Effects of blood flow restriction training on cardiovascular autonomic function in middle-aged and older adults: a meta-analysis

Abstract

This meta-analysis aimed to evaluate the effects of blood flow restriction training (BFRT) on heart rate variability (HRV), blood pressure, and heart rate in middle-aged and older adults. Following the PRISMA 2020 guidelines, randomized controlled trials examining the long-term effects of BFRT on HRV and blood pressure in adults aged 45 years or older were systematically searched in Web of Science, PubMed, Scopus, and Cochrane Library up to December 1, 2025. Meta-analyses were performed using the meta and metafor packages in RStudio. Subgroup analyses and meta-regression were conducted to explore potential sources of heterogeneity. Fourteen randomized controlled trials were included. BFRT significantly improved the root mean square of successive differences between adjacent NN intervals (RMSSD; SMD = 0.46, 95% CI [0.21, 0.71], p < 0.001, I² = 35.7%) and reduced systolic blood pressure (SBP; SMD = -0.67, 95% CI [ -1.05, -0.29], p < 0.0001, I² = 54.5%), diastolic blood pressure (DBP; SMD = -0.37, 95% CI [ -0.72, -0.02], p = 0.04, I² = 46.7%), and heart rate (HR; SMD = -0.30, 95% CI [-0.60, -0.01], p = 0.04, I² = 0%). No significant effects were observed for the standard deviation of normal-to-normal intervals (SDNN), low-frequency power (LF), high-frequency power (HF), low-frequency/high-frequency ratio (LF/HF), or percentage of adjacent normal-to-normal intervals differing by more than 50 ms (pNN50). Exploratory subgroup analyses suggested that participant characteristics and BFRT protocol variables may partly contribute to heterogeneity in blood pressure responses. Meta-regression indicated that intervention duration was associated with the SBP response, although this finding should be interpreted cautiously. BFRT may increase RMSSD and reduce blood pressure in middle-aged and older adults, with a potential modest reduction in heart rate. However, the certainty of evidence was limited, and subgroup findings related to training frequency and exercise modality should be considered hypothesis-generating only. Larger, well-powered randomized trials with complete sex reporting and direct comparisons of different BFRT modalities and intensities are needed to confirm these findings.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-026-60475-0.

Introduction

The autonomic nervous system (ANS) is a central network regulating involuntary physiological processes and plays a crucial role in controlling cardiovascular, digestive, and blood pressure functions, as well as maintaining physiological homeostasis^1,2. With the continued growth of the aging population, cardiovascular health issues among middle-aged and older adults have become an increasing public health concern^3,4. Individuals in this age group often face autonomic dysfunction and an elevated risk of hypertension, which not only impairs daily quality of life but also increases the incidence of chronic diseases⁵. Heart rate variability is an important indicator for assessing cardiovascular autonomic function, reflecting both sympathetic and parasympathetic activity^6,7. Meanwhile, blood pressure serves as a key measure of cardiovascular load and vascular function. In middle-aged and older adults, reduced HRV and elevated blood pressure are closely associated with adverse cardiovascular outcomes, functional decline, and increased mortality risk^8–10. Therefore, systematic assessment of HRV and blood pressure holds significant clinical and practical value.

Exercise interventions have been shown to improve heart rate variability in older adults^2,11. Although moderate-to-high intensity exercise can benefit cardiovascular health^12,13, some middle-aged and older individuals, particularly those with frailty, functional limitations, low exercise tolerance, or chronic cardiovascular and metabolic diseases, may require lower-load and more individualized exercise approaches. In contrast, low-intensity, practical, and safe training approaches may be more suitable as interventions to enhance autonomic function in middle-aged and older adults. Therefore, middle-aged and older adults urgently need feasible and effective strategies to improve autonomic function to counter age-related declines in the autonomic nervous system and associated health issues.

Blood flow restriction training involves applying pressure to the proximal part of a limb to reduce arterial blood flow to the muscles while restricting venous return¹⁴. Previous studies have demonstrated that BFRT combined with low-intensity exercise can significantly improve HRV and resting blood pressure, particularly in hypertensive individuals¹⁵. Consequently, BFRT has emerged as a time-efficient and effective intervention. However, existing findings are not entirely consistent. Some studies report that BFRT significantly reduces blood pressure and enhances cardiovascular autonomic function^16,17, whereas others have observed no significant effects¹⁸. These discrepancies may be attributable to differences in study design, sample characteristics, or intervention protocols. This meta-analysis aims to evaluate the effects of BFRT on cardiovascular autonomic function, particularly HRV, and blood pressure in middle-aged and older adults, and to investigate potential moderators of training effects. Additionally, it seeks to provide more robust evidence to support the application of BFRT in middle-aged and older populations.

Materials and methods

Protocol registration

This systematic review and meta-analysis was conducted in accordance with the PRISMA 2020 guidelines for reporting systematic reviews and meta-analyses (details provided in the Supplementary Materials). The study protocol was registered on PROSPERO (registration number: CRD420261276887) after the search strategy had been finalized.

Literature search strategy and study selection

Prior to the formal search, a preliminary search was conducted by one author. Two independent reviewers then systematically searched PubMed (MEDLINE), Web of Science, Scopus and Cochrane Library databases, with the search conducted up to December 1, 2025. Detailed search strategies are provided in Table 1.

Table 1.

Database search strategy.

