Blood flow restriction resistance training as an alternative to resistance training alone to improve strength in elderly: a systematic review with meta-analysis

Research registration

We conducted this systematic review with meta-analysis in accordance with PRISMA statement (see Supplemental Digital Content 1 that is a table with PRISMA check list) [17]. Search commission of Federal University of Health Sciences from Porto Alegre registration number: 084/2020. PROSPERO registration number: CRD42020220729.

Data sources

The electronic databases used were MEDLINE (PubMed), Web of Science, Cochrane Library, and EMBASE in February 2022. We used a comprehensive search strategy tailored to each database. In cases of missing data, authors of selected papers were contacted. When contacted authors did not answer, data were extrapolated from figures, using Image J software, or using available data and mathematics formulas provided by Cochrane’s handbook as presented at section 2.4 [19].

Search strategy

For identification of relevant studies, a systematic literature search was performed by two blinded researchers (Mallmann, ALS & Doria, LD). Keywords and medical subject headings (MeSH) for the terms “Blood flow restriction”, “Kaatsu”; “Vascular occlusion”, “Strength training”, “Resistance Training”, “Muscle strength”, “Elderly”, “Older” and “Aging” were selected. No filters were used to perform this search. The term OR was used for Union of MeSH terms and “entry terms”, and the term AND was used to attach the terms. The complete string used at PUBMED and adapted for the other databases is fully described at Supplemental Digital Content 2. Study information, including title and abstract, were exported from the databases and stored in a citation manager (Mendeley1, version1.17.9). Before further processing of the studies all duplicates were removed.

Inclusion and exclusion criteria

We included: (1) Participants were older than 60 years, (2) the study design allowed to compare resistance training combined with blood flow restriction to traditional resistance training and/or to a control group (without any intervention or placebo training, such as light stretching), and (3) strength were assessed pre- and post-intervention. No restriction on publication date was imposed.

We excluded: (1) participants received any kind of substance that could interfere on study results, (2) the manuscript was not written in English, Portuguese, or Spanish languages or (3) meta-analysis articles.

Tables with inclusion/exclusion decisions by reviewer and agreements can be accessed at Supplemental Digital Content 3

Data extraction and assessment of reviewer agreement

First, two researchers (Mallmann, ALS and Doria, LD) screened, independently, all the studies titles, then abstracts and, finally, the full text of the included papers. This process was made in three steps and the reviewers were blinded about the coworker screening. In case of any discrepancy, a third reviewer (Dos Santos, LP) were asked to find an agreement and decide to include or not the paper. The sheets of inclusion/exclusion process for Mallmann, ALS and Doria, LD is available at Supplementary material 2. All data from each study were screened using a bibliographic management program (Mendeley1, version1.17.9).

After screening of the studies, all relevant considered articles were assessed for eligibility based on their full texts. At this stage, we extracted information about (1) population characteristics, (2) primary outcome measures, (3) methods, (4) exercise/interventional characteristics and (5) the main result of the study. When intervention effects were assessed at multiple time points, only the very last time point was considered (as post-training value). When available, data were extracted in the form of delta mean (mean_change), delta standard deviation (SD_change), and sample size of the studies to perform the meta-analysis. In case of incomplete raw data availability, we contacted the corresponding author of the manuscript or extrapolated the data from figures, if the authors could not be reached. When the article reported baseline and post-intervention outcomes, however, without mean_change and SD_change, we used the equation (Delta mean = post-training mean– baseline mean) to calculate the delta value. Considering cochrane’s handbook recommendations to calculate the SD_change, we used the correlation equals zero, since none of the selected papers provided the delta data as mean ± pattern deviation [19]. In order to find SD_changefor selected studies, the following formula was used. Where Corr is correlation coefficient in the experimental group, SD_E,baselineis baseline standard deviation in the experimental group, SD_E,final is final standard deviation in the experimental group and SD_E,change is standard deviation of the changes in the experimental group. When data were presented by interquartile range (IQR), it was decided to transform these data in order to standardize the results of all studies in mean_change and SD_change. The equation used to calculate the mean_change is available below [20]. Where q1 is the first quartile, m is the median and q3 is the third quartile. Finally, to find the SD_change presented by IQR, we use the calculation available below [20]. The choice for using these formulas was based on a previous Systematic Review with meta-analysis about effects of BFR training on strength, hypertrophy, and functionality for people with osteoarthritis and rheumatoid arthritis [21].

