← Back

Blood pressure

Everything in the library studied for it, medical and holistic together. Within each kind, what was shown to help comes first, strongest evidence first. What was tested and not shown to help stays on the list.

Strong several reviews agree · Moderate one review or consistent trials · Limited one or a few small studies · Untested no study in people, listed with what it rests on

Medication15 Supplements23 Therapies4 Practices11 Daily life9 Food and drink11 Diets11 Devices and wearables1 Lab markers4

Medication

ACE inhibitorsStrong

Lowering blood pressure

Shown to lower blood pressure about 8/5 mm Hg in adults with high blood pressure, with no difference found between drugs in the class.

The numbers and sources

Effect: Across 92 placebo-controlled trials of 14 ACE inhibitors, doses of half the maximum or more lowered trough blood pressure by about 8 mm Hg systolic and 5 mm Hg diastolic. One eighth to one quarter of the maximum dose gave 60 to 70% of that effect, and doses above the maximum did not lower blood pressure further. No one ACE inhibitor lowered blood pressure more than another [1]. A separate analysis of 354 trials found thiazides, beta-blockers, ACE inhibitors, ARBs and calcium channel blockers produced similar reductions, on average 9.1 mm Hg systolic and 5.5 mm Hg diastolic at standard dose [2].
Population: 12,954 adults with primary hypertension and a baseline blood pressure of 157/101 mm Hg, treated 3 to 12 weeks [1]; about 40,000 treated and 16,000 placebo participants in 354 trials [2].
Sources: [1], [2]
Caveats: Trials lasted 3 to 12 weeks. The lisinopril label reports a smaller blood pressure effect in Black patients than in white patients [3].

ARBs (angiotensin receptor blockers)Strong

Lowering blood pressure

Shown to lower blood pressure about 8/5 mm Hg in adults with high blood pressure, with no difference found between drugs in the class.

The numbers and sources

Effect: Across 46 placebo-controlled trials of 9 ARBs, the largest trials put the trough blood pressure effect of the class at about 8 mm Hg systolic and 5 mm Hg diastolic. One eighth to one quarter of the maximum dose gave 60 to 70% of the full effect, half the maximum gave 80%, and doses above the maximum did not lower blood pressure further. No one ARB lowered blood pressure more than another [1]. A separate analysis of 354 trials found thiazides, beta-blockers, ACE inhibitors, ARBs and calcium channel blockers produced similar reductions, on average 9.1 mm Hg systolic and 5.5 mm Hg diastolic at standard dose [2].
Population: 13,451 adults with primary hypertension and a baseline blood pressure of 156/101 mm Hg [1]; about 40,000 treated and 16,000 placebo participants in 354 trials [2].
Sources: [1], [2]
Caveats: The Cochrane authors found evidence of publication bias and relied on the largest trials for the estimate [1].

Beta-blockersStrong

Lowering blood pressure

Shown to lower blood pressure by about 10/8 mm Hg and heart rate by 11 beats a minute in adults with mild to moderate high blood pressure.

The numbers and sources

Effect: Across 56 trials of beta-1 selective blockers, blood pressure fell by an average of 10/8 mm Hg and heart rate by 11 beats a minute. The largest blood pressure effect came at twice the starting dose, with no graded dose-response for blood pressure across the usual dose range [10]. An analysis of 354 trials found beta-blockers lowered blood pressure by about the same amount as the other main drug classes, on average 9.1/5.5 mm Hg at standard dose [11].
Population: 7,812 people with primary hypertension in 56 trials of eight beta-1 selective blockers [10]; about 40,000 treated and 16,000 placebo participants in 354 trials [11].
Sources: [10], [11]
Caveats: The Cochrane review covers beta-1 selective drugs such as atenolol, metoprolol and bisoprolol, not propranolol or carvedilol [10].

SGLT2 inhibitorsStrong

Blood pressure

Shown to lower systolic blood pressure by about 2.5 mm Hg and diastolic by about 1.5 mm Hg in adults with type 2 diabetes.

The numbers and sources

Effect: Across 43 trials in type 2 diabetes, systolic blood pressure fell by 2.46 mm Hg (95% CI 2.06 to 2.86) and diastolic by 1.46 mm Hg (95% CI 1.09 to 1.82) [16]. Against other diabetes drugs, systolic pressure fell by 4.45 mm Hg (95% CI 3.18 to 5.73) [14]. All three drugs lowered blood pressure against placebo in a network meta-analysis [15]. In adults without diabetes, blood pressure did not change against placebo [17].
Population: 22,528 patients in 43 trials [16]; active-comparator studies [14]; 23,997 participants in 38 trials [15]; 750 adults without diabetes [17].
Sources: [16], [14], [15], [17]
Caveats: Not an approved use. Heterogeneity was moderate [16].

Thiazide diureticsStrong

Lowering blood pressure

Shown to lower blood pressure by dose, about 8/3 mm Hg on hydrochlorothiazide 25 mg and 12/4 mm Hg on chlorthalidone in adults with high blood pressure.

The numbers and sources

Effect: In 33 placebo-controlled trials, hydrochlorothiazide at 6.25, 12.5, 25 and 50 mg a day lowered blood pressure by 4/2, 6/3, 8/3 and 11/5 mm Hg. In 7 trials, chlorthalidone at 12.5 to 75 mg a day lowered it by 12.0/4 mm Hg, with no evidence that higher doses lowered it more [1]. A separate analysis of 354 trials found thiazides lowered blood pressure by about the same amount as the other main drug classes, on average 9.1/5.5 mm Hg at standard dose [2]. In stage 4 chronic kidney disease, chlorthalidone lowered 24-hour systolic pressure by 10.5 mm Hg more than placebo at 12 weeks (95% CI 6.4 to 14.6) [3].
Population: 11,282 people in 60 trials of six thiazides, mean age 55, treated a mean 8 weeks [1]; about 40,000 treated and 16,000 placebo participants in 354 trials [2]; 160 people with stage 4 chronic kidney disease and poorly controlled hypertension [3].
Sources: [1], [2], [3]
Caveats: The chlorthalidone trials started from a different baseline pressure (163/88 against 155/100 mm Hg), so the two drugs' figures are not a direct comparison. Most trials were at unclear or high risk of bias [1].

AllopurinolModerate

Blood pressure

Shown to lower blood pressure by about 3.3/1.3 mm Hg across 10 studies, with larger falls in one trial of adolescents with high uric acid.

The numbers and sources

Effect: Across 10 studies, systolic pressure fell 3.3 mm Hg more on allopurinol than control (95% CI 1.4 to 5.3) and diastolic 1.3 mm Hg (95% CI 0.1 to 2.5), with similar results in the higher-quality randomized trials [14]. In adolescents with newly diagnosed hypertension and uric acid of 6 mg/dL or more, 4 weeks of allopurinol lowered clinic systolic pressure by 6.9 mm Hg against 2.0 on placebo and diastolic by 5.1 against 2.4. Twenty of 30 reached normal blood pressure on allopurinol and 1 on placebo [15].
Population: 738 participants in 10 studies [14]; 30 adolescents aged 11 to 17 in a crossover trial [15].
Sources: [14], [15]
Caveats: Small studies. Not an FDA-approved use [5].

Alpha-blockersModerate

Blood pressure

Shown to lower blood pressure about 8/5 mm Hg more than placebo in high blood pressure, which the review authors called modest and likely an overestimate.

Compared with placebo [16][7]

The numbers and sources

Effect: Across four alpha-blockers, trough blood pressure fell an estimated 8/5 mm Hg more than on placebo, with no drug better or worse than another [16]. In pooled doxazosin label trials, doses of 1 to 16 mg lowered 24-hour blood pressure about 10/8 mm Hg standing and 9/5 mm Hg lying down against placebo [7].
Population: 1,175 adults with primary hypertension, baseline 155/101 mm Hg, in 10 trials of 3 to 12 weeks [16]; about 300 people per group in doxazosin trials [7].
Sources: [16], [7]
Caveats: Only 10 trials, and harms were poorly reported [16]. Tamsulosin, alfuzosin and silodosin are not indicated for blood pressure. For heart outcomes against a diuretic, see the heart failure block under Risks.

Calcium channel blockersModerate

Lowering blood pressure

Shown to lower blood pressure by about 9/5.5 mm Hg at standard dose, similar to the other main classes of blood pressure drug.

The numbers and sources

Effect: Across 354 placebo-controlled trials, calcium channel blockers lowered blood pressure by about the same amount as thiazides, beta-blockers, ACE inhibitors and ARBs, on average 9.1 mm Hg systolic and 5.5 mm Hg diastolic at standard dose and 7.1/4.4 mm Hg at half dose. Symptoms attributed to calcium channel blockers rose with dose [1]. In older adults with isolated systolic hypertension, treatment starting with the dihydropyridine nitrendipine lowered blood pressure by 10.1/4.5 mm Hg more than placebo at 2 years [2].
Population: About 40,000 treated and 16,000 placebo participants in 354 trials of five drug classes [1]; 4,695 adults over 60 with systolic pressure of 160 to 219 mm Hg [2].
Sources: [1], [2]
Caveats: The pooled analysis reports the class as a whole and does not separate the two subgroups [1]. The trial is of a dihydropyridine, with enalapril and hydrochlorothiazide added as needed [2].

ClonidineModerateVery common

High blood pressure

Shown to lower blood pressure: in one trial 50% of men on clonidine reached goal against 25% on placebo, similar to other drug classes, with more stopping for intolerance.

Compared with placebo [1][2]; hydrochlorothiazide, atenolol, captopril, diltiazem and prazosin [1]; spironolactone as a fourth drug [3]

The numbers and sources

How common: Very common: more than 1 in 10 people.
Effect: In a trial of six drugs, 50% on clonidine reached and kept the diastolic goal at one year, against 25% on placebo, 59% on diltiazem, 51% on atenolol, 46% on hydrochlorothiazide and 42% on captopril or prazosin. Drug intolerance was more frequent with clonidine (14%) [1]. With a clonidine patch, blood pressure was controlled in 11 of 15 people against 4 of 15 on placebo [2]. In resistant hypertension, office blood pressure control was similar with clonidine (29.3%) and spironolactone (33.3%), while spironolactone lowered 24-hour blood pressure more [3].
Population: 1,292 men with diastolic pressure 95 to 109 mm Hg, 48% Black, followed at least one year [1]; 30 adults with mild hypertension over 5 weeks [2]; adults with resistant hypertension on three drugs [3].
Sources: [1], [2], [3]
Caveats: No meta-analysis of clonidine for blood pressure was located. The six-drug trial enrolled men only [1]. These trials measured blood pressure, not heart attacks or strokes.