Web of Science	TS=(((“blood flow restriction” OR “blood-flow restriction” OR BFR OR BFRT OR KAATSU OR “KAATSU training” OR “occlusion training” OR (“blood flow” NEAR/5 restrict*) OR (“blood flow” NEAR/5 occlu*)) AND (exercis* OR train* OR “physical activity” OR “physical activities” OR rehabilitation OR “exercise therapy” OR “resistance training” OR “strength training” OR aerobic* OR endurance OR walking OR cycling OR “low load” OR “low-load” OR “low intensity” OR “low-intensity”) AND ((“blood pressure” OR hypertension OR hypertensive OR SBP OR DBP OR “systolic blood pressure” OR “diastolic blood pressure”) OR (autonomic* OR “autonomic nervous system” OR “heart rate variability” OR HRV OR SDNN OR RMSSD OR baroreflex* OR sympath* OR parasympath* OR vagal OR cardiovagal))))
PubMed	((“blood flow restriction“[tiab] OR “blood-flow restriction“[tiab] OR BFR[tiab] OR BFRT[tiab] OR KAATSU[tiab] OR “KAATSU training“[tiab] OR “occlusion training“[tiab]) AND (exercise[tiab] OR exercises[tiab] OR training[tiab] OR “physical activity“[tiab] OR “physical activities“[tiab] OR rehabilitation[tiab] OR “exercise therapy“[tiab] OR “resistance training“[tiab] OR “strength training“[tiab] OR aerobic[tiab] OR endurance[tiab] OR walking[tiab] OR cycling[tiab] OR “low load“[tiab] OR “low-load“[tiab] OR “low intensity“[tiab] OR “low-intensity“[tiab]) AND ((“blood pressure“[tiab] OR hypertension[tiab] OR hypertensive[tiab] OR SBP[tiab] OR DBP[tiab] OR “systolic blood pressure“[tiab] OR “diastolic blood pressure“[tiab]) OR (autonomic[tiab] OR autonomic*[tiab] OR “autonomic nervous system“[tiab] OR “heart rate variability“[tiab] OR HRV[tiab] OR SDNN[tiab] OR RMSSD[tiab] OR baroreflex*[tiab] OR sympath*[tiab] OR parasympath*[tiab] OR vagal[tiab] OR cardiovagal[tiab]))) NOT (animals[MeSH Terms] NOT humans[MeSH Terms])
Scopus	TITLE-ABS-KEY(((“blood flow restriction” OR “blood-flow restriction” OR BFR OR BFRT OR KAATSU OR “KAATSU training” OR “occlusion training” OR (“blood flow” W/5 restrict*) OR (“blood flow” W/5 occlu*)) AND (exercis* OR train* OR “physical activity” OR “physical activities” OR rehabilitation OR “exercise therapy” OR “resistance training” OR “strength training” OR aerobic* OR endurance OR walking OR cycling OR “low load” OR “low-load” OR “low intensity” OR “low-intensity”) AND ((“blood pressure” OR hypertension OR hypertensive OR SBP OR DBP OR “systolic blood pressure” OR “diastolic blood pressure”) OR (autonomic* OR “autonomic nervous system” OR “heart rate variability” OR HRV OR SDNN OR RMSSD OR baroreflex* OR sympath* OR parasympath* OR vagal OR cardiovagal))))
Cochrane	((“blood flow restriction” OR “blood-flow restriction” OR BFR OR BFRT OR KAATSU OR “KAATSU training” OR “occlusion training” OR (“blood flow” NEAR/5 restrict*) OR (“blood flow” NEAR/5 occlu*)) AND (exercis* OR train* OR “physical activity” OR “physical activities” OR rehabilitation OR “exercise therapy” OR “resistance training” OR “strength training” OR aerobic* OR endurance OR walking OR cycling OR “low load” OR “low-load” OR “low intensity” OR “low-intensity”) AND ((“blood pressure” OR hypertension OR hypertensive OR SBP OR DBP OR “systolic blood pressure” OR “diastolic blood pressure”) OR (autonomic* OR “autonomic nervous system” OR “heart rate variability” OR HRV OR SDNN OR RMSSD OR baroreflex* OR sympath* OR parasympath* OR vagal OR cardiovagal)))

Inclusion and exclusion criteria

The inclusion criteria were: (1) Participants aged ≥ 45 years were included because this review focused on middle-aged and older adults¹⁹; (2) studies comparing individuals receiving BFRT with those not receiving BFRT; (3) studies reporting at least one outcome related to HRV or blood pressure; and (4) randomized controlled trials (RCTs) or crossover studies. The exclusion criteria were: (1) studies for which the full text or relevant data were not available; and (2) reviews, conference abstracts, study protocols, and non-human studies.

Data extraction

After deduplication and screening using Zotero, data were extracted from the remaining studies. Data extraction was primarily conducted using Excel, carefully recording each study’s participants’ age, sex, health status, and BMI, intervention duration, cuff width and pressure, exercise type, and pre- and post-intervention values for each outcome measure. When data were presented graphically, values were extracted using WebPlotDigitizer (Copyright 2010–2024, Ankit Rohatgi). When key methodological or intervention details were not clearly reported in the original articles or supplementary materials, we attempted to contact the corresponding authors to request additional information. If no response was received or the requested information could not be obtained, the relevant items were recorded as “NA”.

Risk of bias assessment

The risk of bias for all included studies was assessed using the Cochrane Risk of Bias 2 (RoB 2) tool. The tool evaluates five domains: (1) the randomization process; (2) deviations from intended interventions (including adherence and allocation effects); (3) missing outcome data; (4) outcome measurement; and (5) selection of the reported results. Each domain is rated as “Yes,” “Probably Yes,” “Probably No,” or “No.” According to Cochrane guidelines, the overall risk of bias is categorized as “High,” “Some Concerns,” or “Low.” Two reviewers independently assessed each study across all domains and the overall risk of bias, with any discrepancies resolved by a third reviewer.