The extracted data of included studies are sample characteristics (number of participants, age, Body Mass Index (BMI), Duration of the intervention (weeks), frequency (sessions per week), sets, repetitions, interval (seconds), one maximum repetition percentual, blood flow restriction pressure (BFR mmHg) and the delta strength, resulting from interventions).

Methodological quality assessment

Methodological quality of reports was determined using the Grading of Recommendations Assessment, Development and Evaluation (GRADE). The peer reviewed analysis is available at supplementary material 4. For each of the 7 items of the GRADE scale, two reviewers (Mallmann, ALS and Doria, LD) assessed the studies independently. Disagreements about methodological quality were resolved by a third reviewer (Dos Santos, LP). The GRADE approach considers the risk of bias and the body of evidence to rate the certainty of the evidence into one of four levels:

Risk of bias

The risk of bias of the studies was assessed using the risk of bias tool 2.0 (RoB2) from Cochrane [22]. Two authors (Mallmann, ALS and Doria, LD) independently assessed the risk of bias. In the case of disagreement, the subject was discussed with another author (Dos Santos, LP). The evaluators analyzed the randomization process, deviations from intended interventions, missing outcome data, measurement of the outcome, and selection of the reported results. The studies were classified into low, moderate, or high risk of bias.

Search strategy

The searches were performed in the five databases from beginning to 2022 and returned 6,956 (380 duplicates) (Fig 1). After removing duplicates, reading the titles, abstracts and full texts, eight studies, between 2013 and 2020, were kept for analysis. Six of them compared BFR training to a non-training group (CON) [25-30] and six of them compared BFR training to a group trained with traditional moderate or high intensity RT [25,27,29,31,32].

• Download figure
• Open in new tab

Fig 1.

Flow chart including literature search and selection steps following PRISMA statement.

Meta-analysis data

Fig 2(A) shows the comparison of BFR with RT for selected papers. Only studies with similar methodologies were included in this analysis, minimizing the risk of biasing findings. Studies ere included in our meta-analysis if the assessments of strength were made preferentially with bilateral knee extension (KE) exercise. We also included, for this meta-analysis, papers that evaluated participants with Leg press and unilateral KE exercises. Although, a sensibility analysis, represented by figure 2(B), was made by excluding the study from Letieri et al, 2018 because the training was slightly different from the others for the use of unilateral KE, whereas the other authors compared bilateral KE for both groups. Nevertheless, there were no significative differences between training methodologies, even without considering said study [27].

• Download figure
• Open in new tab

Fig 2.

Forest plot graph of the strength of elderly trained with BFR versus Resistance training. A: comparison for all studies with similar methodology including unilateral and bilateral KE; B: sensibility analysis excluding the paper from Letieri et al, 2018 that used unilateral knee extension; KE: knee extension; BFR: blood flow restriction;I²:inconsistency test (heterogeneity); SD: standard deviation; Std: Standardized, 95% CI: confidence interval, IV: inverse variance, Random: random effects model

Fig 3 shows the comparison between BFR and CON and its sensibility analysis. Figure 3 (A) is the overall comparison between all studies that included groups without any intervention between assessments or applied placebo training (e.g., light stretch training). Figure 3 (B, C and D) illustrates the sensibility analysis we made to minimize the bias of that comparison. Were excluded papers with methodological differences that could interfere on results. First (B) we excluded studies from Letieri et al, 2018 and Vechin et al, 2015 for using unilateral KE and Leg Press (LP), respectively. Second (C) we reintroduced only the study from Vechin et al, 2015 and at least we excluded this one again and reintroduced that from Letieri et al, 2018. Although, the difference between groups remained significative in favor of BFR training.

• Download figure
• Open in new tab

Fig 3.

Forest plot graph of the strength of elderly trained with BFR versus Control group (without training or placebo training).A: comparison for all studies with similar methodology including LP, unilateral and bilateral KE; B: sensibility analysis excluding the papers that used unilateral KE and LP; C: sensibility analysis excluding only the study from Letieri et al, 2018 that used unilateral KE; D: sensibility analysis excluding only the article from Vechin et al, 2015 that used LP; KE: knee extension; BFR: blood flow restriction; I²:inconsistency test (heterogeneity); SD: standard deviation; Std: Standardized, 95% CI: confidence interval, IV: inverse variance, Random: random effects model.