Loop diureticsModerate

Blood pressure

Shown to lower blood pressure about 8/4 mm Hg more than placebo in high blood pressure, an estimate the review authors called likely too high.

Compared with placebo [9]

The numbers and sources

Effect: Loop diuretics lowered systolic blood pressure 7.9 mm Hg (95% CI 5.4 to 10.4) and diastolic 4.4 mm Hg (95% CI 2.8 to 5.9) more than placebo, with no meaningful difference between drugs in the class. Withdrawals for adverse effects and blood chemistry changes did not differ significantly [9]. Torsemide 5 to 10 mg once a day lowered blood pressure in label trials, near maximal by 4 to 6 weeks [1].
Population: 460 people with baseline blood pressure 162/103 mm Hg in 9 trials, mean 8.8 weeks, including furosemide 40 to 60 mg [9]; hypertension trials in the label [1].
Sources: [9], [1]
Caveats: Low quality evidence with likely publication bias. No trial measured heart attacks, strokes or deaths [9]. The furosemide label says people not controlled on thiazides will probably not be controlled on furosemide alone [10].

SpironolactoneModerate

Resistant high blood pressure

Shown to lower home systolic pressure by 8.7 mm Hg more than placebo when added to three drugs in resistant high blood pressure.

The numbers and sources

Effect: In a crossover trial in people whose blood pressure stayed high on three drugs, spironolactone lowered home systolic pressure by 8.70 mm Hg more than placebo (95% CI 7.69 to 9.72), 4.03 mm Hg more than doxazosin and 4.48 mm Hg more than bisoprolol [5]. A meta-analysis of 4 trials of add-on spironolactone in resistant hypertension found systolic pressure fell 16.67 mm Hg (95% CI 5.80 to 27.54) and diastolic 6.11 mm Hg more than placebo [6]. A Cochrane review in hypertension found a reduction of 20.09/6.75 mm Hg across 5 crossover studies, with no further reduction above 50 mg a day [7].
Population: 335 people randomized, 285 of whom received spironolactone, each drug taken for 12 weeks [5]; 869 people in 4 trials, mean 12 weeks [6]; 5 crossover studies in hypertension [7].
Sources: [5], [6], [7]
Caveats: Trials lasted about 12 weeks and did not measure heart attack or stroke. Serious adverse effects or withdrawals tended to be more frequent on spironolactone (OR 2.11, 95% CI 0.98 to 4.53) [6].

Calcium channel blockersModerate

Compared with other blood pressure drugs

Mixed results: fewer strokes than beta-blockers or ACE inhibitors, more heart failure than diuretics or ACE inhibitors, and no difference in deaths from any cause.

The numbers and sources

Effect: In 23 trials of first-line treatment, all-cause mortality did not differ between calcium channel blockers and any other class. Against diuretics, heart failure was more frequent (RR 1.37, 95% CI 1.25 to 1.51). Against beta-blockers, major cardiovascular events (RR 0.84, 95% CI 0.77 to 0.92) and stroke (RR 0.77, 95% CI 0.67 to 0.88) were less frequent. Against ACE inhibitors, stroke was less frequent (RR 0.90, 95% CI 0.81 to 0.99) and heart failure more frequent (RR 1.16, 95% CI 1.06 to 1.28) [4]. In ALLHAT, amlodipine did not differ from chlorthalidone on fatal coronary disease or nonfatal heart attack (RR 0.98, 95% CI 0.90 to 1.07) and had a higher 6-year rate of heart failure (10.2% against 7.7%) [5]. An amlodipine-based regimen brought fewer strokes than an atenolol-based one (HR 0.77, 95% CI 0.66 to 0.89) [6]. Benazepril plus amlodipine brought fewer cardiovascular events than benazepril plus hydrochlorothiazide (9.6% against 11.8%, HR 0.80) [7].
Population: 153,849 people with hypertension in 23 trials, 18 of dihydropyridines and 4 of non-dihydropyridines [4]; 33,357 adults aged 55 or older in ALLHAT, mean 4.9 years [5]; 19,257 adults with hypertension and three other risk factors [6]; 11,506 high-risk adults with hypertension [7].
Sources: [4], [5], [6], [7]
Caveats: Most of the evidence is for dihydropyridines, and the three named trials all used amlodipine. Certainty was rated low to moderate across comparisons [4].

Thiazide diureticsModerate

Chlorthalidone compared with hydrochlorothiazide

Mixed results: indirect comparisons favored chlorthalidone, and the one large head-to-head trial found no difference in heart events between the two.

The numbers and sources

Effect: In a pragmatic trial in adults 65 and older already on hydrochlorothiazide, a major cardiovascular event or non-cancer death occurred in 10.4% switched to chlorthalidone and 10.0% kept on hydrochlorothiazide over a median 2.4 years (HR 1.04, 95% CI 0.94 to 1.16) [7]. In a cohort of first-time users, the composite cardiovascular outcome did not differ (calibrated HR 1.00, 95% CI 0.85 to 1.17) [8]. Network meta-analyses that compared the drugs indirectly across separate trials found 21% fewer cardiovascular events on chlorthalidone (95% CI 12 to 28) [9], and 12% fewer on thiazide-like than thiazide-type diuretics after adjusting for blood pressure [10]. In 9 trials that compared them directly, chlorthalidone lowered systolic pressure by 3.26 mm Hg and diastolic by 2.41 mm Hg more [11].
Population: 13,523 veterans with a mean age of 72 [7]; 730,225 first-time users in three US databases [8]; 9 trials with 50,946 people in the drug-adjusted analysis [9]; randomized trials of at least 1 year in adults with hypertension [10]; 9 head-to-head trials [11].
Sources: [7], [8], [9], [10], [11]
Caveats: In the head-to-head trial, 94.5% were taking hydrochlorothiazide 25 mg at entry and chlorthalidone was given at 12.5 or 25 mg [7]. The indirect comparisons do not randomize between the two drugs. For side effect differences, see the low potassium block below.

ACE inhibitorsModerate

Compared with ARBs

Not shown to differ from ARBs in deaths or cardiovascular events in people with high blood pressure or vascular disease. Fewer people stopped ARBs over side effects.

The numbers and sources

Effect: In 9 head-to-head trials lasting at least a year, total mortality (RR 0.98, 95% CI 0.88 to 1.10), total cardiovascular events (RR 1.07, 95% CI 0.96 to 1.19) and cardiovascular mortality (RR 0.98, 95% CI 0.85 to 1.13) did not differ between ACE inhibitors and ARBs. Withdrawals over adverse effects were lower on ARBs (RR 0.83, 95% CI 0.74 to 0.93), an absolute difference of 1.8% over 4.1 years, mainly from dry cough on ACE inhibitors [14]. In ONTARGET, the composite of cardiovascular death, heart attack, stroke or heart failure hospitalization occurred in 16.5% on ramipril and 16.7% on telmisartan (RR 1.01, 95% CI 0.94 to 1.09) [15].
Population: 11,007 people with primary hypertension in 9 trials [14]; 17,118 people with vascular disease or high-risk diabetes assigned to ramipril or telmisartan, median 56 months [15].
Sources: [14], [15]
Caveats: The Cochrane authors rated the evidence moderate for mortality and low for cardiovascular events, citing possible publication bias [14].

ARBs (angiotensin receptor blockers)Moderate

Compared with ACE inhibitors

Not shown to differ from ACE inhibitors in deaths or cardiovascular events. Fewer people stopped ARBs over side effects, mainly cough.

The numbers and sources

Effect: In 9 head-to-head trials in hypertension, total mortality (RR 0.98, 95% CI 0.88 to 1.10) and total cardiovascular events (RR 1.07, 95% CI 0.96 to 1.19) did not differ, and withdrawals over adverse effects were lower on ARBs (RR 0.83, 95% CI 0.74 to 0.93) [15]. In ONTARGET, the main cardiovascular outcome occurred in 16.7% on telmisartan and 16.5% on ramipril (RR 1.01, 95% CI 0.94 to 1.09) [16]. After a heart attack with heart failure or left ventricular dysfunction, deaths did not differ between valsartan and captopril (979 against 958, HR 1.00, 97.5% CI 0.90 to 1.11) [17]. In heart failure trials, mortality did not differ between ARBs and ACE inhibitors (RR 1.05, 95% CI 0.91 to 1.22), and withdrawals over adverse effects were lower on ARBs (RR 0.63, 95% CI 0.52 to 0.76) [13].
Population: 11,007 people in 9 hypertension trials [15]; 17,118 people assigned to telmisartan or ramipril [16]; 9,818 people assigned to valsartan or captopril after a heart attack, median 24.7 months [17]; heart failure trials within a 22-study review [13].
Sources: [15], [16], [17], [13]
Caveats: The evidence in [16] is for telmisartan and in [17] for valsartan.

Supplements

Black seed oilStrong

Blood pressure

Shown to lower blood pressure by about 3 mmHg systolic and 3 mmHg diastolic over about 8 weeks in adults.

The numbers and sources

Effect: In 11 RCTs (860 people with or without hypertension), systolic pressure fell 3.26 mmHg (95% CI -5.10 to -1.42, I² 59%) and diastolic 2.80 mmHg (95% CI -4.28 to -1.32) more than control over a mean 8.3 weeks. Powder lowered pressure more than oil [4]. The 2025 review also reported lower systolic and diastolic pressure [2].
Population: Hypertensive and normotensive adults [2][4].
Sources: [4], [2]
Caveats: The 2016 review included English-language trials only [4].

GarlicStrong

Blood pressure in people with hypertension

Shown to lower blood pressure about 7 to 8 mmHg systolic and 5 mmHg diastolic in adults with hypertension.

The numbers and sources

Effect: In 12 trials (553 hypertensive participants), garlic lowered systolic pressure by a mean 8.3 ± 1.9 mmHg, and in 8 trials (374 participants) diastolic pressure by 5.5 ± 1.9 mmHg [1]. A second meta-analysis of 7 placebo-controlled trials found systolic pressure fell 6.71 mmHg (95% CI -12.44 to -0.99) and diastolic 4.79 mmHg (95% CI -6.60 to -2.99), with no serious adverse events reported [2].
Population: Adults with hypertension [1][2].
Sources: [1], [2]
Caveats: Trials are small. Review [1] has one author and summarizes trials of Kyolic aged garlic extract. Its abstract does not report funding. The authors of [2] call for longer trials focused on primary endpoints.