Statistical analysis

Statistical analyses were primarily conducted using the meta and metafor packages in RStudio. For each study, change scores were first calculated for the intervention and control groups, and the between-group difference in change scores was then used to estimate the intervention effect. When only the standard error (SE) was reported, the standard deviation (SD) was calculated as follows:

For studies that did not report the SD of change scores, it was estimated using the following formula:

SD_pre and SD_post represent the standard deviations of pre- and post-intervention measurements. When the SD of change scores was not reported, it was estimated using baseline and post-intervention SDs and an assumed pre–post correlation coefficient. Following methodological recommendations for imputing missing change-score variances, a moderate correlation coefficient of r = 0.6 was used in the primary analysis²⁰. This value was selected as a plausible mid-range assumption, slightly above the commonly suggested minimum correlation of 0.5 for change-score imputation, and was used to avoid assuming either weak or very strong pre–post correlation. To examine the robustness of this assumption, sensitivity analyses were conducted using r = 0.4 and r = 0.8.

For outcomes measured on different scales or using different assessment methods, standardized mean differences (SMD) were used²¹. All effect sizes are reported with 95% confidence intervals (CIs). Heterogeneity was assessed using Cochrane’s Q test and the I² statistic. An I² > 50% with P < 0.10 was considered indicative of moderate-to-high heterogeneity, in which case a random-effects model was applied to pool effect sizes; otherwise, a fixed-effects model was used. For outcomes showing moderate or substantial heterogeneity, leave-one-out influence analysis was performed to examine the robustness of the pooled estimates and to identify whether the overall effect was disproportionately influenced by any single study. In this analysis, one study was sequentially omitted at a time, and the pooled effect size, 95% confidence interval, p value and I² were recalculated using the same random-effects model. The direction, magnitude, statistical significance, and heterogeneity of the pooled effect after each omission were compared with the overall pooled estimate. This analysis was used to assess the stability of the findings and to identify studies that may contribute to between-study heterogeneity. Publication bias was evaluated using funnel plots²² combined with Egger’s test²³, with P > 0.05 indicating no significant bias. Subgroup and sensitivity analyses were conducted for the primary outcomes.

Assessment of evidence quality

The quality of scientific evidence was assessed following the recommendations of the GRADE Handbook. The evidence was rated as high, moderate, low, or very low using GRADEpro GDT software, taking into account risk of bias, inconsistency of results, indirectness of evidence, imprecision, and publication bias to evaluate the effects of BFRT on heart rate variability and blood pressure. Initially, evidence was considered of high certainty, but downgrading was applied based on the following criteria: (1) Risk of bias: evidence was downgraded by one level if there were some concerns in outcome measures, or by two levels if high risk was identified. (2) Indirectness: evidence was downgraded by one level when indirectness was present (e.g., differences in population, interventions, comparators, or outcomes), or by two levels if multiple sources of indirectness were identified. (3) Inconsistency: evidence was downgraded by one level if between-study heterogeneity (I² > 50%) was high or if confidence intervals showed poor overlap. (4) Publication bias: downgraded by one level when evident. Evidence could be upgraded if the following conditions were met: (1) a large magnitude of effect; (2) a dose-response relationship was present; or (3) potential residual confounding or bias was likely to reduce, rather than exaggerate, the observed effect²⁴.

Results

Literature search results

A total of 1,427 records were initially retrieved, and after deduplication using Zotero, 793 records remained. The remaining records were screened by title, abstract, and full text, resulting in 10 studies meeting the preliminary inclusion criteria. Additionally, four more studies were identified through supplementary searches in other databases, such as Google Scholar, yielding a total of 14 studies included in the review. The detailed study selection process is illustrated in Fig. 1.

Fig. 1.

Literature search flowchart.

Study characteristics

Detailed data were extracted from the 14 included studies, encompassing demographic characteristics and BFRT-related parameters (Table 2). All included studies were long-term randomized controlled trials (RCTs), comprising a total of 432 participants: 60 males, 80 females, and 292 participants whose sex was not reported. The high proportion of missing sex information may limit the interpretability of subgroup analyses.

Table 2.

Study characteristics and BFRT protocols of the included studies.

Study	Characteristics					Blood flow restriction training protocol				Experimental protocol
	Age	BMI	Health status	Sex	N	Position	Cuff pressure	Type of exercise	Intensity	Recording duration	Duration	Frequency	Week	Outcome
Kambič et al. (2019)31	64.9	30.15	Coronary Artery Disease	Mix	12	Thigh	15–20 + SBP	Resistance training	35% 1-RM	NA	45 min	3	8	SBP DBP
Bane et al. (2024)28	71.5	26.29	Parkinson’s disease	Mix	38	Thigh	60% AOP	Resistance training	20% 1-RM	NA	NA	3	4	RMSSD SBP DBP
Baniasadi et al. (2025)26	54.68	29.71	Knee osteoarthritis	Female	38	Thigh	60% AOP	Resistance Training	9–11 Borg scale (6–20)	5 min	NA	3	8	RMSSD SDNN LF/HF
Cezar et al. (2016)34	63.75	27.28	Hypertension	Mix	27	Arm	70% SBP	Resistance training	30% 1-RM	NA	NA	2	8	SBP DBP HR
De Deus et al. (2021)30	58.06	33.32	Patients with Kidney Disease	Mix	70	Thigh	50% SBP	Resistance training	60% 1-RM	30 min	60 min	3	24	RMSSD pNN50 SDNN
Doustaki Zaboli et al. (2025)16	55.2	28.3	Hypertension	Mix	45	Thigh	30% AOP	Cycling	50–60% VO2 peak	NA	55 min	3	10	RMSSD pNN50 LF HF LF/HF HR
Ferreira Junior et al. (2019)27	52.4	28.7	Healthy	Male	21	Thigh	100 mmHg	Walking	70% HRmax	10 min	20 min	3	6	LF/HF SDNN RMSSD LF HF
Junior et al. (2019)29	52	28.6	Healthy	Male	21	Thigh	100 mmHg	Walking	NA	10 min	19 min	3	6	SDNN RMSSD LF HF LF/HF HR
Kim et al. (2025)32	52.5	22.8	Hypertension	Male	18	Thigh	275 mmHg	Running	40–60% HRR	NA	20 min	2	8	HR SBP DBP
Lopes et al. (2021)25	72	26.3	Healthy	Female	12	Arm and thigh	50% AOP	Resistance training	70% 1-RM	30 min	50 min	3	12	SDNN RMSSD pNN50 LF HF LF/HF HR
Ma et al. (2024)36	70	26.4	Diabetes Mellitus	Mix	34	Arm	40–50% AOP	Resistance training	20–30% 1-RM	NA	50 min	3	24	SBP DBP
Yasuda et al. (2016)35	70	20.8	Healthy	Female	30	Thigh	161	Resistance training	5.6–8.4 OMNI-RES	NA	NA	2	12	SBP DBP
Zhang et al. (2024)33	71.25	20.16	Sarcopenia	Mix	21	Thigh	50% AOP	Resistance training	20–30% 1-RM	NA	NA	3	12	SBP DBP HR
Zhao et al. (2022)15	62.5	23.2	Hypertension	Mix	45	Thigh	130% SBP	Resistance training	30% 1-RM	15 min	NA	3	12	RMSSD LF/HF HR SBP DBP

