Exercise training and resting blood pressure, a network meta-analysis of 270 RCTs (15,827 adults)
Isometric exercise training lowered resting BP by 8.24/4.00 mmHg against no-exercise controls and ranked first of five modes for systolic pressure (SUCRA 98.3%); by sub-mode, wall squat −10.47/−5.33 and handgrip −7.10/−3.46 mmHg pairwise, a gap the network could not separate
Edwards and colleagues pooled every randomised trial since 1990 that compared at least 2 weeks of exercise training with a no-exercise control and reported resting blood pressure. They included 270 trials and 15,827 participants (358 effect sizes).
Against controls, every main mode lowered resting systolic and diastolic pressure: aerobic training −4.49/−2.53 mmHg, dynamic resistance training −4.55/−3.04, combined training −6.04/−2.54, high-intensity interval training −4.08/−2.50 and isometric exercise training −8.24/−4.00 (all p < 0.001).
In the network meta-analysis, isometric training ranked first for systolic pressure (SUCRA 98.3%) and was significantly more effective than aerobic, interval and dynamic resistance training, by about 3.7-4.0 mmHg; its edge over combined training was not significant.
Among sub-modes, the isometric wall squat ranked first for systolic pressure (90.4%), ahead of isometric leg extension and handgrip, and running ranked first for diastolic pressure (91.3%). The wall squat's lead over handgrip was a ranking, not a significant difference. Reductions were larger in people with higher baseline pressure.
The authors conclude that isometric training should be considered a primary recommended mode for blood pressure, while advising caution because the isometric evidence is far smaller and has no direct comparisons.
Included studies
Eligibility: RCTs published 1990 to February 2023; adults with no restriction on health status; exercise of at least 2 weeks against a non-intervention control with no meaningful diet, counselling or exercise influence; no co-interventions; peer-reviewed only.
Volume: 270 RCTs, 15,827 participants (7,632 controls), 358 effect sizes: aerobic 182 (89 walking, 28 cycling, 21 running, 44 other), dynamic resistance 57, combined 46, interval 49 (7 sprint, 13 aerobic interval) and isometric 24 (17 handgrip, 4 wall squat, 3 leg extension). The discussion counts 19 isometric RCTs.
Isometric protocols (as the authors describe the typical prescription; overcoming holds against a fixed handle or a wall): 4 × 2-min contractions with 1-4 min rest, 3 times a week. Handgrip usually at 30% of maximal voluntary contraction.Wall squat and leg extension usually at a joint angle or force that brings heart rate to 95% of the peak reached in a laboratory incremental isometric test; or a self-selected wall-squat knee angle targeting RPE 3.5-4.5 of 10 in bout 1, 5-6 in bout 2, 6.5-7.5 in bout 3 and 8-9 in bout 4.
Methods: pairwise random- or fixed-effects meta-analyses (Comprehensive Meta-Analysis); Bayesian network meta-analysis in MetaInsight (random effects, node-splitting for inconsistency, none found); meta-regression on duration, frequency, compliance and baseline pressure; TESTEX risk of bias on a random 10% of trials per mode.
