REF-1151 · Hansford et al., 2021 · Hypertension Research · Paper · PubMed

Effectiveness and safety of isometric resistance training in adults with high blood pressure

Across 24 RCTs (mostly handgrip, 4 × 2 min at 30% of maximum) office BP fell 6.97/3.86 mmHg against controls with no rise in adverse events (7 minor vs 1, about one per 38,444 bouts), but every trial was at high risk of bias and certainty is very low to low

A-fairly strong evidenceMeasured on non-athletes

Hansford and colleagues updated the evidence on isometric resistance training for raised blood pressure and added the first pooled data on central and 24-hour ambulatory pressure and on safety.

Twenty-four RCTs (1,286 randomised; 1,143 in the published abstract; mean age 56, 56% women) compared at least 3 weeks of isometric training with non-exercise, sham or lifestyle-advice controls in adults with high-normal pressure or grade 1 hypertension.

Twenty-one used handgrip, most often 4 × 2 min at 30% of maximal voluntary contraction, for a mean of 9.8 weeks at 3.2 sessions a week.

Office systolic pressure fell by 6.97 mmHg and diastolic by 3.86 mmHg against controls; central systolic and diastolic pressure fell by 7.48 and 3.75 mmHg; 24-hour diastolic fell by 2.39 mmHg, but 24-hour systolic (−2.77 mmHg) did not change significantly.

Adverse events were not more frequent (risk ratio 1.12, 0.47 to 2.68): 7 minor events with isometric training (mostly joint or muscle pain, one dyspnoea and tachycardia) against 1 in controls, about one per 38,444 bouts, and no serious events. In three trials against aerobic training there was no significant difference.

Every trial was at high risk of bias, and the authors rate the certainty very low (office and ambulatory BP, central systolic) to low (central diastolic, adverse events). They suggest isometric training as an adjunct for high-normal BP or grade 1 hypertension until better trials exist.

Included studies

Eligibility: RCTs including crossovers, any language, published or unpublished; adults with mean office BP high-normal (130-139/85-89 mmHg), grade 1 or grade 2 hypertension; isometric training for at least 3 weeks with no restriction on dose; comparators aerobic exercise, dynamic resistance exercise, non-exercise, sham isometrics or lifestyle advice.
Trials: 24 (median n=25, range 8-400); 677 women and 541 men; 88 dropouts (48 isometric, 40 control). Comparators: non-exercise 20, sham 4, lifestyle advice 4; three trials also had an aerobic arm and one ongoing trial a dynamic-resistance arm. Two unpublished trials; two non-English.
Isometric protocols (overcoming holds): handgrip in 21 of 24 trials; leg isometrics (knee extension or wall sit) in the rest. Most frequent prescription 4 × 2 min at 30% of maximal voluntary contraction. Mean 9.8 weeks, 3.2 sessions a week, 14.75 min a session. Frequency and intensity varied too little to analyse.
Measurement: mostly validated oscillometric devices; 88% of trials reported standardised BP testing.
Quality: all trials high risk of bias (mainly unblinded participants and personnel, then selective reporting); GRADE very low for office BP, ambulatory BP and central systolic, low for central diastolic and adverse events.

Key Results

Outcome (isometric vs control)TrialsMean difference, mmHg (95% CI)p; heterogeneity
Office systolic24−6.97 (−8.77 to −5.18)p < 0.0001; I² 75.8%
Office diastolic24−3.86 (−5.31 to −2.41)p < 0.0001; I² 73.3%
Central systolic4−7.48 (−14.89 to −0.07)p = 0.035 in the abstract, 0.048 in the results text; I² 63.8%
Central diastolic2 (3 comparisons)−3.75 (−6.38 to −1.12)p = 0.005; I² 0%
24-hour systolic7−2.77 (−6.80 to 1.25) abstract; −2.74 (−6.74 to 1.25) results textp = 0.18; I² 77.7%
24-hour diastolic8−2.39 (−4.28 to −0.40) abstract; −2.39 (−4.38 to −0.40) results textp = 0.02; I² 51.6%
Handgrip, office systolic / diastolic23 comparisons−6.45 (−8.21 to −4.7) / −4.03 (−5.67 to −2.39)both p < 0.0001
Leg isometrics, office systolic / diastolic4 comparisons (3 trials)−8.68 (−14.17 to −3.18) / −2.99 (−6.42 to 0.43)p = 0.0019 / p = 0.087 (not significant)
Aged ≥65, office systolic / diastolicpost hoc−7.17 (−10.89 to −3.46) / −3.51 (−5.47 to −1.55)I² 46.6% / 4.8%
Grade 1 hypertension / high-normal, office systolic—−6.84 (−9.91 to −3.76) / −7.05 (−9.31 to −4.79)—
Isometric vs aerobic training, office systolic / diastolic33.36 / 2.17 mmHg favouring aerobic, both not significantp = 0.53 / p = 0.1
Adverse events (17 comparisons)—risk ratio 1.12 (0.47 to 2.68)p = 0.8; 7 events vs 1; ≈1 per 38,444 bouts
Serious adverse events—none with isometric trainingone myocardial infarction in a dynamic-resistance arm of 8
Clinical thresholds used: 5 mmHg systolic and 2 mmHg diastolic.
Adverse-event reporting: only 7 of the 24 trials clearly reported adverse events; 10 of 21 contacted authors replied. No events occurred in the two lower-limb trials or in the older-adult trials that supplied data.