Note: SBP: Systolic Blood Pressure; DBP: Diastolic Blood Pressure; HR: Heart Rate; RMSSD:Root mean square of successive differences between adjacent NN intervals; SDNN: Standard Deviation of normal-to-normal (NN) intervals; LF: Low Frequency power; HF: High Frequency power; pNN50: Percentage of adjacent NN intervals differing by more than 50 ms, AOP: arterial occlusion pressure, LOP: limb occlusion pressure. NA: Not available.

Risk of bias assessment

The overall risk of bias across all included studies was rated as “Some Concerns.” According to the RoB 2.0 tool, the main contributing factors were: (1) issues with the randomization process, and (2) incomplete or unclear reporting of attrition and its handling in the published text. The detailed risk of bias assessment is presented in Fig. 2.

Fig. 2.

Risk of bias assessment.

Meta-analysis results

RMSSD

A total of eight studies^{15,16,25–30} were included. Meta-analysis indicated that blood flow restriction training (BFRT) significantly improved RMSSD in middle-aged and older adults (SMD = 0.46 [0.21, 0.71], p < 0.001, I² = 35.7%). Subgroup analyses showed that exercise type and health status did not significantly moderate the effect of BFRT on RMSSD (p > 0.05), as presented in Table 3. Meta-regression analyses revealed that BMI (β = -0.01, 95% CI [-0.13, 0.12]), age (β = 0.001, 95% CI [-0.04, 0.05]), intervention duration in weeks (β = -0.03, 95% CI [-0.08, 0.02]) and measurement duration (β =-0.02, 95% CI [-0.06,0.01]) did not significantly influence the effect of BFRT on RMSSD.

Table 3.

RMSSD analysis results.

Subgroup	K (N)	SMD	95% CI	P d	Q	I2(%)	P m
Type of exercise							0.12
Resistance	5 (90)	0.61	[0.15; 1.08]	< 0.05	8.48	52.8
Walking	2 (23)	0.53	[-0.07; 1.13]	> 0.05	0.42	0
Cycling	1 (15)	0.46	[0.20; 0.71]	n/a	n/a	n/a
Health status							0.13
Normal	3 (35)	0.41	[-0.08; 0.90]	> 0.05	0.86	0
Abnormal	5 (93)	0.48	[0.18; 0.77]	< 0.05	10.37	61.4

Note: K refers to the number of included studies; SMD represents the pooled effect size; Pd is the P-value for the pooled effect size in each subgroup; Q and I² are indicators of heterogeneity; Pm is the P-value for between-subgroup differences; n/a indicates not applicable.

Systolic blood pressure

A total of nine studies^{15,26–28,31–35} were included. Results indicated that BFRT significantly reduced SBP in middle-aged and older adults (SMD = -0.67 [-1.05, -0.29], p < 0.0001, I² = 54.5%). Subgroup analyses suggested that exercise type, participants’ health status, baseline blood pressure level, and cuff pressure prescription may partly contribute to the heterogeneity in the effect of BFRT on SBP, as presented in Table 4. However, these subgroup findings should be interpreted cautiously because some subgroup levels contained only one study. Meta-regression analyses indicated that BMI (β = -0.03, 95% CI [-0.16, 0.10]) and age (β = 0.02, 95% CI [-0.03, 0.07]) did not significantly influence the effect of BFRT on SBP (p > 0.05). In contrast, intervention duration in weeks was significantly associated with the effect of BFRT on SBP (β = 0.05, 95% CI [0.01, 0.09], p < 0.05), suggesting that intervention duration may partly explain the variability in SBP responses. Meta-regression results are shown in Fig. 3.

Table 4.

SBP analysis results.