Risk of Bias

All selected studies were analyzed for their risk of bias according to the GRADE Approach tool, by Cochrane Collaboration. The data presented were analyzed using RoB 2 software.

Among the analyzed fields, shown in Figure 5, the greatest risk of bias was found in the random sequence generation field, with one study that did not present such variable (12%); at the field about measurement of the outcome, most studies (75%) were classified as “some concerns” for risk of bias, because the methods for assess the outcomes were not the very best way to do that. All selected studies were included in the systematic review, independently of quality assessment results.

For more details about risk of bias, see figs 4 and 5 (for complete peer reviewed risk of bias see the figure at Supplemental Digital Content 4).

• Download figure
• Open in new tab

Fig 4.

Risk of bias graph considering all studies pooled.

• Download figure
• Open in new tab

Fig 5.

Risk of bias for individual studies and according to the different criteria assessed.

Quality assessment

Overall studies quality assessment show that the evidence present low quality, in accordance with GRADE tool. The lowest quality shown is for risk of bias domain since a study presented high risk of bias for randomization process. For complete data, see table at Supplemental Digital Content 5 with each reviewer quality assessment.

Studies Data

In table 1 the characteristics of sample are presented. It is possible to notice that are homogeneity between groups in each study in age and body mass index (BMI). The mean age across studies was 69,2±5,4, and the mean BMI 26,2±2,8. The total sample analyzed was 232 older (≥ 60 years old) men (76) and women.

Table 2 shows that there is no pattern in the duration of the training period (6 to 16 weeks), in the number of repetitions (from 10 to failure) or in the BFR pressure used (from 71 to 270 mmHg) for the included papers. However, it is possible to observe that the number of sets (2 to 4) and the interval between them (30 to 60 seconds) followed the ACSM recommendation for resistance training [4]. It is also possible to see that for resistance training using the BFR resource, the percentages of a maximum repetition (1RM) were lower (20-30% 1RM) than those of traditional resistance models (70-80% 1RM).

Short training protocols describing for each study are presented at Supplemental Digital Content 6

Blood flow restriction vs Resistance training alone

Our results show, by meta-analysis (Figure 2), that there are no statistically significant differences between groups trained with BFR or RT alone and these findings are in consistency with previous literature [33].

In some cases, BFR showed bigger strength improvements than RT alone (Table 3), while in other studies the results from RT were better than BFR for strength improvement and these cases must be analyzed. Study from Letieri et al, (2018) show results favorable to BFR, when compared to RT and the study from Bigdeli et al., (2020) show results favorable to BFR only for leg extension exercise [25, 27]. In the other hand, in the studies from Cook et al., (2017) and from Vechin et al. (2015), RT was more effective than BFR for strength gains [26,28]. Differences between studies that could explain why BFR is better in some papers, as the study from Letieri et al, (2018), remains on AOP, training intensity (1RM%), number of repetitions per set, also known as training volume, weekly frequency, between sets intervals [4].

First, Letieri et al. (2018) show the best results in favor of BFR and, not coincidently, this is the study with biggest duration, between all selected papers for this review [27]. In line with that, the study from Loenneke et al., (2012) show that BFR present better results for strength improvement, with a bigger effect size, when intervention has longer duration, with ten or more weeks from baseline to post-intervention assessments, and this fact could explain our findings at the studies from Bigdeli et al., (2020), Cook et al., (2017) and Vechinet al., (2015), because of the short duration of this studies [25, 26, 28, 33]. It is important to note that the study from Bigdeli et al., (2020) was conducted for only 6 weeks, while the studies from Cook et al., (2017) and Vechin et al., (2015) lasted for 12 weeks, that is, only 2 weeks more than the study from Loenneke et al., (2012) found as a time cut point for better strength improvements resulting from BFR training [25, 26, 28, 33]. Another note that deserves our attention is the fact that most of studies about BFR training lasted from 4 to 12 weeks of training, including all kinds of population and this is a reason to difficult a better understanding about this training modality and its impacts on strength due to its short durations.

Although BFR presents less neural drive, our findings show that there is no difference in strength gains between this methodology and RT. A systematic review with meta-analysis from Centner and Benedikt, (2020) found inconclusive results comparing muscle activation between low load BFR training (LL-BFR) and HL-RT, but most of the studies included in this paper lasted for less than 10 weeks, that is the time that, as seen, when BFR starts to show bigger improvement on muscle strength compared to RT [34].

