1.4% in LL-BFRT (d = 0.22, small) and by + 10.1% in HL-RT (d = 0.44, small). Left extension changes were + 2.6% (LL-BFRT, d = 0.20) and + 18.1% (HL-RT, d = 0.47). Flexion gains were also modest, with effect sizes ranging from small to moderate (d = 0.25–0.41). In Table 8, no statistically significant group differences were observed; therefore, both groups demonstrated meaningful within-group strength gains. Both LL-BFRT and HL-RT significantly improved 1RM performance in biceps and triceps exercises, with moderate to large effect sizes (d = 0.53–0.81). In Table 9, both LL-BFRT and HL-RT groups demonstrated significant improvements in one-repetition maximum (1RM) performance for both biceps curls and triceps pushdowns. The LL-BFRT group showed a 21.5% increase in biceps strength (d = 0.58, moderate) and a 28.5% increase in triceps strength (d = 0.81, large). Similarly, the HL-RT group achieved a 19.9% improvement in biceps strength (d = 0.53, moderate) and a 29.5% improvement in triceps strength (d = 0.80, large). These results indicate that both training methods are effective in enhancing maximal dynamic strength, with comparable gains across upper limb exercises. To account for baseline differences in age and body fat percentage between groups, additional analyses were performed using analysis of covariance (ANCOVA). After adjustment for baseline values of the corresponding outcome, age, and body fat percentage, no statistically significant group effects were observed for any of the primary outcomes (all *p* > 0.05). The overall pattern and direction of the results were consistent with those obtained from the repeated-measures ANOVA. A summary of pre–post changes, percentage improvements, confidence intervals, and effect sizes across all outcome measures is provided in Table 10. Table 10Summary of pre–post changes, 95% CIs, and effect sizesOutcomeGroupΔ Change (post–pre)% Change95% CIEffect size (Cohen’s d)Total Work – Right Arm (Nm)LL-BFRT+ 77.66+ 11.3%553.22–971.440.65 (moderate)HL-RT+ 75.60+ 10.7%648.37–922.830.39 (small)Total Work – Left Arm (Nm)LL-BFRT+ 117.09+ 16.3%582.25–1085.590.68 (moderate)HL-RT+ 64.10+ 8.4%665.96–989.640.36 (small)Peak Torque – Right Extension (Nm)LL-BFRT+ 2.66+ 4.5%46.20–77.621.24 (large)HL-RT+ 5.30+ 9.6%48.47–71.930.73 (moderate)Peak Torque – Left Extension (Nm)LL-BFRT+ 0.75+ 1.1%44.74–87.581.39 (large)HL-RT+ 7.90+ 13.4%54.20–80.000.80 (moderate)Peak Torque – Right Flexion (Nm)LL-BFRT−0.16−0.4%29.67–56.830.02 (trivial)HL-RT+ 4.00+ 9.3%37.01–56.590.25 (small)Peak Torque – Left Flexion (Nm)LL-BFRT+ 6.50+ 17.2%29.12–59.700.33 (small)HL-RT+ 2.80+ 6.7%33.84–54.960.18 (small)Relative Strength – Right Extension (Nm/kg)LL-BFRT+ 0.01+ 1.4%0.53–0.930.22 (small)HL-RT+ 0.07+ 10.1%0.68–0.840.44 (small)Relative Strength – Left Extension (Nm/kg)LL-BFRT+ 0.02+ 2.6%0.61–0.950.20 (small)HL-RT+ 0.13+ 18.1%0.75–0.950.47 (moderate)Relative Strength – Right Flexion (Nm/kg)LL-BFRT+ 0.05+ 11.6%0.35–0.610.29 (small)HL-RT+ 0.06+ 11.3%0.51–0.670.41 (small)Relative Strength – Left Flexion (Nm/kg)LL-BFRT+ 0.08+ 19.0%0.35–0.650.33 (small)HL-RT+ 0.05+ 10.6%0.42–0.620.25 (small)1RM – Biceps (kg)LL-BFRT+ 3.94+ 21.5%17.09–27.370.58 (moderate)HL-RT+ 3.63+ 19.9%18.11–25.590.53 (moderate)1RM – Triceps (kg)LL-BFRT+ 21.78+ 28.5%72.31–124.010.81 (large)HL-RT+ 22.97+ 29.5%74.93–127.090.80 (large)Values are presented as mean change (post–pre), percentage change, 95% confidence intervals (CI), and Cohen’s d effect sizes. Effect sizes were interpreted as small (d = 0.2), moderate (d = 0.5), and large (d ≥ 0.8) ## Dıscussıon The present study examined the effects of low-load blood flow restriction training (LL-BFRT) and high-load resistance training (HL-RT) on upper-limb strength outcomes. Both training modalities resulted in significant improvements in isokinetic and dynamic strength parameters; however, most group × time interactions were not statistically significant. These findings should be interpreted with caution, as the study was conducted with a relatively small sample size, which may have limited statistical power to detect small-to-moderate between-group differences. In addition, despite randomisation, significant baseline differences in age and body fat percentage were observed between groups, which may have influenced neuromuscular adaptation, recovery capacity, and training responsiveness. Although additional analyses adjusted for these baseline variables did not materially alter the overall pattern of results, residual confounding cannot be fully excluded. Therefore, the observed non-significant interaction effects should be considered preliminary rather than indicative of equivalence between LL-BFRT and HL-RT. Upper-limb muscles may respond differently to blood flow restriction compared with lower limbs due to anatomical, vascular, and functional differences, as well as the relatively limited evidence base for upper-extremity BFR applications [41]. Smaller muscle mass, distinct loading patterns, and limb-specific vascular and occlusion characteristics are likely to influence the balance between metabolic stress– and mechanical load–driven adaptations in the upper extremities [4, 9, 23, 25, 39]. Additionally, methodological factors such as cuff design and arm-specific pressure application may further contribute to variability in upper-limb responses to BFR [52]. Despite these considerations, emerging evidence suggests that LL-BFRT can induce meaningful strength adaptations in