Subgroup	K (N)	SMD	95% CI	P d	Q	I2(%)	P m
Type of exercise							< 0.01
Resistance	7 (206)	-1.13	[-2.51; -0.18]	< 0.05	42.80	86
Walking	1 (21)	-0.48	[-1.36; 0.39]	n/a	n/a	n/a
Running	1 (33)	-0.81	[-1.52; -0.09]	n/a	n/a	n/a
Health status							< 0.01
Normal	2 (41)	-0.30	[-0.92; 0.32]	> 0.05	0.34	0
Abnormal	7 (219)	-0.65	[-1.03; -0.44]	< 0.05	41.24	85.5
Baseline blood pressure level							0.003
Normal	6 (178)	-0.33	[-0.63; -0.04]	0.03	6.48	22.9
Hypertension	3 (82)	-1.19	[-1.79; -0.66]	p < 0.01	2.51	20.5
Training frequency							0.23
3	5 (168)	-0.45	[-0.7661; -0.1382]	p < 0.01	10.91	63.3
2	4 (92)	-0.78	[-1.2154; -0.3488]	p < 0.01	5.22	42.6
Cuff pressure							0.02
Individualized	3 (116)	-0.35	[-0.94; 0.23]	0.24	4.46	55.2
Non-individualized	6 (144)	-0.84	[-1.20; -0.50]	p < 0.01	7.78	35.7

Note: K refers to the number of included studies; SMD represents the pooled effect size; Pd is the P-value for the pooled effect size in each subgroup; Q and I² are indicators of heterogeneity; Pm is the P-value for between-subgroup differences; n/a indicates not applicable.

Fig. 3.

Meta-regression results on systolic blood pressure.

Diastolic blood pressure

Nine studies^{15,27,28,31–36} were included in the meta-analysis. The results indicated that BFRT significantly reduced DBP in participants (SMD = -0.37 [ -0.72, -0.02], p = 0.04, I² = 46.7%). Subgroup analyses suggested potential between-subgroup differences according to exercise type, health status, and training frequency; however, these findings should be interpreted cautiously because several subgroup levels contained only one study, as presented in Table 5. Meta-regression analyses did not identify any significant moderators (p > 0.05).

Table 5.

DBP analysis results.

Subgroup	K (N)	SMD	95% CI	P d	Q	I2(%)	P m
Type of exercise							< 0.01
Resistance	7 (206)	-0.14	[-0.43; 0.14]	> 0.05	17.77	66.2
Walking	1 (21)	-0.19	[-1.05; 0.66]	n/a	n/a	n/a
Running	1 (33)	-1.16	[-1.91; -0.41]	n/a	n/a	n/a
Health status							< 0.01
Normal	2 (41)	-0.19	[-0.92; 0.32]	> 0.05	0	0
Abnormal	7 (219)	-0.28	[-0.56; 0.00]	< 0.05	23.95	74.9
Baseline blood pressure level							0.05
Normal	6 (178)	-0.16	[-0.46; 0.12]	0.26	3.81	0.0
Hypertension	3 (82)	-0.73	[-1.20; -0.26]	0.01	7.29	72.6
Training frequency							0.02
3	5 (168)	-0.12	[-0.43; 0.18]	0.44	3.67	0.0
2	4 (92)	-0.73	[-1.16; -0.30]	0.02	6.24	51.9
Cuff pressure prescription							0.17
Individualized	3 (116)	-0.13	[-0.51; 0.23]	0.56	3.54	43.5
Non-individualized	6 (144)	-0.49	[-0.83; -0.15]	0.03	9.63	48.1

Note: K refers to the number of included studies; SMD represents the pooled effect size; Pd is the P-value for the pooled effect size in each subgroup; Q and I² are indicators of heterogeneity; Pm is the P-value for between-subgroup differences; n/a indicates not applicable.

Heart rate

A total of seven studies were included in the quantitative synthesis of heart rate^{15,16,25,29,32–34}. The pooled analysis demonstrated that BFRT significantly reduced HR in participants (SMD = -0.30 [-0.60, -0.01], p = 0.04, I² = 0%). Subgroup analyses showed that none of the predefined subgroup factors significantly moderated the effect of BFRT on HR. Meta-regression analyses further indicated that the examined covariates were not significantly associated with the pooled HR effect (all p > 0.05).

Other outcomes

Among other outcomes, no significant improvements were observed for SDNN (SMD = 0.09, p = 0.54), LF (SMD =-0.07, p = 0.72), HF (SMD = 0.20, p = 0.34), LF/HF (SMD = 0.02, p = 0.94), or pNN50 (SMD = 0.06, p = 0.71). However, these analyses included a small number of studies, particularly for pNN50, which was based on only three studies. Therefore, the findings for these secondary HRV outcomes should be interpreted cautiously because of limited statistical power. The forest plot results of all outcome measures are shown in Fig. 4.

Fig. 4.

Forest Plot of outcome measures. Note: SMD: Pooled effect size; SBP: Systolic Blood Pressure; DBP: Diastolic Blood Pressure; HR: Heart Rate; RMSSD: Root Mean Square of Successive Differences between adjacent normal; RR (NN) intervals; SDNN: Standard Deviation of normal-to-normal (NN) intervals; LF: Low Frequency power; HF: High Frequency power; pNN50: Percentage of adjacent NN intervals differing by more than 50 ms. BFRT: Blood Flow Restriction Training.

Sensitivity analyses and assessment of publication bias

Because fewer than 10 studies were included for this outcome, publication bias assessment should be interpreted cautiously. Exploratory Egger’s test suggested possible small-study effects, and the funnel plot appeared asymmetric. Therefore, trim-and-fill analysis was conducted as an exploratory sensitivity analysis. The trim-and-fill method estimated zero missing studies, and the pooled effect remained unchanged. These findings suggest that the pooled estimate was not materially altered by the trim-and-fill adjustment, although the possibility of publication bias cannot be excluded because of the small number of included studies. Sensitivity analyses indicated that the pooled results for RMSSD, SBP, DBP, and HR were stable. The funnel plot is shown in Fig. 5. Egger’s test indicated that SBP may be subject to publication bias (p < 0.05), whereas no significant publication bias was detected for RMSSD, DBP, or HR (p > 0.05).

Fig. 5.

Funnel plot. Note: A: Systolic Blood Pressure; B: Diastolic Blood Pressure; C: Heart Rate; D: Root Mean Square of Successive Differences between adjacent normal.