L-arginineStrong

Blood pressure

Shown to lower blood pressure about 5 to 6 mmHg systolic and 2.6 mmHg diastolic in adults. Doses above 9 g a day and obese participants showed no change.

The numbers and sources

Effect: Across 11 trials with 387 participants taking 4 to 24 g a day, systolic pressure fell 5.39 mmHg (95% CI -8.54 to -2.25) and diastolic 2.66 mmHg (95% CI -3.77 to -1.54) [1]. Across 22 RCTs, systolic fell 6.40 mmHg (95% CI -8.74 to -4.05) and diastolic 2.64 mmHg. Doses above 9 g a day, trials longer than 24 days and obese participants showed no significant change [2]. In 5 RCTs with 202 participants using ambulatory monitoring, 24-hour systolic fell 4.23 mmHg (95% CI -5.87 to -2.58) and diastolic 3.04 mmHg [3].
Population: Adults with and without hypertension in RCTs [1][2][3].
Sources: [1], [2], [3]
Caveats: Trials were small. The ambulatory analysis included 5 trials [3].

MagnesiumStrong

Blood pressure

Shown to lower blood pressure about 2 mmHg systolic and 2 mmHg diastolic, with larger drops of about 4 mmHg in people with insulin resistance or chronic disease.

The numbers and sources

Effect: Across 34 double-blind trials (2,028 participants), a median 368 mg a day for a median 3 months lowered systolic pressure by 2.00 mmHg (95% CI 0.43 to 3.58) and diastolic by 1.78 mmHg (95% CI 0.73 to 2.82) [1]. In 11 trials of people with insulin resistance, prediabetes or other chronic disease (543 participants), systolic pressure fell by 4.18 mmHg and diastolic by 2.27 mmHg (SMD -0.20 and -0.27) [2].
Population: Normotensive and hypertensive adults [1]; adults with preclinical or chronic disease given 365 to 450 mg a day for 1 to 6 months [2].
Sources: [1], [2]
Caveats: Residual heterogeneity remained after stratification in [1]. Effects in [2] are in people with existing disease.

PotassiumStrong

Blood pressure

Shown to lower blood pressure about 3.5 mmHg systolic and 2 mmHg diastolic in people with hypertension, with little effect in people without it.

The numbers and sources

Effect: Increased potassium intake lowered systolic pressure 3.49 mm Hg (95% CI 1.82 to 5.15) and diastolic 1.96 mm Hg (95% CI 0.86 to 3.06), an effect seen in people with hypertension and not in those without it [1]. Salt substitutes lowered systolic pressure 4.61 mm Hg (95% CI 3.14 to 6.07) and diastolic 1.61 mm Hg in one meta-analysis [2], and 4.76 mm Hg systolic and 2.43 mm Hg diastolic in a Cochrane review [3]. In untreated primary hypertension, oral potassium supplements lowered clinic systolic pressure 5.3 mm Hg (95% CI 4.4 to 6.3) and diastolic 4.5 mm Hg (95% CI 3.9 to 5.1). The drop in 24-hour ambulatory pressure was smaller, 2.8 mm Hg systolic and 3.2 mm Hg diastolic [4]. In a dose-response analysis of 10 RCTs, each 50 mmol a day rise in urinary potassium went with 5.3 mm Hg lower systolic pressure in people with hypertension and 0.5 mm Hg lower in people without it [5].
Population: 22 RCTs with 1,606 adults [1]; 19 trials reporting blood pressure [2]; 20 RCTs with 21,414 adults [3]; 19 trials with 803 adults, median 4 weeks [4]; 10 RCTs from 2000 to 2024 [5].
Sources: [1], [2], [3], [4], [5]
Caveats: Each 10% lower share of sodium chloride in a salt substitute was linked to a further 1.53 mm Hg drop in systolic pressure [2]. In children, the change was 0.28 mm Hg and not significant [1]. Heterogeneity was substantial and follow-up short in [4]. The dose-response review rested on few trials [5].

TaurineStrong

Blood pressure

Shown to lower blood pressure about 3 to 4 mmHg systolic and 1.5 to 2.5 mmHg diastolic in adults.

The numbers and sources

Effect: In 7 studies, taurine lowered systolic pressure by Hedges' g -0.70 (95% CI -0.98 to -0.41) and diastolic by g -0.62, a mean of about 3 mmHg each [3]. Across 25 RCTs, systolic fell by 4.0 mmHg (95% CI -7.3 to -0.7) and diastolic by 1.5 mmHg (95% CI -2.5 to -0.5) [4]. Across 34 RCTs, systolic fell by 4.4 mmHg (95% CI -7.3 to -1.5) and diastolic by 2.5 mmHg (95% CI -4.0 to -1.1) [5].
Population: 103 people of varied age and health [3]; 1,024 participants at 0.5 to 6 g a day for 5 to 365 days [4]; adults with cardiometabolic risk factors [5].
Sources: [3], [4], [5]
Caveats: The three reviews overlap in the trials they pool. The earliest pooled only 103 people.

Beta-glucanModerate

Blood pressure

Shown to lower systolic blood pressure by about 1.4 mmHg in adults with overweight or obesity. Diastolic pressure did not change.

The numbers and sources

Effect: In people with a BMI of 25 or more, oat beta-glucan lowered systolic blood pressure by 1.38 mmHg (95% CI 0.09 to 2.66). Diastolic blood pressure did not change. Across cereal beta-glucan, total cholesterol fell by 0.24 mmol/L and LDL by 0.19 mmol/L [8].
Population: 49 studies, 3,854 people with overweight or obesity [8].
Sources: [8]
Caveats: One meta-analysis. One author is affiliated with a food science company [8].

Beta-sitosterolModerate

Blood pressure and other heart risk markers

Shown to lower systolic blood pressure by about 1.6 mmHg for plant sterols and stanols as a class. HDL, blood sugar and weight did not change.

The numbers and sources

Effect: Plant sterols and stanols lowered systolic blood pressure by 1.55 mmHg (95% CI 0.42 to 2.67), diastolic by 0.84 mmHg, apoB by 0.06 g/L and hs-CRP by 0.25 mg/L. HDL cholesterol, fasting glucose, HbA1c, body weight, waist and fat mass did not change [5].
Population: Adults of mixed health status in trials pooled across 14 meta-analyses. Blood pressure came from one of them [5].
Sources: [5]
Caveats: Class-level evidence, not beta-sitosterol alone. These are risk markers, not heart attacks or strokes. Certainty was moderate to high for most outcomes, downgraded for inconsistency, imprecision or publication bias [5].

Coenzyme Q10Moderate

Blood pressure

Shown to lower systolic blood pressure by about 3.4 mmHg in adults. Diastolic pressure did not change, and no change was seen in people with diabetes.

The numbers and sources

Effect: Across 45 RCTs, CoQ10 lowered systolic pressure by 3.44 mmHg (95% CI -5.13 to -1.55). Diastolic pressure and heart rate did not change significantly [8].
Population: Adults in 48 effect sizes [8].
Sources: [8]
Caveats: Larger reductions in subgroups with lower doses and longer trials [8]. In 20 RCTs in people with prediabetes or type 2 diabetes, systolic and diastolic pressure did not change [9].

Fisetin and quercetin (senolytics)Moderate

Blood pressure (quercetin alone)

Shown to lower blood pressure about 3 mmHg systolic and 2.6 mmHg diastolic with quercetin alone, mainly above 500 mg a day.

The numbers and sources

Effect: Across 7 trials, quercetin lowered systolic pressure by 3.04 mmHg (95% CI -5.75 to -0.33) and diastolic pressure by 2.63 mmHg (95% CI -3.26 to -2.01). The effect appeared limited to, or larger at, more than 500 mg a day [9].
Population: 587 patients in 7 RCTs with 9 arms [9].
Sources: [9]
Caveats: One meta-analysis from 2016. This is quercetin taken alone, not as a senolytic.

Green tea extractModerate

Blood pressure

Shown to lower blood pressure about 2 mmHg in adults, more with extracts and in those starting at 130 mmHg or higher.

The numbers and sources

Effect: Across 13 studies, green tea lowered systolic pressure by 2.08 mmHg (95% CI -3.06 to -1.05) and diastolic by 1.71 mmHg (95% CI -2.86 to -0.56). Reductions were larger with baseline systolic pressure of 130 mmHg or more and with extracts [4].
Population: Adults in RCTs from 1995 to 2013 [4].
Sources: [4]
Caveats: One meta-analysis.

InositolModerate

Weight, blood sugar and lipids

Shown to lower BMI slightly, about 0.6 kg/m², and glucose, triglycerides and blood pressure in adults across several conditions, at low certainty for some.

The numbers and sources

Effect: In 18 RCTs (898 participants) of at least 4 weeks, inositol lowered BMI by 0.57 kg/m² (95% CI -1.10 to -0.03), waist circumference by 2.36 cm, glucose by 7.25 mg/dL, HOMA-IR by 1.21, triglycerides by 29.80 mg/dL, LDL cholesterol by 5.15 mg/dL, and systolic and diastolic pressure by 5.34 and 6.12 mm Hg [11].
Population: Adults in RCTs across several conditions [11].
Sources: [11]
Caveats: Certainty was low to very low for BMI, waist and blood pressure, and moderate for glucose, insulin and HOMA-IR. Heterogeneity was high for most outcomes, and the authors note a high risk of bias in the trials [11].

NMNModerate

Blood pressure

Shown to lower diastolic pressure about 2 mmHg, with no clear change in systolic pressure except a 4 mmHg drop in adults 60 and older.

The numbers and sources

Effect: Across 10 RCTs, NMN lowered resting diastolic pressure by 2.15 mmHg (95% CI -3.68 to -0.61). Systolic pressure did not change significantly overall. In participants aged 60 and older, systolic pressure fell by 3.94 mmHg (95% CI -7.06 to -0.82) [9]. A second meta-analysis found no change in systolic pressure and a slight fall in diastolic pressure [8].
Population: 349 participants in 10 RCTs with 11 arms [9]; 15 RCTs [8].
Sources: [9], [8]
Caveats: The age subgroup is one of several subgroup analyses [9]. The two reviews overlap in the trials they pool.

Vitamin CModerate

Blood pressure

Shown to lower blood pressure about 3.8 mmHg systolic and 1.5 mmHg diastolic in short trials of adults.