the upper limbs when appropriately applied, supporting its use as a viable alternative to high-load training [8, 20–22, 41, 60]. Although the group × time interaction effects were largely non-significant, the magnitude of improvements frequently favoured one training protocol over the other, depending on the parameter assessed. In particular, the LL-BFRT group demonstrated greater percentage increases and larger effect sizes in several measures, suggesting meaningful practical adaptations despite the absence of statistical interactions. These results indicate that LL-BFRT may elicit comparable or, in some metrics, even superior responses to HL-RT when applied under controlled conditions. It is important to note that a substantial proportion of the existing LL-BFRT literature is based on lower-extremity training protocols, particularly involving large muscle groups such as the quadriceps. Upper-limb muscles differ in muscle mass, vascular structure, occlusion pressure requirements, and fatigue characteristics, which may influence the magnitude and nature of training adaptations. Therefore, findings derived from lower-limb LL-BFRT studies should be interpreted with caution when applied to upper-extremity performance. The present results contribute to this limited body of evidence by specifically demonstrating how LL-BFRT affects upper-limb isokinetic strength and work-related outcomes, which may not fully mirror adaptations observed in the lower extremities. ### Total work In the study, the total work capacity of the right arm increased by 16.3% in the LL-BFRT group, while an 8.4% increase was observed in the HL-RT group. This finding is consistent with studies by Scott et al. [4] and Takarada et al. [17, 61], which reported that low-intensity LL-BFRT can increase total work capacity by inducing metabolic stress. Furthermore, Loenneke et al. [8] emphasised that LL-BFRT can also provide significant increases in work production with low loads. However, no statistically significant difference was found between the two groups in our current study. Although the η² values are moderate, and the absolute difference appears small, these gains may be practically meaningful, especially in individuals with limited tolerance to high loads. While no significant group-by-time interactions were found, LL-BFRT demonstrated notably greater percentage improvements in total work and flexion peak torque than HL-RT. The consistent direction of these changes, together with moderate effect sizes, suggests that LL-BFRT may elicit a distinct adaptive response that may be relevant in applied settings for certain performance variables. Such differences, although not statistically significant, could be clinically important for populations unable to tolerate high mechanical loads, such as in rehabilitation or early-phase strength programs. ### Peak torque Although the group × time interaction for peak torque was not statistically significant, the LL-BFRT group still exhibited larger percentage improvements in several torque measures (e.g., left-arm flexion + 17% vs. +6% in HL-RT), indicating greater percentage improvements, although these differences did not reach statistical significance. A 4% increase was observed in right arm extension in the LL-BFRT group. Pope (2013) and Loenneke (2017) reported that LL-BFRT can increase the muscle’s maximal power production capacity despite being applied with low loads [5, 62]. However, Schoenfeld [63] and Rong et al. [53] has reported that HL-RT may be more advantageous in peak torque and maximal strength gains. Enkeleda et al. [64] reported that LL-BFRT can produce effects similar to HL-RT on jump strength, while McKee et al. [65] noted that it improves sprint performance. In contrast, Fostiak et al. [66] reported no significant change in peak torque. These findings suggest that peak torque increase can be observed in both LL-BFRT and HL-RT, but that mechanical stress (HL-RT) and metabolic stress (LL-BFRT) processes may exert their effects through different mechanisms. Furthermore, the fact that most η² values are small to moderate indicates that changes in peak torque may be modest from a clinical/practical standpoint but still noteworthy. ### Mechanistic perspective The greater relative improvements observed in several LL-BFRT parameters, despite the substantially lower mechanical load, appear to be closely linked to the distinct physiological stimuli elicited by blood flow restriction. In the present study, LL-BFRT frequently demonstrated larger percentage changes in outcomes such as total work and relative strength, which are performance variables strongly influenced by metabolic fatigue resistance and sustained force production capacity. These findings are consistent with the pronounced metabolic stress induced by LL-BFRT through the accumulation of metabolites such as lactate, hydrogen ions, and inorganic phosphate, which can enhance muscle fibre recruitment, particularly of type II fibres typically activated under high-load conditions [4, 8, 9]. This metabolically demanding environment stimulates anabolic signalling pathways, including mTOR, MAPK, and HIF-1α activation, and promotes elevated secretion of growth hormone and IGF-1, collectively accelerating muscle protein synthesis [10, 61, 67]. In addition, the hypoxic conditions associated with BFR may increase motor unit firing frequency and activate previously dormant satellite cells, further facilitating