Assessment of evidence quality

The certainty of evidence was assessed using the GRADE framework. Evidence from randomized controlled trials was initially rated as high certainty and was subsequently downgraded according to risk of bias, inconsistency, indirectness, imprecision, and publication bias. For SBP, the certainty of evidence was downgraded to very low because of risk of bias, inconsistency, and potential publication bias. However, because the pooled effect showed a relatively large magnitude, the evidence was upgraded by one level, resulting in a final rating of low certainty. The certainty of evidence for RMSSD was not downgraded for publication bias, as Egger’s test was non-significant (p = 0.27). Although the funnel plot showed slight visual asymmetry, this was not considered sufficient evidence of publication bias given the small number of included studies (< 10), and the result was therefore interpreted cautiously. For DBP, HR, RMSSD, SDNN, LF, HF, LF/HF, and pNN50, the certainty of evidence was downgraded mainly because of methodological limitations, small sample sizes, inconsistency, or imprecision. Overall, the certainty of evidence for all outcomes was rated as low. The detailed criteria for these ratings are presented in Table 6.

Table 6.

Evidence grade assessment.

	Risk of bias		Inconsistency		Indirectness		Publication bias		Certainty
HR	Some concerns	↓	I2 = 0%	↔	Inconsistent control groups	↓	p = 0.43	↔	Low
SBP	Some concerns	↓	I2 = 54.5%	↓	Inconsistent control groups	↓	p = 0.03	↓	Low
DBP	Some concerns	↓	I2 = 46.7%	↔	Inconsistent control groups	↓	p = 0.23	↔	Low
RMSSD	Some concerns	↓	I2 = 35.7%	↔	Inconsistent control groups	↓	p = 0.27	↔	Low
SDNN	Some concerns	↓	I2 = 0%	↔	Inconsistent control groups	↓	p = 0.22	↔	Low
LF	Some concerns	↓	I2 = 0%	↔	Inconsistent control groups	↓	p = 0.36	↔	Low
HF	Some concerns	↓	I2 = 0%	↔	Inconsistent control groups	↓	p = 0.98	↔	Low
LF/HF	Some concerns	↓	I2 = 28.0%	↔	Inconsistent control groups	↓	p = 0.75	↔	Low
pNN50	Some concerns	↓	I2 = 0%	↔	Inconsistent control groups	↓	p = 0.17	↔	Low

Note: SBP: Systolic Blood Pressure; DBP: Diastolic Blood Pressure; HR: Heart Rate; RMSSD: Root Mean Square of Successive Differences between adjacent normal; RR (NN) intervals; SDNN: Standard Deviation of normal-to-normal (NN) intervals; LF: Low Frequency power; HF: High Frequency power; pNN50: Percentage of adjacent NN intervals differing by more than 50 ms.

Discussion

By pooling data from 14 studies, the present study systematically evaluated the effects of blood flow restriction training on heart rate variability and blood pressure in middle-aged and older adults. The results showed that blood flow restriction training significantly improved both systolic and diastolic blood pressure in this population. This finding is consistent with that reported by Feng et al.³⁷, who also found that blood flow restriction training significantly improved blood pressure in middle-aged and older adults. In addition, the present study found that blood flow restriction training significantly increased RMSSD, a time-domain index of heart rate variability, in middle-aged and older adults. Overall, these findings suggest a potential favorable influence of BFRT on cardiovascular autonomic regulation in middle-aged and older adults; however, the interpretation should be made cautiously given the limited number and heterogeneity of included studies.

Heart rate variability

Middle-aged and older adults are often characterized by reduced cardiac vagal activity, typically manifested as decreased parasympathetic activity and increased sympathetic activity^2,38,39. RMSSD is a time-domain index of HRV that reflects short-term, beat-to-beat fluctuations in heart rate and is widely used as an indicator of vagal (parasympathetic) regulation^40,41. In the present study, BFRT significantly increased RMSSD in middle-aged and older adults. This finding suggests that blood flow restriction training may improve vagal activity and enhance parasympathetic regulation, thereby exerting a positive effect on autonomic cardiovascular control. These effects may have potential clinical and health-promoting significance for middle-aged and older adults. One possible explanation is that BFRT increases shear stress during exercise and activates endothelial nitric oxide synthase, thereby enhancing shear stress-mediated nitric oxide production in endothelial cells^42–44. This may improve endothelium-dependent vasodilation and the hemodynamic environment. In addition, the hypoxic metabolic stress induced by BFRT may alter angiogenesis-related molecular signaling. Over time, these adaptations may improve microcirculation and resting metabolic stress, thereby indirectly promoting parasympathetic predominance⁴⁵. Moreover, the local hypoxic metabolic stress induced by BFRT may activate signaling pathways related to angiogenesis and neural regulation, which may contribute to long-term improvements in autonomic function⁴⁶. This study also found that blood flow restriction training combined with resistance exercise produced significant improvements. Previous studies have shown that resistance exercise can improve autonomic indicators such as HRV^47,48. Resistance exercise may increase local metabolic stress and promote the expression of VEGF, HIF-1α, and eNOS, thereby improving microcirculatory perfusion and cardiovascular function^49,50. It may also improve HRV indices and sympathovagal balance by activating parasympathetic activity and suppressing sympathetic activity. Resistance exercise combined with BFRT may further amplify the local hypoxic metabolic stress induced by blood flow restriction, thereby enhancing the regulation of autonomic function.

In the present study, BFRT did not show significant effects on SDNN, LF, HF, LF/HF, or pNN50, which may be attributable to differences in recording duration across studies. For example, Nussinovitch et al.⁵¹ systematically evaluated HRV parameters under different recording durations in 70 healthy participants and found that indices such as SDNN and LF were highly dependent on recording length. In the present meta-analysis, the measurement durations for the included HRV indices were not standardized, which may have been a major reason for the non-significant findings. Second, the sample sizes for studies reporting these HRV parameters were relatively small, which may also have contributed to the lack of statistically significant differences. Similarly, Lopes et al.²⁵ reported no significant changes in either time-domain or frequency-domain HRV indices when comparing BFRT performed at different resistance training intensities, which they attributed to the small sample size. Taken together, blood flow restriction training may help prevent HRV-related dysfunction and associated geriatric syndromes to some extent by improving RMSSD.