The numbers and sources

Effect: Across 29 trials, systolic pressure fell 3.84 mm Hg (95% CI 2.38 to 5.29) and diastolic 1.48 mm Hg (95% CI 0.10 to 2.86) against control. In people with hypertension, systolic fell 4.85 mm Hg, and the diastolic change of 1.67 mm Hg was not significant [2].
Population: Trials of 10 to 120 participants, median dose 500 mg a day, median 8 weeks [2].
Sources: [2]
Caveats: Short trials of blood pressure, not of heart attack or stroke. Effects shrank when 9 trials with imputed data were added [2].

GABALimited

Blood pressure

One trial found GABA-rich Chlorella lowered systolic pressure more than placebo in adults with high-normal blood pressure. GABA alone was not tested.

The numbers and sources

Effect: In 80 adults with high-normal blood pressure or borderline hypertension, GABA-rich Chlorella (20 mg of GABA twice a day) for 12 weeks lowered systolic pressure more than placebo (p below 0.01). Diastolic pressure tended to fall [5]. A USP review found some studies with a transient, moderate drop in blood pressure of less than 10% [6].
Population: 80 Japanese adults [5]. Clinical studies reviewed by USP [6].
Sources: [5], [6]
Caveats: The trial tested GABA inside Chlorella, so the effect cannot be assigned to GABA alone. The abstract does not give the size of the change [5].

CalciumModerate

Preeclampsia and high blood pressure in pregnancy

Mixed results: the 2025 Cochrane update found little to no difference in preeclampsia, while the 2018 version and earlier reviews found fewer cases at 1 g a day or more.

The numbers and sources

Effect: Sources disagree. The 2025 Cochrane update found little to no difference in preeclampsia against placebo (RR 0.83, 95% CI 0.67 to 1.04, 6 RCTs, 15,364 women, low certainty). In the 4 trials of 500 or more women (14,730), the RR was 0.92 (95% CI 0.79 to 1.05, high certainty), and preterm birth before 37 weeks did not differ (RR 0.97, 95% CI 0.85 to 1.11). Baseline calcium intake and preeclampsia risk did not change the findings, and low and high doses did not differ (RR 0.96, 2 RCTs, 22,000 women) [8]. The 2018 version of the review found that 1 g a day or more lowered high blood pressure (RR 0.65, 95% CI 0.53 to 0.81) and preeclampsia (RR 0.45, 95% CI 0.31 to 0.65), with a clear effect in women with low calcium diets (RR 0.36), and that preterm birth fell (RR 0.76, 95% CI 0.60 to 0.97) [9]. An umbrella review of earlier meta-analyses reported a pooled RR for preeclampsia of 0.53 (95% CI 0.42 to 0.68, I² 84%) [10]. Calcium started before pregnancy did not change preeclampsia in 1,355 women with prior preeclampsia (RR 0.84, 95% CI 0.62 to 1.14) [11].
Population: 10 trials with 37,504 women [8]. 13 trials with 15,730 pregnant women [9]. 1,355 parous women in one trial [11].
Sources: [8], [9], [10], [11]
Caveats: The 2025 update removed 20 previously included trials, 11 because eligibility criteria changed and 9 that await classification over trustworthiness concerns. It was part-funded by WHO [8]. The umbrella review's search ended in December 2023, before those exclusions [10]. Heterogeneity was high in the 2018 review and small-study effects were possible. In its largest trial, preeclampsia fell only 8% and the interval included no effect [9].

CitrullineModerate

Blood pressure

Mixed results: reviews in older adults and during cold exposure found systolic pressure about 4 to 9 mmHg lower, while an earlier, smaller analysis found no change.

The numbers and sources

Effect: In 15 RCTs (24 datasets, 415 participants) of L-citrulline or watermelon in middle-aged and older adults, systolic pressure fell 4.02 mmHg (95% CI -6.54 to -1.50) and diastolic 2.54 mmHg (95% CI -4.27 to -0.81). L-citrulline combined with L-arginine lowered systolic pressure 10.44 mmHg (95% CI -13.57 to -7.31) [7]. In 6 RCTs (162 participants) measuring blood pressure during cold exposure, systolic pressure was 9.28 mmHg lower (95% CI -10.66 to -7.90) and diastolic 5.33 mmHg lower (95% CI -9.38 to -1.27) with L-citrulline [8]. In a systematic review of 12 RCTs in 360 postmenopausal women, most of the 7 trials that measured blood pressure reported lower systolic and mean arterial pressure [9]. An earlier analysis of 5 interventions found no change in brachial systolic (0.28 mmHg, 95% CI -2.87 to 2.31), diastolic (-1.56 mmHg, 95% CI -4.32 to 1.20) or aortic pressure [10]. A 2018 meta-analysis reporting a 7.54 mmHg systolic drop across 15 trials has been retracted by its journal [11].
Population: Middle-aged and older adults [7]; adults under cold exposure [8]; postmenopausal women aged 50 to 75 in 4 to 8 week US trials [9]; adults, with body weight a source of heterogeneity [10].
Sources: [7], [8], [9], [10], [11]
Caveats: Sources disagree. The reductions come from analyses of older adults and of cold exposure, and the earlier, smaller analysis found no change [10]. The older-adult analysis pooled L-citrulline, watermelon and citrulline with arginine [7]. Two of the analyses come from one author group [7][8]. The postmenopausal review did not pool results [9]. The retracted paper [11] is listed so that readers who meet its numbers elsewhere know its status.

Grape seed extractModerate

Blood pressure

Mixed results: reviews found lower blood pressure, by up to about 6 mmHg systolic, but disagree on whether systolic or diastolic pressure falls.

The numbers and sources

Effect: In 16 trials (810 people), grape seed extract lowered systolic pressure by 6.08 mmHg (95% CI -10.74 to -1.42) and diastolic pressure by 2.80 mmHg (95% CI -4.42 to -1.19) [1]. An earlier meta-analysis of 9 RCTs (390 people) found systolic pressure 1.54 mmHg lower (95% CI -2.85 to -0.22) and no change in diastolic pressure [2]. A 2022 meta-analysis of 19 trials found diastolic pressure 2.20 mmHg lower (95% CI -3.79 to -0.60) and no significant change in systolic pressure (-3.55 mmHg, 95% CI -7.59 to 0.49) [3]. A meta-analysis of proanthocyanidins from any source (6 RCTs, 376 people) found systolic pressure 4.60 mmHg lower and diastolic pressure 2.75 mmHg lower [4].
Population: Adults, including people with obesity, metabolic syndrome or raised blood pressure [1][2][3][4].
Sources: [1], [2], [3], [4]
Caveats: The reviews disagree on which pressure falls. Heterogeneity in the 2022 analysis was high (I² 88.8% diastolic, 97.4% systolic) [3]. The proanthocyanidin review included sources other than grape seed [4].

TurmericModerate

Blood pressure

Mixed results: one review found systolic pressure about 2.7 mmHg lower in adults with prediabetes or diabetes, while a larger pooled analysis found no effect.

The numbers and sources

Effect: In 15 RCTs (855 adults with prediabetes or type 2 diabetes), systolic pressure fell 2.69 mmHg (95% CI -3.84 to -1.55, I² 30.1%), and 3.41 mmHg in those starting at 130 or above. Diastolic pressure did not change significantly (-1.20 mmHg, 95% CI -2.84 to 0.44) [35]. The 63-trial pooled analysis found no significant blood pressure effect [7].
Population: Adults with prediabetes, type 2 diabetes or hyperglycemia [35][7].
Sources: [35], [7]
Caveats: The two analyses disagree. Diastolic heterogeneity was 84.3% [35].

ResveratrolStrong

Blood pressure

Not shown to lower blood pressure overall, with drops seen at 300 mg a day or more and in people with type 2 diabetes.

The numbers and sources

Effect: Across 17 trials, resveratrol did not significantly change systolic pressure (WMD -2.5 mmHg, 95% CI -5.5 to 0.6), diastolic pressure (WMD -0.5 mmHg, 95% CI -2.2 to 1.3) or mean pressure. Significant effects appeared in subsets with 300 mg a day or more and with diabetes [3]. A second meta-analysis found no change in systolic (WMD 0.82 mmHg, 95% CI -8.86 to 10.50) or diastolic pressure [4]. An umbrella review of 45 systematic reviews rated as high certainty a drop in people with type 2 diabetes of 7.97 mmHg systolic (95% CI 5.31 to 10.63) and 3.55 mmHg diastolic (95% CI 1.93 to 5.18) [5].
Population: 17 RCTs [3]; 10 RCTs with 11 arms [4]; meta-analyses of RCTs [5].
Sources: [3], [4], [5]
Caveats: Heterogeneity was moderate, with I² of 62% for systolic pressure [3]. The subgroup findings come from smaller sets of trials.

Tart cherryStrong

Blood pressure and cardiometabolic markers

Not shown to change blood pressure, cholesterol or body fat in adults. Fasting glucose and CRP fell very slightly.

The numbers and sources

Effect: Across 10 RCTs, tart cherry juice did not change blood pressure, cholesterol, triglycerides or body fat. Fasting glucose fell by 0.51 mg/dL [4]. Across 21 trials, blood pressure, heart rate, hs-CRP and IL-6 did not change, while CRP fell by 0.39 mg/L (95% CI -0.74 to -0.05, moderate certainty) [5]. A 3-month trial in 50 middle-aged adults found no change in vascular or metabolic measures [6].
Population: Adults across 10 to 21 RCTs [4][5]; middle-aged adults, 30 mL concentrate twice a day [6].
Sources: [4], [5], [6]
Caveats: A 0.51 mg/dL glucose change is very small. The 3-month trial was funded by the Cherry Research Committee [6].

ReishiModerate

Blood sugar, cholesterol and blood pressure

Not shown to change HbA1c, fasting glucose, cholesterol or blood pressure against placebo, with one spore oil trial finding lower lipids.

The numbers and sources

Effect: In a Cochrane review of 5 trials (398 participants), 1.4 to 3 g a day for 12 to 16 weeks did not change HbA1c (WMD -0.10%, 95% CI -1.05 to 0.85), total cholesterol, LDL or BMI against placebo. Fasting glucose, blood pressure and triglycerides did not differ. Post-meal glucose results were inconsistent [2]. A 2025 meta-analysis of 17 RCTs (971 participants) at 200 to 11,200 mg a day for 1 to 24 weeks found no effect on blood pressure, fasting glucose or the lipid profile. BMI fell by 0.43 (95% CI 0.10 to 0.77) and heart rate by 3.92 beats a minute (95% CI 0.40 to 7.45) [3]. In 110 adults with dyslipidemia, a spore oil extract for 12 weeks lowered total cholesterol, triglycerides and LDL and raised HDL compared with placebo. The abstract gives no effect sizes [4].
Population: Adults with type 2 diabetes in the analyzed trials [2]. Healthy, at-risk and chronically ill adults [3]. Adults with dyslipidemia [4].
Sources: [2], [3], [4]
Caveats: Data in [2] came from 3 small trials, and risk of bias was unclear in 4 of 5 [2]. The 2025 review rated the certainty of evidence very low for every outcome [3]. The spore oil trial tested a different product and disagrees with both reviews on lipids [4].