Key Results
| Mode against no-exercise control | Systolic reduction, mmHg (95% CI) | Diastolic reduction, mmHg (95% CI) | Note |
|---|---|---|---|
| Isometric exercise training, all | 8.24 (6.5 to 10.0) | 4.00 (2.7 to 5.3) | Egger's test significant for diastolic |
| — isometric wall squat | 10.47 | 5.33 | 4 effect sizes |
| — isometric leg extension | 10.05 | 4.23 | 3 effect sizes |
| — isometric handgrip | 7.10 | 3.46 | 17 effect sizes |
| Combined aerobic + resistance | 6.04 (3.2 to 8.9) | 2.54 (1.1 to 4.0) | ranking fell in sensitivity analysis |
| Dynamic resistance | 4.55 (3.2 to 5.9) | 3.04 (2.2 to 3.9) | — |
| Aerobic, all | 4.49 (3.5 to 5.5) | 2.53 (1.8 to 3.2) | Egger's test significant |
| — walking / cycling / running | 2.85 / 6.88 / 6.83 | 1.44 / 3.20 / 5.67 | — |
| High-intensity interval, all | 4.08 (2.6 to 5.5) | 2.50 (1.2 to 3.8) | sprint intervals significant, 4 × 4-min intervals not |
| Network meta-analysis (systolic) | Result |
|---|---|
| Ranking, main modes (SUCRA) | isometric 98.3%, combined 75.7%, dynamic resistance 46.1%, aerobic 40.5%, interval 39.4% |
| Isometric vs aerobic / interval / dynamic resistance | 3.86 (1.19 to 6.54) / 3.95 (0.93 to 7.03) / 3.68 (0.71 to 6.66) mmHg more reduction with isometrics |
| Isometric vs combined | 2.31 (−0.77 to 5.42), not significant |
| Sub-modes against control | wall squat −10.52 (−16.3 to −4.7); leg extension −9.84 (−16.72 to −3.03); handgrip −7.29 (−10.33 to −4.21) |
| Wall squat vs handgrip | 3.24 (−3.38 to 9.83) mmHg, not significant |
| Ranking, sub-modes (SUCRA) | wall squat 90.4%, leg extension 84.7%, handgrip 73.1%, cycling 69.9%, running 66.1% … walking 17.4% |
| Diastolic ranking | main modes: isometric 89.0%; sub-modes: running 91.3%, wall squat 86.1%, handgrip 57.1%, leg extension 56.2% |
| Baseline pressure | each 1 mmHg higher control-group baseline systolic → 0.10 mmHg (0.05 to 0.15) larger reduction |
Normal vs high blood pressure: every main mode lowered systolic pressure significantly in normotensive groups, but all reductions were substantially larger in hypertension (Figure 4; numbers not in the text). The isometric sub-modes could not be stratified by baseline pressure.
Wall squat vs handgrip, separately: pairwise, the wall squat (4 effect sizes) lowered systolic pressure by 10.47 mmHg and handgrip (17) by 7.10; in the network, 10.52 vs 7.29. The 3.24 mmHg difference has a credible interval from −3.38 to 9.83, so the ranking does not establish that the wall squat is better.
Mechanisms
Not tested: the review measured no mechanism. Hansford 2021 (REF-1151) lists the proposed ones (cardiac autonomic regulation, vagal tone, vascular adaptation, oxidative stress) and calls the evidence for each weak.
Authors' interpretation: the weaker effects of walking and of 4 × 4-min aerobic intervals compared with running, cycling and sprint intervals point to intensity as a driver of the reduction.
Inference: the larger effect with higher baseline pressure means a normotensive endurance athlete has less room to fall; the pooled −8.24 mmHg should not be quoted to them as an expected change.
Practical Application
Does a wall sit lower blood pressure? Of five modes, isometric training (19 of 270 trials) lowered resting BP most: −8.2/−4.0 mmHg vs no exercise; wall squat −10.5, handgrip −7.1 systolic (4 vs 17 effect sizes), mostly untrained adults.
The usual dose in these trials: 4 × 2-min holds with 1-4 min rest, 3 times a week, for weeks; handgrip at about 30% of maximum, wall squat at a depth that gets hard by the fourth hold (RPE about 8-9 of 10).
Wall squat or handgrip? The wall squat ranked higher but was not significantly better (3.24 mmHg, interval −3.38 to 9.83), and it rests on 4 effect sizes. Either is reasonable; the leg version needs no equipment.
For a trained endurance athlete with normal BP: expect a smaller change; the large effects come from people with hypertension. Not from this paper; the base's own practice: for a masters athlete with raised BP, isometric holds are an add-on to aerobic training and to the physician's plan, not a replacement for either; avoid holding the breath during holds.