Blood pressure

Elevated blood pressure is a common risk encountered in middle-aged and older adults. The American Heart Association (AHA) has indicated that higher blood pressure is associated with an increased risk of cardiovascular diseases such as myocardial infarction and heart failure⁵². In the present meta-analysis, BFRT was associated with reductions in both systolic and diastolic blood pressure in middle-aged and older adults, suggesting a potential beneficial effect on blood pressure regulation. However, given the limited certainty of evidence and the heterogeneity across included studies, these findings should be interpreted cautiously. From a clinical perspective, even modest reductions in systolic blood pressure may be relevant, as previous evidence has suggested that a 5 mmHg reduction in systolic blood pressure is associated with an approximately 10% lower risk of major cardiovascular events, with corresponding reductions in the risks of stroke, heart failure, ischemic heart disease, and cardiovascular mortality^53,54.

Further subgroup analyses suggested that exercise modality, health status, cuff pressure prescription, training frequency, and baseline blood pressure level may influence the blood pressure-lowering effects of BFRT. BFRT combined with resistance training showed a relatively pronounced antihypertensive effect, which may be partly explained by the blood pressure-lowering effect of resistance training itself^55,56. When low-load resistance training is combined with BFRT, local metabolic stress, metabolite accumulation, muscle pump activity, reactive hyperemia, and reperfusion-related shear stress may be further enhanced, thereby promoting endothelial nitric oxide-mediated vasodilation^57,58. These responses may contribute to improved endothelial function, peripheral circulation, and vascular resistance. In contrast, walking- or cycling-based BFRT may rely more on systemic aerobic adaptations and autonomic regulation^59,60. Training intensity may also influence the cardiovascular responses to BFRT. Low-load resistance protocols, such as 20–30% 1RM, may enhance local metabolic and vascular stimulation under blood flow restriction while maintaining relatively low mechanical stress. In contrast, higher-load protocols, such as 60–70% 1RM, may provide stronger training stimuli but may also increase acute pressor responses and sympathetic activation⁶¹. Therefore, differences across BFRT modalities may partly reflect differences in intensity prescription rather than exercise type alone. However, because the number of studies was unevenly distributed across exercise modalities and some modality subgroups included only a small number of studies, the current evidence is insufficient to determine which BFRT modality produces the greatest blood pressure-lowering effect. Therefore, modality-related findings should be interpreted as exploratory and hypothesis-generating.

Because only a limited number of studies reported sex-specific data, the current evidence remains inconclusive and should be interpreted with caution. Further studies with adequate sex stratification are needed to confirm whether sex modifies the blood pressure response to BFRT. Subgroup analyses also suggested that participants with non-healthy conditions exhibited more pronounced blood pressure reductions. This may be partly explained by their higher baseline blood pressure levels compared with healthy participants, leaving greater room for improvement after intervention. Similar findings were reported by Cornelissen et al.⁶², who showed that blood pressure reductions following exercise training were greater in hypertensive individuals than in normotensive individuals. In addition, individuals with chronic diseases are often characterized by increased sympathetic activity and impaired autonomic regulation⁶³. In this context, BFRT may reduce sympathetic outflow and improve sympathovagal balance, thereby producing a more pronounced antihypertensive response in non-healthy populations than in healthy individuals.

Training frequency and cuff pressure prescription may also contribute to heterogeneity in blood pressure outcomes. For SBP, both twice-weekly and three-times-weekly training protocols produced significant reductions, and no significant difference was observed between the two subgroups. For DBP, the twice-weekly subgroup showed a greater reduction than the three-times-weekly subgroup. However, this finding should not be interpreted as evidence that a lower training frequency is superior, because training frequency may be confounded by intervention duration, exercise modality, baseline blood pressure, and participants’ health status. Regarding cuff pressure prescription, significant reductions in both SBP and DBP were observed in studies using non-individualized pressure protocols, whereas the individualized pressure subgroup showed smaller and non-significant reductions. This result should be interpreted cautiously. Non-individualized pressure protocols included heterogeneous approaches, such as fixed absolute pressures and systolic blood pressure-based pressures, and this subgroup also included more participants with higher baseline blood pressure. Previous methodological recommendations suggest that cuff pressure should ideally be individualized according to arterial occlusion pressure or limb occlusion pressure to account for inter-individual differences in limb circumference, blood pressure, cuff width, and vascular characteristics⁶⁴.

Overall, BFRT may reduce blood pressure in middle-aged and older adults, with more pronounced effects observed in individuals with elevated baseline blood pressure or chronic diseases. The magnitude of this effect may be influenced by multiple factors, including exercise modality, training intensity, training frequency, cuff pressure prescription, and participant characteristics. Future adequately powered clinical trials are needed to directly compare different BFRT modalities and intensities under standardized cuff pressure and training-volume conditions.

Heart rate

The present study found that BFRT significantly reduced HR in middle-aged and older adults. This finding suggests that BFRT may exert beneficial effects on cardiac autonomic regulation and cardiovascular efficiency in this population. A reduction in HR after intervention generally reflects reduced sympathetic excitability, enhanced parasympathetic modulation, or improved cardiovascular adaptation to exercise training. This interpretation is supported by Junior et al.²⁹, who investigated the effects of blood flow-restricted walking training on HR kinetics, HRV dynamics, and recovery responses. Their findings suggest that BFR combined with aerobic exercise may influence not only peripheral vascular responses but also heart rate regulation.