Nicotinamide ribosideLimited

Blood pressure

Small trials found no drop in blood pressure against placebo in older adults, including with walking exercise.

The numbers and sources

Effect: In adults 55 and older with high daytime systolic pressure, 1,000 mg a day of NR with walking exercise for 6 weeks did not lower systolic pressure more than placebo with the same exercise (+5.19 against -2.71 mmHg) [10]. Blood pressure did not change in a 12-week trial in mild cognitive impairment [9].
Population: 54 sedentary adults, mean age 67 [10]; 42 older adults with amnestic mild cognitive impairment [9].
Sources: [10], [9]
Caveats: Pilot trials. A nighttime pressure trend in [10] came from a post hoc analysis.

Therapies

ReflexologyModerate

Blood pressure and heart rate

Shown to lower systolic blood pressure by about 4.6 mm Hg and heart rate by about 5 beats a minute in the short term, in pooled trials.

Compared with control groups the source abstract does not describe [14]

The numbers and sources

Effect: Foot reflexology lowered systolic blood pressure (mean difference -4.62 mm Hg, 95% CI -5.58 to -3.66), diastolic blood pressure (-3.32 mm Hg), heart rate (-4.76 beats a minute) and breathing rate, and raised oxygen saturation (0.95%) [14].
Population: 819 patients in 13 trials [14].
Sources: [14]
Caveats: Short-term measurements only. A pooled analysis in pregnant women found no change in blood pressure [13].

Massage therapyModerate

Stress hormones, blood pressure and immune cells after one session

Mixed results: a single session lowered blood pressure and heart rate in pooled trials, and cortisol fell in one small trial but not in pooled trials.

Compared with control conditions in randomized studies the source abstract does not describe [12]; light touch [18]

The numbers and sources

Effect: Single sessions reduced state anxiety, blood pressure and heart rate, and did not reduce cortisol [12]. In healthy adults, 45 minutes of Swedish massage produced a large drop in arginine vasopressin and a small drop in blood cortisol against light touch. Circulating lymphocytes rose (effect sizes 0.14 to 0.43) and stimulated cytokine production fell [18].
Population: 37 randomized studies [12]. Healthy adults aged 18 to 45 [18].
Sources: [12], [18]
Caveats: The trial calls its data preliminary, and these were changes over one session [18]. Lymphocyte counts in blood do not show fewer infections.

CuppingLimited

High blood pressure, wet cupping

Limited evidence: a few small trials suggested wet cupping lowered blood pressure against blood pressure drugs, and reviewers drew no firm conclusion.

Compared with antihypertensive medication [11]

The numbers and sources

Effect: In 7 randomized trials, a few small studies suggested wet cupping alone lowered blood pressure more than antihypertensive medication. The reviewers wrote that no firm conclusions can be drawn [11].
Population: Adults with hypertension in 7 trials [11].
Sources: [11]
Caveats: Most trials were not of high methodological quality [11].

Chiropractic and spinal manipulationModerate

Asthma, high blood pressure and period pain

Not shown to treat asthma in children, high blood pressure or period pain. Every high- or acceptable-quality trial found manipulation no better than sham.

Compared with sham interventions [10]

The numbers and sources

Effect: All randomized trials of acceptable or high quality found spinal manipulation was not superior to sham for infantile colic, childhood asthma, hypertension, primary dysmenorrhea and migraine. No trial tested whether manipulation prevents any of these conditions [10].
Population: 6 randomized trials with 534 participants [10].
Sources: [10]
Caveats: Few trials for each condition. Six of the 50 researchers at the review summit did not approve the final report [10].

Practices

MeditationModerate

Blood pressure

Shown to lower office blood pressure by about 5 to 9 mm Hg systolic in pooled trials. One review found no significant fall in out-of-office readings.

Compared with control groups the source abstract does not describe [11]; control arms the source abstract does not describe [12]; treatment as usual, waiting list, and active programs such as relaxation training, yoga, social support and health education [13]

The numbers and sources

Effect: Transcendental Meditation lowered systolic pressure by 4.7 mm Hg (95% CI -7.4 to -1.9) and diastolic by 3.2 mm Hg (95% CI -5.4 to -1.3) against control in 9 trials, with similar results in high-quality trials and people with hypertension [11]. Mindfulness-based stress reduction lowered office systolic pressure by 6.64 mm Hg and diastolic by 2.47 mm Hg after the program, with no significant reduction in 24-hour, daytime or asleep pressure [12]. Across 12 trials of mindfulness programs, systolic pressure fell 9.12 mm Hg (95% CI -12.18 to -6.05) and diastolic 5.66 mm Hg. Against waiting list alone the difference was not significant [13]. An American Heart Association statement rated Transcendental Meditation Class IIB and other meditation techniques Class III for lowering blood pressure [14].
Population: 9 randomized trials [11]. 12 trials in people with elevated blood pressure or hypertension [12]. 715 people with prehypertension or hypertension in 12 trials [13].
Sources: [11], [12], [13], [14]
Caveats: Only one of 12 trials was at low risk of bias, and out-of-office readings came from few trials [12]. Heterogeneity was high (I² 84%) [13]. The Transcendental Meditation trials were rated from low to high quality [11].

SaunaModerate

Blood pressure

Shown to lower systolic and diastolic blood pressure by about 4 mmHg across heat therapy trials. An 8-week sauna trial in adults with coronary artery disease found no change.

The numbers and sources

Effect: Across 15 studies of repeated heat therapy, systolic pressure fell 3.94 mmHg (95% CI -7.22 to -0.67, 10 studies) and diastolic pressure fell 3.88 mmHg (95% CI -6.13 to -1.63, 9 studies) against control. Resting heart rate did not change [1]. A meta-analysis of 16 sauna articles found systolic and diastolic pressure 5.55 and 6.50 mmHg lower in the 30 minutes after a session, and 5.26 and 4.14 mmHg lower after 2 to 4 weeks of sessions [2]. In 47 sedentary adults, a 15-minute sauna after each exercise session for 8 weeks lowered systolic pressure by 8.0 mmHg (95% CI -14.6 to -1.4) compared with the same exercise alone [3]. In 41 adults with stable coronary artery disease, 4 sauna sessions a week for 8 weeks did not change systolic (P = 0.951) or diastolic pressure (P = 0.292) relative to control [4]. In 16 people with untreated hypertension, the authors report that exercise followed by sauna had positive effects on 24-hour systolic and mean pressure, without giving figures in the abstract [5].
Population: Healthy and clinical populations in 15 studies, 10 to 36 sessions of 30 to 90 minutes [1]. Participants in 16 articles [2]. 47 adults aged 49 on average with low activity and at least one cardiovascular risk factor [3]. 41 adults with coronary artery disease, mean age 62 [4]. 16 people with untreated hypertension [5].
Sources: [1], [2], [3], [4], [5]
Caveats: The pooled review covers heat therapy as a whole, and its abstract does not separate sauna from other heating methods. Certainty was rated moderate for systolic and diastolic pressure [1]. The abstract of [2] does not say how many of its articles were controlled trials. The trial in [3] compared exercise plus sauna against exercise, with no sauna-only group.

Tai chi and qigongModerate

Blood pressure, against no exercise or usual care

Shown to lower blood pressure in people with hypertension by about 12 to 13 mm Hg systolic, in two meta-analyses, one of tai chi and one of Baduanjin qigong.

Compared with control groups the source abstract does not describe [5]; no intervention [6]; antihypertensive drugs alone [6]

The numbers and sources

Effect: Tai chi was associated with lower systolic pressure (WMD -12.47 mm Hg, 95% CI -16.00 to -8.94) and diastolic pressure (WMD -6.46 mm Hg, 95% CI -8.28 to -4.64) and better quality of life [5]. Baduanjin qigong lowered systolic pressure by 13.00 mm Hg (95% CI -21.24 to -4.77) and diastolic by 6.13 mm Hg against no intervention, and did not differ from antihypertensive drugs. Added to drugs, it lowered systolic pressure 7.49 mm Hg more than drugs alone [6].
Population: Adults with essential hypertension in trials from 15 databases [5]. Adults with hypertension in 8 trials [6].
Sources: [5], [6]
Caveats: The two reviews cover different practices, tai chi and Baduanjin qigong. Both searched Chinese databases, and neither abstract reports trial quality or heterogeneity [5][6].

YogaModerate

Blood pressure, against waiting list or usual care

Shown to lower blood pressure about 8/5 mm Hg against waiting list or usual care in people with raised blood pressure, in pooled trials rated very low in quality.

Compared with waiting list [8]; usual care [9]

The numbers and sources

Effect: Across 26 trials, systolic pressure fell 7.95 mm Hg (95% CI -10.24 to -5.66) and diastolic 4.93 mm Hg (95% CI -6.25 to -3.60) against waiting list, with high heterogeneity [8]. An earlier review found systolic pressure 9.65 mm Hg lower (95% CI -17.23 to -2.06) and diastolic 7.22 mm Hg lower against usual care, with effects in trials of people with hypertension and not in trials mixing hypertension and prehypertension [9].
Population: 2,283 people with prehypertension or hypertension in 30 trials [8]. 452 patients in 7 trials [9].
Sources: [8], [9]
Caveats: The two reviews come from overlapping authors, and the later one updates the earlier. Evidence quality was very low and heterogeneity high (I² 90%) [8][9].

SaunaLimitedStudied in men

New high blood pressure

Linked to fewer new cases of hypertension in Finnish men, with a hazard ratio of 0.53 at 4 to 7 sessions a week against one.

The numbers and sources

Effect: Over a median 24.7 years, 251 of 1,621 men (15.5%) developed hypertension. After adjustment, hazard ratios were 0.83 (95% CI 0.59 to 1.18) at 2 to 3 sessions a week and 0.53 (95% CI 0.28 to 0.98) at 4 to 7 sessions, compared with one [14].
Population: 1,621 men aged 42 to 60 without hypertension at baseline [14].
Sources: [14]
Caveats: Observational, one cohort of men. The upper confidence limit sits close to 1.