hypertrophy and neuromuscular adaptations [9, 13, 68]. These mechanisms provide a physiological basis for the observation that LL-BFRT elicited meaningful improvements in several performance metrics despite the lower external loads applied. In contrast, high-load resistance training primarily relies on mechanical tension as the dominant stimulus for neuromuscular adaptation. While HL-RT is traditionally associated with superior maximal strength gains, the absence of significant between-group differences in peak torque and 1RM outcomes in the present study suggests that, over the relatively short intervention period, metabolic stress–driven adaptations induced by LL-BFRT may have compensated for the lower mechanical stimulus. Accordingly, the present findings support the notion that LL-BFRT and HL-RT may promote strength adaptations through partially distinct but converging physiological pathways, with metabolic stress playing a more prominent role in LL-BFRT and mechanical tension remaining central to HL-RT [8, 10]. ### Relative strength Relative strength is a critical indicator for performance athletes because it is the ratio of muscle strength to body weight. In this study, a 19.04% increase in left arm flexion relative strength was observed in the LL-BFRT group, but the difference between the groups was not statistically significant. These findings are consistent with the results of meta-analyses conducted by Lixandrao et al. [27] and Centner et al. [10]. Most studies examining the increase in strength between the LL-BFRT and HL-RT groups report similar strength gains in both groups. However, some studies suggest that HL-RT provides a higher increase, while others suggest that LL-BFRT provides greater gains [29]. Loenneke et al. [8] reported that LL-BFRT can also provide strength gains with low loads, while Yasuda et al. [60] reported significant hypertrophy increases in short-term upper extremity applications. However, it is noted that HL-RT is often advantageous in terms of absolute strength gains [63], but LL-BFRT can approach HL-RT when individualised pressure, session number, and cuff protocol are optimised [22]. The η² values were particularly moderate in left arm flexion, and this change may be considered meaningful in terms of athletic performance. ### 1RM strength Significant increases in biceps and triceps 1RM values were recorded in both the LL-BFRT and HL-RT groups. A 21.5% increase for biceps and a 28.5% increase for triceps were reported in the LL-BFRT group. These findings are consistent with the results reported by Takarada et al. [61] and Centner et al. [10, 17]. However, no significant differences were found between the groups. While studies in the literature have found that LL-BFRT and HL-RT training produce similar results, there are also studies indicating that HL-RT training yields greater gains [29, 69]. Additionally, Loenneke et al. [8] emphasised that low-intensity LL-BFRT can provide comparable strength and hypertrophy gains to high-intensity HL-RT, while Yasuda et al. [60] reported that short-term LL-BFRT applications in upper extremity muscles contribute significantly to hypertrophy. However, the literature generally indicates that HL-RT is superior in terms of maximal strength gains [63], but as noted in the studies by Chang et al. [22] and Schoenfeld (2017), LL-BFRT can approach these gains with appropriate protocols. The findings of this study also reveal that LL-BFRT can produce results close to HL-RT in terms of strength development rate, but there is no significant difference between the groups. Most η² values fall within the small to moderate range, and it should be emphasised that these increases in strength may be functionally meaningful, particularly in rehabilitation and clinical populations. Therefore, these results are preliminary in nature, and it is inappropriate to draw definitive conclusions about superiority based on the results of a single small-sample study. It is important to acknowledge that the two groups differed significantly in baseline age and body fat percentage. Such disparities may have influenced the magnitude and rate of physiological adaptation, since both age and body composition can affect recovery capacity, hormonal responses, and metabolic efficiency during resistance training. Although randomisation was applied, these baseline differences might partly explain the variability observed in strength gains. Future studies with larger and more homogenous samples are recommended to confirm these findings and control for potential confounding effects. ### Limitations and future research The limitations of this study include the small sample size (n = 22), short follow-up period (7 weeks), and limited generalizability of the results. Although an a priori power analysis suggested that the sample size would be adequate for detecting medium effects, the study was not sufficiently powered to detect smaller between-group differences. This limitation may partially explain the non-significant interaction effects and highlights the need for larger controlled trials. In addition, significant baseline differences in age and body fat percentage between groups may have confounded the training responses, further limiting the ability to draw strong causal inferences. Although baseline differences in age and body fat percentage were statistically controlled using ANCOVA, residual confounding cannot be entirely excluded due to the small