Safety was not assessed in this review; Hansford 2021 (REF-1151) found no rise in adverse events with mostly handgrip training in people with raised BP.
Limitations
The isometric evidence is a small corner of a large network: 19 trials, 24 effect sizes, only 4 for the wall squat and 3 for leg extension; no isometric arm was compared directly with another exercise mode, so its ranking rests on indirect comparisons through the control node.The authors themselves advise caution and note that trials outside their criteria have shown conflicting results against guideline exercise. The discussion nonetheless says the network "highlight[s] the IWS as more effective than the traditionally employed IHG"; the difference was not significant.
Significant heterogeneity for most analyses; significant publication bias for aerobic systolic and diastolic pressure and for isometric diastolic pressure.
Risk of bias (TESTEX) scored for a random 10% of trials per mode; common problems were unmonitored control-group activity, no intention-to-treat analysis and unblinded participants and investigators.
Missing change-score SDs were imputed with a correlation of 0.8; compliance analyses are limited to adherers, as most trials excluded low attenders.
Participants were general-population adults, many hypertensive; no athletes. The sub-modes could not be stratified by baseline pressure.
Two authors (Wiles, O'Driscoll) have run isometric training trials, including a lower-limb (wall-sit) trial, Taylor 2019, which REF-1151 cites; the authors describe this analysis as supporting their previous work.
Read from the author manuscript, not the typeset article; figures and supplementary tables were not seen.
Funding: no specific funding; one author supported by the UK National Institute for Health and Care Research. Competing interests: none relevant declared.
How this entry was checked and changed
Why this level
270 RCTs make the overall network strong, but the isometric claims rest on 24 effect sizes from 19 trials (17 handgrip, 4 wall squat, 3 leg extension), with no trial comparing isometric training head to head with another mode, significant heterogeneity, significant publication bias for isometric diastolic pressure, and risk of bias scored on a random 10% of trials only; read from the author manuscript (Canterbury Christ Church figshare copy, labelled a submitted version; its headline numbers match the published abstract)
Full text (checked 2026-10-16)
270 exercise training RCTs, analysed sample of 15,827 (7632 controls); 358 effect sizes, of which 24 IET (17 IHG, 4 IWS, 3 ILE) (full text).
IET programmes commonly employ 4 x 2-minute contractions, 1–4-minute rest, 3x per week; IHG at 30% maximum voluntary contraction; IWS and ILE at 95% of the peak heart rate; self-selected wall squat RPE 3.5-4.5, 5-6, 6.5-7.5 and 8-9 for bouts 1 to 4 (full text).
IET overall reduction of 8.24mmHg (95% CI = 6.5-10.0) systolic, 7.10mmHg for IHG, 10.05mmHg ILE and 10.47mmHg for IWS (full text).
IET diastolic reduction of 4.0 (95% CI = 2.7-5.3), 3.46mmHg for IHG, 4.23 ILE and 5.33 for IWS; Egger's test significant for overall IET dBP publication bias (full text).
SUCRA sBP: IET 98.3%, CT 75.7%, RT 46.1%, AET 40.53%, HIIT 39.44% (full text).
IET vs AET WMD -3.86mmHg (1.19-6.54), vs HIIT -3.95mmHg (0.93-7.03), vs RT -3.68mmHg (0.71-6.66) (full text).
Secondary sBP SUCRA: IWS 90.4%, ILE 84.7%, IHG 73.1%; dBP: Running 91.3%, IWS 86.1% (full text).
Table 2, against control: IWS -10.52 (-16.3, -4.7); IHG -7.29 (-10.33, -4.21); IWS vs IHG 3.24 (-3.38, 9.83) (full text).
Baseline sBP moderator: 0.10mmHg (95% CI= 0.05-0.15) per unit (full text).
"As seen in Figure 5, the NMA included no direct comparative IET data" (full text).
Funding: "No sources of funding were used to assist in the preparation of this article"; NJC supported by NIHR (full text).