Slow breathing and breathworkLimited

Blood pressure, resistance breathing training against sham

One trial of 36 adults found 30 daily breaths against high resistance lowered systolic pressure from 135 to 126 mm Hg over 6 weeks. Sham training did not.

Compared with low-resistance sham training [11]

The numbers and sources

Effect: High-resistance inspiratory muscle strength training at 75% of maximal inspiratory pressure lowered casual systolic pressure from 135 to 126 mm Hg (P < 0.01), about 75% of which remained 6 weeks after training stopped, and diastolic pressure from 79 to 77 mm Hg (P = 0.03). Twenty-four hour systolic pressure was lower after training than after sham (P = 0.01). Blood pressure did not change with sham training at 15% of maximum. Brachial artery flow-mediated dilation rose about 45% [11].
Population: 36 adults aged 50 to 79 with systolic pressure of 120 mm Hg or higher, 18 per group [11].
Sources: [11]
Caveats: One small double-blind trial using a handheld resistance device. This is strength training of the breathing muscles, not slow breathing [11].

Tai chi and qigongLimited

Blood pressure, against aerobic exercise

One trial of 342 adults with prehypertension found a year of tai chi lowered office systolic pressure 2.4 mm Hg more than aerobic exercise of the same schedule.

Compared with aerobic exercise, four 60-minute supervised sessions a week [7]

The numbers and sources

Effect: At 12 months, office systolic pressure fell 7.01 mm Hg with tai chi and 4.61 mm Hg with aerobic exercise, a difference of -2.40 mm Hg (95% CI -4.39 to -0.41). Twenty-four hour ambulatory systolic pressure (-2.16 mm Hg) and nighttime systolic pressure (-4.08 mm Hg) were also lower with tai chi [7].
Population: 342 adults aged 18 to 65 with prehypertension at 2 hospitals in China, mean age 49, 51.5% women [7].
Sources: [7]
Caveats: One single-blinded trial. People with full hypertension were not included [7].

Slow breathing and breathworkModerate

Blood pressure, against non-sham controls

Mixed results: pooled trials found blood pressure fell about 5.6/3.0 mm Hg with slow breathing, but device trials showed no effect once manufacturer-linked trials were removed.

Compared with control groups the source abstract does not describe [6]; control groups in 8 device trials, not sham-controlled in most [7]

The numbers and sources

Effect: In 17 randomized trials of slow breathing at 10 breaths a minute or fewer, with or without a guiding device, systolic pressure fell 5.62 mm Hg (95% CI -7.86 to -3.38) and diastolic 2.97 mm Hg (95% CI -4.28 to -1.66), with high heterogeneity [6]. Across 8 trials of the RESPeRATE device, systolic pressure fell 3.67 mm Hg and diastolic 2.51 mm Hg. Excluding the 5 trials sponsored by or involving the manufacturer left no overall effect [7]. The American Heart Association rated device-guided breathing Class IIA for lowering blood pressure in 2013 [8].
Population: People with hypertension, prehypertension or on blood pressure medication in 17 trials [6]. 494 adults in 8 trials, none longer than 9 weeks [7].
Sources: [6], [7], [8]
Caveats: Heterogeneity was high in every analysis [6]. The AHA rating predates the blinded trials in the next block [8][9].

YogaModerate

Blood pressure, against exercise and other active comparisons

Mixed results: against active comparisons such as exercise, systolic pressure did not differ, and diastolic pressure fell about 2 mm Hg in one review.

Compared with active control interventions [8]; exercise [9]

The numbers and sources

Effect: Against active controls, systolic pressure differed by 4.16 mm Hg (95% CI -10.76 to 2.44, p = 0.22) and diastolic fell 1.88 mm Hg (95% CI -3.41 to -0.36), from 5 trials [8]. Against exercise, no evidence was found for effects on systolic or diastolic pressure [9]. An American Heart Association statement rated yoga Class III for lowering blood pressure in 2013 [10].
Population: 306 participants in 5 trials [8]. Patients with prehypertension or hypertension [9].
Sources: [8], [9], [10]
Caveats: Five small trials with very low quality evidence [8].

Infrared saunaLimited

Blood pressure

Mixed results: blood pressure fell right after sessions in small studies, while pooled heart failure trials found no lasting change.

The numbers and sources

Effect: In 31 people with treated hypertension, systolic pressure fell from 118.5 to 115.1 mmHg and diastolic from 70.5 to 65.9 mmHg after a single session (P < .001). After repeated sessions the fall was not significant overall, and was significant in 11 non-smokers (124.2 to 108.8 mmHg systolic) [12]. In 12 healthy adults, mean arterial pressure was 9 mmHg lower immediately after 45 minutes at 65°C [13]. Pooled heart failure trials found no significant effect on systolic or diastolic pressure [1]. A single session in 10 healthy women did not change blood pressure compared with rest [11]. A 2009 review of 9 papers found limited moderate evidence for normalizing blood pressure [14].
Population: 31 people with hypertension, mean age 64, no control group [12]. 12 healthy adults [13]. People with heart failure [1]. 10 healthy women [11].
Sources: [12], [13], [1], [11], [14]
Caveats: The hypertension study had no control group, and its subgroup finding in non-smokers rests on 11 people [12]. No randomized trial of infrared sauna for high blood pressure was located.

Slow breathing and breathworkModerate

Blood pressure, device-guided breathing against sham

Not shown to lower office blood pressure more than a sham device or music, in a pooled analysis of blinded trials and a double-blind trial.

Compared with a sham device playing music without pacing the breath [9][10]; music therapy [9]

The numbers and sources

Effect: In individual patient data from blinded, randomized trials with active controls, office systolic pressure differed by 2.2 mm Hg in favor of the control group (95% CI -2.7 to 7.0) and diastolic by 0.2 mm Hg in favor of the device (95% CI -2.8 to 3.1) [9]. In 48 people with type 2 diabetes and hypertension using the device or a sham for 15 minutes a day for 8 weeks, systolic pressure differed by 2.35 mm Hg in favor of sham (95% CI -6.50 to 11.20) and diastolic by 2.25 mm Hg in favor of the device (95% CI -2.16 to 6.67) [10].
Population: Adults with hypertension in 5 eligible trials, 3 with individual data [9]. 48 adults with type 2 diabetes and hypertension in the Netherlands [10].
Sources: [9], [10]
Caveats: All trials were short [9]. The sham used music, which may itself relax, so these trials ask whether slowing the breath adds anything beyond sitting quietly with music.

Daily life

Physical activityStrong

Blood pressure

Shown to lower resting blood pressure by about 3.5 to 4.5 mmHg systolic with aerobic training in pooled trials, and by about 8 mmHg in people with hypertension.

Compared with non-exercising control groups [11][12]

The numbers and sources

Effect: Endurance training lowered systolic pressure 3.5 mmHg (95% CI -4.6 to -2.3) and diastolic 2.5 mmHg. In people with hypertension the fall was 8.3/5.2 mmHg, and in people with normal pressure 0.75/1.1 mmHg, with the systolic change not significant [11]. A larger analysis found aerobic training lowered pressure by 4.49/2.53 mmHg, high-intensity interval training by 4.08/2.50 and isometric training by 8.24/4.00 [12].
Population: 5,223 healthy adults in 93 trials of at least 4 weeks [11]. 15,827 participants in 270 trials [12].
Sources: [11], [12]
Caveats: Trials ran weeks to months, and blinding to exercise is not possible.

Sleep durationStrong

Heart disease and stroke

Linked to more heart disease and stroke at both short and long sleep in pooled cohorts, with the lowest risk near 7 hours a night.

Compared with people sleeping 7 to 8 hours a night, the reference in most studies [1]; normal sleepers [3]; people with normal sleep duration [4]

The numbers and sources

Effect: Each hour under 7 was linked to 6% more total cardiovascular disease (RR 1.06, 95% CI 1.03 to 1.08) and 7% more coronary heart disease (RR 1.07, 95% CI 1.03 to 1.12). Each hour over 7 was linked to 12% more cardiovascular disease (RR 1.12) and 18% more stroke (RR 1.18, 95% CI 1.14 to 1.21) [1]. Short sleepers had higher risk of cardiovascular disease (RR 1.16, 95% CI 1.10 to 1.23), coronary heart disease (RR 1.26) and hypertension (RR 1.17) [3]. Long sleepers had higher risk of cardiovascular disease (RR 1.25), stroke (RR 1.46, 95% CI 1.26 to 1.69) and coronary heart disease (RR 1.24), and no higher risk of hypertension (RR 1.01, 95% CI 0.95 to 1.07) [4].
Population: 3,582,016 participants in generally healthy populations, follow-up 2.3 to 34 years [1]. 5,172,710 participants in 153 studies [3]. 5,134,036 participants in 137 cohorts [4].
Sources: [1], [3], [4]
Caveats: Observational, with self-reported sleep. Confounding and reverse causation, where early heart disease or other illness changes sleep, cannot be ruled out. The three reviews draw on overlapping cohorts.

Strength trainingStrong

Blood pressure

Shown to lower resting blood pressure by about 2 to 4.5 mmHg systolic and 3 mmHg diastolic with dynamic resistance training, and more with isometric training, in pooled trials.

Compared with non-exercising control groups [12][13]

The numbers and sources

Effect: Dynamic resistance training lowered blood pressure by 1.8/3.2 mmHg, most in people with prehypertension (4.0/3.8 mmHg). Isometric resistance training lowered it by 10.9/6.2 mmHg in a few studies [12]. A larger analysis found 4.55/3.04 mmHg with dynamic resistance training and 8.24/4.00 mmHg with isometric training [13].
Population: 29 dynamic and 5 isometric resistance groups among 5,223 adults in 93 trials [12]. 15,827 participants in 270 trials [13].
Sources: [12], [13]
Caveats: Trials lasted weeks to months. The isometric estimates rest on far fewer trials [12].

Time in natureModerate

Blood pressure and heart rate

Shown to lower systolic blood pressure by about 3 mmHg and diastolic by about 2 mmHg in forest settings, in one meta-analysis of 20 trials, with heart rate also lower.