A study from 2021 [35] compared two groups with a repeated measures cross-over design intervention, performed in two sessions of isometric exercises of the knee extensors on two separate days in random and counterbalanced order. The low load group (LL) trained with 20% of the maximum voluntary contraction (MVC) and the low intensity group with blood flow restriction (LL-BFR) trained at 20% of the (MVC), but with 50% of its arterial occlusion pressure (AOP), which was determined in a personalized way before the beginning of the intervention. The findings of this study suggests that BFR is better than traditional RT with low loads for strength improvements and this is an interesting line of thinking, keeping in mind that people with older ages may not be able to accomplish sets of RT with optimal loads for strength gains [35].

We hypothesized that, due to the typical physical limitations of the population within this age, the BFR training could generate gains similar or even superior to the RT since, theoretically, this population would not be able to displace the same percentages of 1RM as a young population. Therefore, they would end up training with low intensity resistance training (LL-RT), a technique that seems to be less efficient, according to the study by Hughes et al., (2017), than BFR training for strength gains [14]. In this 2017 review, the authors sought studies that compared BFR to RT for strength gains in patients with clinical musculoskeletal condition. The research strategy included studies dealt with patients with osteoarthritis of the knees (n=3), ligament injuries (n=3), sporadic inclusion body myositis (n=1) and older adults susceptible to sarcopenia (n=13), corroborating our hypothesis that the older population, which, as stated, has a greater inability to perform training with higher intensities [14].

Although our hypothesis considered that in some cases this population, over 60 years of age, could not perform strength exercises with high loads, our search strategy, which considered medically stable elderly individuals, returned studies that compared BFR training to high intensity RT. And the results of the comparison, performed in this review (Figure 2), suggests that strength gains may be similar to moderate and high intensity RT.

If the neural drive, as mentioned, is smaller with the application of BFR training than with RT, it is important to try to understand how BFR brings strength improvements similar to RT. The literature has shown the physiological mechanisms that could explain these gains and, thus, enabling new hypotheses for future research with greater direction regarding BFR training to control the loss of strength in the elderly [35,36]. Studies show that there appears to be a strong relationship between the metabolic response, arising from BFR training, and gains, both in strength and muscle hypertrophy. The study by Loenneke et al., (2010) was the first to clarifies that the main mechanisms related to gains arising from BFR are apparently the accumulation of metabolites, such as blood lactate, plasma lactate and muscle cell lactate, in addition to being related to activation of fast twitch muscle fibers (FT), even when training intensity is low, and last but not least, gains appear to be related to increased expression of mammalian target of rampamycin (m-TOR) [36].

Blood flow restriction vs non-training

A meta-analysis comparing a BFR group with a control group was performed. The result of this analysis suggests that BFR training can be an effective alternative for strength gain in people over 60 years, since we found a significative difference between groups in favor of the BFR training one (Figure 3).

This direction has been consistently shown in the literature. A study from Centner et al., (2019) compared BFR combined with collagen hydrolysate (BFR+CH) against BFR combined with placebo (BFR+PLA) and to a third group that only consumed the protein supplementation (CH). Both groups with BFR training show improvements in strength capacity while CH group demonstrated a decrease in strength gains, suggesting that BFR is better than non-training to strength capacity, even when participants have protein supplementation [35].

Keeping in mind that there are some elderly people who is not able to perform resistance exercises with high loads, specially following ACSM’s [4] orientations for muscle strength and hypertrophy, due to joint disorders, pain, lack of proprioception and other common conditions of this population, it is important to find therapies or training strategies capable of guaranteeing satisfactory results on the ability to produce muscle strength, insofar as this is a functional capacity that may reduce the risk of falls and mortality as well as increase or maintain the quality of life for elderly who needs to rise from chairs, beds, pick up some objects from the ground and other daily activities that are important for self-autonomy [37, 38].

Our results show that BFR can be an alternative to physiotherapists and other health professionals who needs to find a tolerable training strategy to work with patients with joint injury, pain or any disorder that makes RT not recommendable. This is the main finding of our review because, since muscle strength can be improved with the application of the BFR and this, in turn, requires a much lower intensity than the RT, it can be assumed that this modality would become making exercises with weights more tolerable for people with osteoarticular, neurological and/or musculoskeletal disorders.