Mechanisms

Not tested: outside the review's scope.
Authors' interpretation: proposed mechanisms — cardiac autonomic regulation, vagal tone, vascular adaptation and oxidative stress — each lack strong evidence; trials were too small to identify mediators.
Authors' interpretation: lowering central as well as brachial pressure may reduce cardiovascular risk more than lowering brachial pressure alone; central BP has not been shown to predict risk better.

Practical Application

Is isometric training safe for high blood pressure? In 24 small trials, mostly handgrip 4 × 2 min at 30% of max, office BP fell about 7/4 mmHg with no rise in adverse events; certainty very low, all trials at high risk of bias.
Where it fits: the authors suggest it as an adjunct for high-normal BP or grade 1 hypertension, not a replacement for aerobic exercise or medication; against aerobic training it was not significantly different in three trials.
Safety scope: the reassurance covers handgrip at 30% for 4 × 2 min. Longer, heavier or leg-based holds were not tested enough to say; the authors flag the wall sit as promising and under-studied.
24-hour pressure moved less (diastolic −2.39 mmHg, systolic not significant) than office pressure, so the everyday effect may be smaller than the office numbers suggest. Our inference.
Not from this paper; the base's own practice: a masters athlete with raised BP clears any new programme with their physician, keeps their aerobic training, and breathes through each hold.

Limitations

All 24 trials at high risk of bias, mainly from unblinded participants and personnel; GRADE very low to low. Heterogeneity was high and unexplained for most outcomes.
Small trials (median n=25); 73% of comparisons needed change scores calculated from endpoint data.
Safety rests on sparse reporting: 7 of 24 trials clearly reported adverse events, and the "one per 38,444 bouts" figure depends on the bouts in trials that supplied data. Grade 2 hypertension is essentially untested.
Handgrip dominates (21 of 24); leg isometrics have 4 comparisons, with a non-significant diastolic effect.
The age and BP-class subgroups were partly post hoc.
Abstract vs results text (the manuscript abstract matches the published one): 24-hour systolic −2.77 vs −2.74, 24-hour diastolic lower limit −4.28 vs −4.38, risk difference 1.02 vs 1.01, and central systolic p 0.035 vs 0.048 (0.035 is the heterogeneity p in the results); the table above shows both values where they differ.
Funding: one author held national and university scholarships. Disclosures: AES reports speaker honoraria from Takeda, Servier, Novartis and Omron Healthcare and serves as a scientific adviser for Abbott; the other authors declare no conflicts.
How this entry was checked and changed
Why this level

systematic review of 24 RCTs, but the trials are small (median n=25), all at high risk of bias, heterogeneity is high (I² about 75% for office BP) and GRADE certainty is very low to low; the safety estimate rests on the 7 of 24 trials that clearly reported adverse events plus data supplied by authors; read from the author manuscript (UNSWorks copy, labelled a submitted version), whose results text differs slightly from its own abstract and the published abstract in four places (Limitations)

Full text (checked 2026-10-16)
Twenty-four trials randomizing 1286 participants; median n=25, range 8 to 400; females n=677, 56% (full text).
Mean duration of IRT was 9.8 weeks, 3.2 sessions per week, average session 14.75mins; 21 of the 24 trials used handgrip; most frequent prescription 4 x 2 minutes at 30% maximal voluntary contraction (full text).
Office SBP -6.97 mmHg (95%CI -8.77 to -5.18), I2=75.8%; office DBP -3.86 mmHg (95%CI -5.31 to -2.41), I2=73.3% (full text).
Handgrip SBP -6.45 mmHg (95%CI -8.21 to -4.7); isometric leg exercise (n=4) SBP -8.68 mmHg (95%CI -14.17 to -3.18), DBP -2.99 mmHg (95%CI -6.42 to 0.43, p=0.087) (full text).
24hr ambulatory SBP -2.74mmHg (95%CI -6.74 to 1.25, p=0.18); ambulatory DBP -2.39 mmHg (95%CI -4.38 to -0.40, p=0.02) (full text).
Only seven (29%) clearly reported adverse events; risk ratio=1.12 [95%CI 0.47 to 2.68]; 7 adverse events in IRT and 1 in control; one adverse event per 38,444 bouts (full text).
Against aerobic exercise: office SBP mean difference 3.36 mmHg (p=0.53), office DBP 2.17mmHg (p=0.1), favouring aerobic exercise (full text).
All included trials were deemed to be at a high risk of bias (full text).
Disclosures: AES reports speaker honoraria from Takeda, Servier, Novartis and Omron Healthcare, and serves as scientific advisory for Abbott (full text).