Compared with non-forest environments [5]; people with lower greenspace exposure [2]; before the forest bathing, in the same people [6]; control conditions, mostly passive [7]

The numbers and sources

Effect: Systolic blood pressure was lower in the forest than in non-forest environments (MD -3.15 mmHg, 95% CI -4.12 to -2.18, I² 1%), as was diastolic pressure (MD -1.75 mmHg, 95% CI -2.38 to -1.13) [5]. Greenspace exposure was linked to lower diastolic pressure (-1.97 mmHg) and heart rate (-2.57 beats a minute) [2]. Heart rate fell after forest bathing (MD -4.00, 95% CI -6.90 to -1.09) at very low certainty [6]. Across reviews, nature-based programs lowered heart rate (SMD -0.70, 95% CI -1.17 to -0.23) [7].
Population: 732 participants in 20 trials, 17 in Japan [5]. 143 studies [2]. 11 studies [6]. 30 reviews in a second-order meta-analysis [7].
Sources: [5], [2], [6], [7]
Caveats: Trials were short, and 17 of the 20 were in Japan [5]. Walking, quiet and cleaner air come with forest time and cannot be separated from it. No trial located tested whether the drop lasts.

SunlightLimited

Blood pressure

One small study found UVA light on the skin lowered blood pressure in 24 healthy volunteers, with signs of nitric oxide released from the skin.

Compared with a control condition the source abstract does not describe [12]

The numbers and sources

Effect: Two standard erythemal doses of UVA on the skin lowered blood pressure, with falls in circulating nitrate and rises in nitrite. Forearm blood flow rose, independently of nitric oxide synthase [12].
Population: 24 healthy volunteers [12].
Sources: [12]
Caveats: A UV lamp in a laboratory, not sunlight. The abstract gives no size for the fall, and the effect was measured over hours, not weeks.

Daily stepsModerate

Blood pressure in step-count programs

Mixed results: one review found systolic pressure fell 3.8 mmHg with pedometer programs, while reviews of activity trackers found blood pressure effects small and often not significant.

Compared with before the program in the same people, and control participants, pooled together [8]; control groups the source does not describe [9]

The numbers and sources

Effect: Pedometer program participants lowered systolic blood pressure by 3.8 mmHg (95% CI 1.7 to 5.9), more so when starting pressure was higher [8]. Across 39 reviews of activity trackers, effects on blood pressure, cholesterol and HbA1c were typically small and often not significant [9].
Population: 2,767 participants in 26 studies [8]. 163,992 participants in 39 reviews [9].
Sources: [8], [9]
Caveats: The blood pressure figure pools randomized and observational studies [8]. The umbrella review gives no pooled blood pressure number in its abstract [9].

Pet ownershipModerate

Blood pressure

Mixed results: pet owners had slightly lower systolic pressure and heart rate in one meta-analysis but no less hypertension, and getting a dog did not change blood pressure.

Compared with people without pets [2]; people who delayed getting a dog and people with no plans to get one [6]

The numbers and sources

Effect: Pet owners had lower heart rate (mean difference -2.32), mean arterial pressure (-2.60) and systolic pressure (-1.69 mmHg, 95% CI -3.06 to -0.31), with no difference in diastolic pressure (-0.23) or in the share with hypertension [2]. Systolic and diastolic pressure, resting heart rate and VO2max did not change after getting a dog [6].
Population: 26 studies [2]. 71 adults followed 8 months [6].
Sources: [2], [6]
Caveats: The pooled differences are between groups measured once, and are small. The controlled study was small [6].

Sleep durationModerate

Sleep extension and blood pressure, glucose and insulin

Mixed results: pooled sleep extension trials found no overall change in blood pressure, fasting glucose or insulin resistance. Blood pressure fell in people with raised pressure.

Compared with control or baseline conditions the source does not describe [14]; control or baseline conditions the source does not describe [15]

The numbers and sources

Effect: Sleep extension lengthened total sleep by 51 minutes (95% CI 39 to 63) and did not change blood pressure overall. In people with pre-hypertension or stage 1 hypertension, 24-hour systolic pressure fell 7.8 mmHg (95% CI -10.6 to -4.9) and diastolic pressure 4.2 mmHg. Pooled effects on fasting glucose and insulin resistance were not significant [14]. A systematic review of 7 extension studies reported better direct and indirect measures of insulin sensitivity, less appetite and less desire for sweet and salty foods [15].
Population: 13 studies from 11 trials, 3 days to 6 weeks, adults without a sleep disorder [14]. 138 participants in 7 studies, mostly short sleepers [15].
Sources: [14], [15]
Caveats: Small and short trials. The blood pressure fall is a subgroup result. The review of 7 studies gave no pooled estimates [15].

Food and drink

Dark chocolateStrong

Blood pressure

Shown to lower blood pressure by about 1.8 mmHg with flavanol-rich cocoa products in short trials, by about 4 mmHg systolic in people with hypertension.

Compared with flavanol-free products or low-flavanol cocoa powder [2]; control conditions the abstract does not describe [3]

The numbers and sources

Effect: Across 40 comparisons, flavanol-rich cocoa products lowered systolic pressure by 1.76 mmHg (95% CI 0.43 to 3.09) and diastolic by 1.76 mmHg (95% CI 0.94 to 2.57) over 2 to 18 weeks. Systolic pressure fell 4 mmHg in people with hypertension and did not change significantly in people with normal blood pressure. Removing trials with authors employed by industry sponsors made the effect slightly smaller [2]. A second meta-analysis found diastolic pressure fell 1.60 mmHg (95% CI 0.43 to 2.77), with larger effects above 50 mg of epicatechin a day [3].
Population: 1,804 mainly healthy adults in 35 trials [2]. 1,297 participants in 42 trials of up to 18 weeks [3].
Sources: [2], [3]
Caveats: Unexplained heterogeneity lowered the Cochrane rating to moderate quality, and unblinded trials tended to show larger effects [2]. Many trials used cocoa drinks or extracts rather than chocolate bars.

SaltStrong

Blood pressure, from eating less salt

Shown to lower blood pressure, by about 4 mmHg systolic for a 130 mmol daily sodium cut, more in people with hypertension and in older adults.

Compared with usual sodium intake [3]; usual salt intake [4]; a high-sodium diet [5]; higher sodium intakes in the same trials [6]

The numbers and sources

Effect: Across 133 trials, a 130 mmol cut in 24-hour urinary sodium lowered systolic pressure by 4.26 mmHg (95% CI 3.62 to 4.89) and diastolic by 2.07 mmHg (1.67 to 2.48). Each 50 mmol cut lowered systolic pressure 1.10 mmHg, with larger falls in older people, non-white populations and those with higher starting pressure. Trials under 15 days showed less than half the effect of longer trials [3]. A Cochrane review of 34 trials found a 4.4 g a day salt cut lowered systolic pressure 5.39 mmHg in people with hypertension and 2.42 mmHg in people without it [4]. In white participants, cutting sodium from 203 to 65 mmol a day lowered systolic pressure 5.71 mmHg with hypertension and 1.14 mmHg without it, and did not change diastolic pressure in people with normal blood pressure [5]. A dose-response analysis of 85 trials found a roughly straight-line relation between sodium intake and blood pressure from 0.4 to 7.6 g a day, with no flattening at the lowest intakes [6].
Population: 12,197 adults in 133 randomized trials [3]. 3,230 adults in 34 trials of at least 4 weeks [4]. About 10,000 white participants in up to 96 trials, plus black and Asian participants in smaller sets of trials [5]. 85 trials of at least 4 weeks [6].
Sources: [3], [4], [5], [6]
Caveats: The same trials appear in several of these reviews. Effects in people without hypertension were small, and evidence in black and Asian participants was rated low to moderate quality [5]. Heterogeneity was high in [4].

SaltStrong

Blood pressure, with salt substitutes

Shown to lower systolic blood pressure by about 4.4 to 4.8 mmHg with potassium-enriched salt substitutes in place of regular salt, in two meta-analyses.

Compared with regular salt [7][8]; usual salt [9]

The numbers and sources

Effect: A Cochrane review found low-sodium salt substitutes lowered systolic pressure by 4.76 mmHg (95% CI 3.50 to 6.01, 20 trials) and diastolic pressure by 2.43 mmHg (95% CI 1.36 to 3.50, 19 trials) [7]. A network meta-analysis of 34 trials found moderate-potassium, low-sodium substitutes (25% to 40% potassium chloride) lowered systolic pressure by 4.39 to 4.64 mmHg [8]. In older adults with normal blood pressure in care homes, new hypertension was less frequent with a salt substitute over 2 years (11.7 against 24.3 per 100 person-years, HR 0.60, 95% CI 0.39 to 0.92), with a net systolic difference of 8.0 mmHg [9].
Population: 21,414 adults in 20 trials [7]. 37,063 participants in 34 trials in 15 countries, mean age 62 [8]. 611 Chinese older adults with normal blood pressure in 48 care facilities [9].
Sources: [7], [8], [9]
Caveats: The two reviews share most of their trials. All trials excluded people for whom extra potassium is known to be harmful [7]. The care home analysis was post hoc, and a salt company supplied the salt [9].

DairyModerate

High blood pressure

Linked to 6% lower high blood pressure for each extra 200 g a day of milk or low-fat dairy, in pooled cohorts.

Compared with 200 g a day less milk or low-fat dairy, in a dose-response model [9]

The numbers and sources

Effect: Each extra 200 g a day was linked to lower hypertension for low-fat dairy (RR 0.94, 95% CI 0.90 to 0.98) and milk (RR 0.94, 95% CI 0.92 to 0.97). Total dairy showed a nonlinear association [9].
Population: Cohort studies within a review of 42 articles [9].
Sources: [9]
Caveats: Observational. One review.

GrapefruitModerate

Blood pressure

Shown to lower systolic blood pressure by about 2.4 mmHg in overweight adults, in one meta-analysis of 3 small trials.

Compared with control groups in 3 randomized trials [2]; a control diet low in fruits and vegetables [4]

The numbers and sources

Effect: Pooled across 3 trials, systolic blood pressure fell more with grapefruit than with controls (MD -2.43 mmHg, 95% CI -4.77 to -0.09, I² 0%) [2]. In the 6-week trial, blood pressure did not differ between groups, and systolic pressure fell 3.21 mmHg from baseline in the grapefruit group (P = .03) [4].
Population: 250 overweight and obese adults in 3 trials [2]. 74 overweight adults [4].
Sources: [2], [4]
Caveats: The upper end of the pooled interval sits close to zero, and the trials were short [2]. The 6-week trial found no difference against its control group [4].

Green teaModerate

Blood pressure

Shown to lower blood pressure by about 2 mmHg in pooled trials of green tea drinks and extracts, with no separate result for the drink.