Blood flow restriction parameters

A secondary finding of our study was to compare different protocols of BFR and its effects on strength in elderly. Although, we could not perform any kind of statistical analysis due to heterogeneity of selected papers for BFR protocols, as well as the small number of studies about BFR and its effects on strength for this specific population, making it difficult to carry out subgroups analysis.

It should be observed that the volume (sets x repetitions) and frequency found at the primary studies selected for this review are similar to the guidelines provided by ACSM, except for Yasuda et al., (2013) that studied sets up to 30 repetitions [30]. We could not compare different training volumes, intensities and rest intervals. In the same line, it must be noted that there were no comparisons between different occlusion protocols. The only study that made it was the one from Letieri et al., (2018) that compared two different levels of pressure applied on individuals trained with BFR [27]. Although, it was not possible to elicit the best protocol with just one study. A systematic review from Lixandrão et al., (2018) did this comparison, with a different meta-analysis for each of the BFR characteristics (i.e., cuff width, absolute occlusion pressure, test specificity and occlusion pressure prescription method) and found results in favor of high load resistance training (HL-RT) for all the comparisons. Although, it must be observed the population of said paper, which was composed of young people [34]. Our group sought to perform the same comparison, but with the elderly population, in order to determine whether musculoskeletal deterioration, inherent to advancing age, could interfere with the strength gains arising from BFR training.

A systematic review from 2012 compared several variables of the BFR training, in order to ascertain their influence on the results obtained with this modality. The authors’ first analysis demonstrates that BFR training could be better utilized when associated with low-intensity resistance exercise than when associated with high-intensity. The researchers compared different occlusion pressures and, according to the survey data, there was no significant difference between groups using higher pressures and groups with lower pressures in strength gains and hypertrophy. Thus, it could be suggested that the pressure used for training may perhaps be much lower than that practiced in other studies [33].

By comparing these recommendations to the data obtained from our research, we can draw some conclusions. Regarding training intensity, all studies selected for this review evaluated BFR training associated with similar intensities, ranging between 20 and 35% of 1RM, with the exception of the study from Cook et al., (2017), which used between 30 and 50% of 1RM and, when compared with the traditional RT, it proved to be less effective than this one for increasing strength, corroborating the research carried out in 2012 [26, 33]. As for the occlusion pressure, there are some inconsistencies between our findings and those of that paper. The article by Letieri et al, (2018) obtained greater strength gains using BFR training with both high and low occlusion pressure when compared to traditional RT [27]. On the other hand, the study by Vechin et al. (2015) was the one that obtained the worst outcomes from the BFR compared to the RT, although it used the lowest occlusion pressure of all selected studies [28]. Another important training variable is the weekly frequency that is, in most of selected studies composed by 2 sessions for week, with the exception of two papers and, casually, these works found a tendency of higher strength improvements for BFR groups than the RT [25, 27]. A previous systematic review with similar population also found papers with 2 to 3 weeks of duration and, although the authors have not statistically analyzed this variable, results did not show trends to any duration variance as a predictor of better results for strength improvement [39]. It is important to understand the influence of time under ischemic conditions. From selected studies, four training protocols [28-30, 32] made participants remain with inflated cuffs during exercise sets and intervals with an approximate time under ischemic condition of 11 minutes, while three training protocols [26, 27, 31] maintained participants under ischemia during exercises and intervals, but deflated during exercises transitions with a approximate time under pressure of 5 minutes per exercise. Last but not least, one training protocol [25] deflated cuffs between sets and between exercises. Comparing our findings with previous literature [40, 41], we find no pattern for these training variables that may have some influence on training results. This inconsistency suggests future research in order to investigate, in greater depth, the influence of each of the training variables associated with the BFR and its effects for strength improvement.

A potential limitation for this systematic review is the heterogeneity of BFR training and assessments protocols. In some cases, we needed to perform analysis with different exercises (e.g. Leg press and Knee Extension). New papers should be conducted with similar protocols, making it viable to other researchers to compare the same exercises and bring new insights to scientific literature.

In the same line, we could not compare BFR parameters, such as AOP, cuff positioning, time under occlusion, high or low intensity with BFR and others because of the wide range of protocols in a few studies. More research must be made with different protocols and, at the same time, similar protocols, so that systematic reviews can be performed comparing subgroups with different BFR training parameters.