sample size. These baseline differences are relevant because older individuals generally demonstrate slower recovery capacity and altered hormonal responses compared to younger adults, which may influence neuromuscular adaptations and total work capacity. Similarly, individuals with higher body fat percentages may experience reduced metabolic efficiency and differential training-induced adaptations. These factors should therefore be considered when interpreting the comparative outcomes between LL-BFRT and HL-RT groups. Another limitation of the present study is the lack of blinding of participants and outcome assessors. Due to the nature of the exercise interventions, blinding of participants to group allocation was not feasible. Similarly, assessors were not blinded during outcome measurements, which may have introduced a potential risk of measurement bias. However, standardized testing procedures and objective outcome measures were used to minimize this risk. Furthermore, body composition measurements using gold-standard methods such as dual-energy X-ray absorptiometry (DXA) were not performed, which limits the interpretation of muscle hypertrophy–related adaptations beyond strength outcomes. Similar limitations have been noted in the literature, and studies with small samples are frequently reported [8, 26, 70–72]. Additionally, most of the η² values obtained in this study indicate that small percentage differences may be statistically limited but could still be meaningful in clinical and practical contexts. Another limitation that should be acknowledged is the potential for learning or familiarisation effects associated with repeated isokinetic testing. Although standardised testing and familiarisation procedures were applied and both groups were equally exposed to the testing protocol, repeated exposure to isokinetic assessments may have contributed to performance improvements independent of training-induced adaptations. In addition, the study sample consisted of relatively young, resistance-trained male participants, which limits the generalisability of the findings to other populations, such as females, older adults, or clinical groups. Future studies including more heterogeneous samples are therefore warranted to confirm the applicability of these findings across different populations. Future studies should include larger samples, long-term follow-ups, and data on body composition. Additionally, designs that standardise protocol elements such as individualised AOP (50–80% AOP range), pressure reduction between sets, 2–3 sessions per week, and 20–40% 1RM loading, and include longer follow-up periods are recommended. Additionally, the absence of statistically significant group differences should be interpreted in the context of the small sample size; the observed effect sizes indicate that clinically meaningful adaptations may have been masked by limited statistical power. ## Conclusion This study extends the existing literature by examining the effects of low-load blood flow restriction resistance training (LL-BFRT) on upper-limb isokinetic strength and dynamic strength outcomes. The findings indicated that LL-BFRT can elicit meaningful within-group improvements in strength-related parameters and may serve as a viable training alternative when high-load resistance training (HL-RT) is not feasible, particularly in rehabilitation or load-compromised populations. Although improvements were observed across multiple outcome measures, most between-group comparisons did not reach statistical significance, and therefore, no between-group conclusions can be drawn. While HL-RT remains the preferred approach for maximal strength development, LL-BFRT may offer a practical option in situations where high mechanical loading cannot be safely or effectively applied. Taken together, the present findings suggest that LL-BFRT may be incorporated as a complementary or alternative strategy under specific conditions; however, the results should be interpreted with caution due to the relatively small sample size and short intervention duration. Future studies with larger and more diverse samples, longer follow-up periods, and comprehensive assessments of muscle morphology are required to confirm these findings and to further clarify the role of LL-BFRT in upper-limb strength training. Although between-group statistical significance was not observed, improvements were numerically greater in the LL-BFRT group across several parameters. However, these effect size estimates and percentage changes should be interpreted with caution, as the absence of statistically significant group × time interactions does not allow confirmation of between-group superiority. This pattern suggests that LL-BFRT may induce meaningful practical gains under certain conditions, without allowing conclusions regarding between-group superiority. Although effect sizes were reported to describe the magnitude of changes, their clinical or practical relevance should be interpreted cautiously and within the context of the study design, sample characteristics, and outcome variability. In conclusion, the present findings support the potential role of blood flow restriction exercise as part of rehabilitation and muscle strength–building programmes when high-load resistance training is not feasible. Nevertheless, prospective studies should evaluate the long-term effects of this method and its applicability to different populations.