From a physiological perspective, this effect may be partly explained by the unique vascular and metabolic stimuli induced by blood flow restriction. During BFRT, partial restriction of arterial inflow and compression of venous return create a local hypoxic and metabolically stressful environment in the exercising muscles. Although such stimuli may acutely activate sympathetic responses, repeated long-term exposure may induce favorable adaptations in autonomic regulation. Zhao et al.¹⁵ reported that low-intensity resistance training combined with BFR significantly improved heart rate recovery speed and reduced the low-frequency/high-frequency power ratio in patients with hypertension, suggesting improved sympathovagal balance and autonomic nervous system regulation.

In addition, BFRT may indirectly reduce HR by improving peripheral vascular regulation. Repeated cycles of blood flow restriction and reperfusion may enhance reactive hyperemia and vascular shear stress, thereby promoting endothelial and vascular adaptations⁶⁵. Previous studies have shown that BFR exercise may improve vascular-related outcomes, including endothelial function and arterial compliance^66,67. Improved peripheral perfusion and reduced vascular resistance may decrease cardiac afterload and reduce the HR required to maintain adequate circulatory supply⁶⁸. However, because the number of studies included in the present HR analysis was limited, this finding should be interpreted cautiously. Future large-scale studies with standardized HR measurement protocols are needed to further confirm the long-term effects of BFRT on HR regulation in middle-aged and older adults.

Future research directions

Future studies should further standardize BFRT protocols and clearly report key intervention variables, including exercise modality, intensity, training volume, cuff width, cuff pressure, and whether pressure was individualized according to arterial or limb occlusion pressure. In particular, future randomized controlled trials should directly compare different BFRT modalities, such as resistance training, walking, cycling, and running, as well as different exercise intensities, such as 20–30% 1RM versus higher-load protocols, under comparable training volumes and cuff-pressure conditions. These comparisons are needed to determine the most effective and safest BFRT prescription for improving blood pressure and cardiovascular autonomic function in middle-aged and older adults.

In addition, future meta-analyses should consider multilevel models to account for multiple time points or intervention arms within the same study. More studies including female participants are also needed to clarify potential sex-specific effects of BFRT.

Limitations

This study has several potential limitations. First, the relatively small sample size of the included studies may limit the interpretation of the findings⁶⁹. Second, subgroup analysis of cuff pressure in BFRT was not performed because of the limited sample size and the lack of standardized pressure settings across studies. However, previous research has suggested that cuff pressure may influence the effects of BFRT differently across populations¹⁴. Third, there was marked heterogeneity in exercise modality, intensity, and training volume across the included studies. BFRT was combined with resistance training, walking, cycling, or running, and exercise intensity was prescribed using different metrics, including %1RM, HRmax, HRR, VO₂peak, Borg scale, and OMNI-RES. These differences limited our ability to determine the independent effects of exercise modality or intensity on blood pressure and HRV. Moreover, some subgroups included only one study; therefore, these subgroup findings should be considered exploratory rather than confirmatory. HRV recording duration was not standardized across the included studies, ranging from 5 to 30 min, and several studies did not report this information. Although sensitivity analysis was performed using studies with reported HRV duration, the reduced number of eligible studies limited interpretation, particularly for frequency-domain outcomes. Therefore, SDNN, LF, HF, LF/HF, and pNN50 should be interpreted cautiously. Finally, the outcome measures used in the included studies were not direct indicators of autonomic nervous system function. Although these indices may reflect trends in autonomic regulation, they are susceptible to multiple confounding factors, such as respiratory rate, tidal volume, postural changes, and circadian rhythm. Therefore, these factors should be considered when interpreting the results.

Conclusion

BFRT may improve RMSSD and reduce blood pressure and heart rate in middle-aged and older adults. However, the certainty of evidence remains limited, and the findings should be interpreted cautiously, especially for outcomes supported by a small number of studies. Subgroup findings related to exercise modality, training frequency, cuff pressure prescription, and participant characteristics should be considered exploratory and hypothesis-generating only. Larger, well-powered randomized trials with complete sex reporting and standardized BFRT protocols are needed to confirm these effects and determine the optimal BFRT prescription for this population.

Supplementary Information

Below is the link to the electronic supplementary material.

Abbreviations

SBP

Systolic blood pressure

DBP

Diastolic blood pressure

HR

Heart rate

RMSSD

Root mean square of successive differences between adjacent NN intervals

SDNN

Standard Deviation of normal-to-normal (NN) intervals

LF

Low Frequency power

HF

High Frequency power

pNN50

Percentage of adjacent NN intervals differing by more than 50 ms

AOP

Arterial occlusion pressure

LOP

Limb occlusion pressure

BFRT

Blood flow restriction training

HRV

Heart rate variability

Author contributions

Chenghao Liu contributed to study concept and design; curated and analyzed the data; prepared the figures, results, and tables; and wrote and revised the manuscript. Zhenyu Zhang contributed to study concept and design; extracted and analyzed the data; prepared figures; and critically revised the manuscript.Yun Xie supervised the study and contributed to critical review and revision of the manuscript, and performed the final approval of the article. Tao Liu and Mingnan Zhuang contributed to developing the literature search strategy. All authors reviewed and approved the final version, and no other person made a substantial contribution to the paper.

Funding

No funding was obtained for this study.

Data availability

The datasets employed and/or analyzed in the current study can be provided upon a reasonable request from the corresponding author.

Competing interests

The authors declare no competing interests.

Declaration of generative AI and AI-assisted technologies in the writing process

During the preparation of this work the author(s) used ChatGPT in order to translate the content of the article. After using this tool, the author(s) reviewed and edited the content as needed and take(s) full responsibility for the content of the publication.