Compared with control groups, including placebo capsules, water and low-catechin green tea drinks [5]

The numbers and sources

Effect: Systolic pressure fell by 1.98 mmHg (95% CI 1.01 to 2.94) and diastolic by 1.92 mmHg (95% CI 0.68 to 3.17). Falls were larger in people with stage 1 hypertension (3.52 mmHg systolic) and were seen only in trials that removed the effect of caffeine, used lower polyphenol doses or lasted longer [5].
Population: 1,367 participants in 13 randomized trials, several using green tea drinks [5].
Sources: [5]
Caveats: The pooled trials mixed drinks and capsules. Polyphenol doses ranged from 208 to 1,207 mg a day [5].

Fermented foodsModerate

Blood pressure

Mixed results: kefir did not lower blood pressure in a meta-analysis of 7 trials, while kimchi lowered it about 3.5 mmHg after excluding outlying trials.

Compared with people who did not drink kefir [6]; control groups the abstract does not describe [9]

The numbers and sources

Effect: Kefir did not change systolic (WMD -1.76 mmHg, 95% CI -5.21 to 1.69) or diastolic pressure [6]. After excluding heterogeneous studies, kimchi lowered systolic pressure by 3.48 mmHg (95% CI -5.95 to -1.01) and diastolic by 2.68 mmHg [9].
Population: 385 adults in 7 kefir trials [6]. 205 participants in 5 kimchi trials [9].
Sources: [6], [9]
Caveats: The kimchi result depends on excluding some studies [9]. Both reviews pooled few small trials.

Red meat and processed meatModerate

Cholesterol and blood pressure in trials

Not shown to change cholesterol or blood pressure against all comparison diets combined. Against plant proteins, LDL fell less on red meat. Against fish, it fell more.

Compared with diets replacing red meat with legumes, soy or nuts, fish, poultry, mixed animal protein, refined carbohydrates or the usual diet [4]

The numbers and sources

Effect: Against all comparison diets combined, red meat made no significant difference to total, LDL or HDL cholesterol, apolipoproteins A1 and B, or blood pressure, and gave a smaller fall in triglycerides (0.065 mmol/L). Against high-quality plant proteins, total cholesterol fell 0.264 mmol/L less and LDL 0.198 mmol/L less. Against fish, LDL fell 0.173 mmol/L more and HDL 0.065 mmol/L more. Against refined carbohydrates, triglycerides fell 0.181 mmol/L more [4].
Population: 1,803 participants in 36 randomized trials [4].
Sources: [4]
Caveats: The result depends on what replaced the red meat. Trials measured blood markers, not heart events.

CoffeeModerate

Blood pressure

Not shown to change blood pressure in pooled trials. Cohorts linked coffee to no change or slightly lower risk of high blood pressure.

Compared with less or no coffee [14]; people drinking no coffee [15]

The numbers and sources

Effect: Pooled randomized trials found no significant change in systolic (-0.55 mmHg, 95% CI -2.46 to 1.36) or diastolic pressure (-0.45 mmHg, 95% CI -1.52 to 0.61). Pooled cohorts found no link with developing hypertension (RR 1.03, 95% CI 0.98 to 1.08) [14]. A later dose-response analysis of cohorts found 2% lower hypertension risk per daily cup (RR 0.98, 95% CI 0.98 to 0.99), and 0.92 at 6 cups a day [15].
Population: 10 randomized trials of at least 1 week and 5 cohorts [14]. 243,869 people and 58,094 hypertension cases in 10 cohort studies [15].
Sources: [14], [15], [12]
Caveats: The trial evidence was rated low in quality [14]. Caffeinated coffee raised blood pressure within the hour in a single-morning trial [12], and caffeine on its own raises it slightly (see the caffeine page).

NutsModerate

Blood pressure

Not shown to lower blood pressure in a meta-analysis of 61 tree nut trials.

Compared with control arms without nuts [4]

The numbers and sources

Effect: Tree nut intake had no statistically significant effect on blood pressure, while cholesterol, ApoB and triglycerides fell [4].
Population: 2,582 adults without cardiovascular disease in 61 trials [4].
Sources: [4]
Caveats: Trials were designed around blood lipids, and the abstract gives no blood pressure estimate.

Olive oilModerate

Blood pressure

Not shown to lower blood pressure in pooled trials of adults with normal or mildly raised blood pressure.

Compared with control groups in randomized trials [13]

The numbers and sources

Effect: Pooled trials found no significant change in systolic (mean difference -0.48 mmHg, 95% CI -1.76 to 0.80) or diastolic pressure (-0.47 mmHg, 95% CI -1.33 to 0.39) [13]. A review of high-polyphenol olive oil reported blood pressure improvements only in subgroups and single trials [10].
Population: 17 randomized trials, 14 of olive oil and 3 of flaxseed oil, in adults with normal blood pressure or at risk of hypertension [13].
Sources: [13], [10]
Caveats: The pooled analysis combined olive oil and flaxseed oil trials. The authors call for longer, higher-dose trials [13].

Diets

DASH dietStrong

Blood pressure

Shown to lower systolic blood pressure about 3 to 7 mmHg against control diets in adults, about 11 mmHg in people with hypertension in the original feeding trial.

Compared with control diets that were not DASH-like, otherwise not described [1]; control diets the source does not describe [2], [3]; a control diet low in fruits, vegetables and dairy, with fat typical of the US diet [4]; a control diet typical of US intake, eaten at high sodium [5]

The numbers and sources

Effect: In 30 RCTs (5,545 participants), DASH lowered systolic blood pressure by 3.2 mmHg (95% CI -4.2 to -2.3) and diastolic by 2.5 mmHg (95% CI -3.5 to -1.5), at moderate certainty. Hypertension status did not change the effect, and it was larger in trials with sodium above 2,400 mg a day and in people under 50 [1]. In 17 RCTs (2,561 participants), systolic fell 6.74 mmHg (95% CI -8.25 to -5.23) and diastolic 3.54 mmHg (95% CI -4.29 to -2.79) [2]. In 20 trials (1,917 participants), systolic fell 5.2 mmHg (95% CI -7.0 to -3.4) and diastolic 2.6 mmHg (95% CI -3.5 to -1.7) [3]. In the original 8-week feeding trial, DASH lowered blood pressure by 5.5/3.0 mmHg against a typical US control diet, 11.4/5.5 mmHg in 133 people with hypertension and 3.5/2.1 mmHg in 326 without [4]. In DASH-Sodium, DASH with low sodium gave systolic pressure 7.1 mmHg lower in people without hypertension and 11.5 mmHg lower in people with hypertension than the control diet with high sodium [5].
Population: Adults with and without hypertension in 30 trials [1], 17 trials [2] and 20 trials of 2 to 24 weeks [3]; 459 adults with systolic pressure under 160 mmHg and diastolic 80 to 95 mmHg [4]; 412 adults [5].
Sources: [1], [2], [3], [4], [5]
Caveats: Heterogeneity was high in [2] (I² 78.1% for systolic). Effect sizes differ about twofold between reviews. In the two NEJM feeding trials, all food was supplied and body weight was held steady, so they measure the diet eaten as written [4][5].

Keto dietStrong

Blood pressure

Shown to lower diastolic blood pressure about 1.4 mmHg more than comparison diets in adults.

Compared with diets that did not restrict carbohydrates to the level of ketosis

The numbers and sources

Effect: Diastolic pressure fell 1.43 mmHg more than on low-fat diets (95% CI 0.37 to 2.49, 1,298 patients) [1] and 1.41 mmHg more than on control diets (95% CI 0.26 to 2.57) [2]. In type 2 diabetes, systolic pressure fell 2.85 mmHg (95% CI 0.71 to 4.99) and diastolic 1.40 mmHg (95% CI 0.13 to 2.66) [7].
Population: Adults with overweight or obesity [1][2]; adults with type 2 diabetes in 29 trials [7].
Sources: [1], [2], [7]
Caveats: Differences of 1 to 3 mmHg are small. Heterogeneity was 49.1% for diastolic pressure in [2].

Paleo dietStrong

Blood pressure

Shown to lower blood pressure more than comparison diets in adults, by about 3 to 4 mmHg diastolic and 4 to 6 mmHg systolic.

Compared with control diets the source does not describe [1], [3]; a control group the source does not describe [5]

The numbers and sources

Effect: Systolic pressure fell 4.24 mmHg more (95% CI 1.38 to 7.11) and diastolic 2.95 mmHg more (95% CI 1.18 to 4.72) [1]. In people with metabolic disorders, systolic fell 5.89 mmHg (95% CI 1.86 to 9.97) and diastolic 4.01 mmHg (95% CI 1.80 to 6.21) [5]. Diastolic fell 3.28 mmHg in 19 RCTs (95% CI 2.01 to 4.55) [3].
Population: Adults in 8 RCTs [1]; adults with metabolic disorders in 10 RCTs [5]; adults in 19 RCTs [3].
Sources: [1], [3], [5]
Caveats: In [1], the blood pressure result changed when single studies were removed.

Alternate-day fastingModerate

Blood pressure

Shown to lower systolic and diastolic blood pressure against control in a pooled analysis of 7 trials, with no difference from daily calorie restriction in a year-long trial.

Compared with eating every day, with no fasting days or daily fasting window

The numbers and sources

Effect: Systolic (p 0.003) and diastolic (p 0.007) blood pressure fell against control [2]. Blood pressure did not differ between alternate-day fasting and daily calorie restriction at 6 or 12 months [3].
Population: 269 adults in 7 trials [2]; 100 adults with obesity [3].
Sources: [2], [3]
Caveats: The review reports p values without the size of the fall [2].

Intermittent fastingModerate

Blood pressure

Shown to lower blood pressure by about 2 mm Hg against unrestricted eating in adults without diabetes, more in those starting with high pressure.

Compared with eating every day, with no fasting days or daily fasting window

The numbers and sources

Effect: Time-restricted eating without calorie restriction lowered systolic pressure by 1.79 mm Hg (95% CI -3.30 to -0.27), diastolic by 1.75 mm Hg (95% CI -3.07 to -0.43) and heart rate by 2.19 beats a minute, with larger falls when baseline pressure was high [11]. An early 8-hour window plus calorie restriction lowered diastolic pressure 4 mm Hg (95% CI -8 to 0) more than calorie restriction with a 12-hour window [6]. Blood pressure did not differ between an 8-hour window, calorie restriction and control in type 2 diabetes [12], nor between time-restricted eating with calorie restriction and calorie restriction alone [3].
Population: 11 trials with 653 adults without diabetes [11]; 90 adults with obesity over 14 weeks [6]; 75 adults with type 2 diabetes [12]; 139 adults with obesity [3].
Sources: [3], [6], [11], [12]
Caveats: The pooled falls are small. When calorie intake was matched, blood pressure did not differ [3].

Mediterranean dietModerate

Blood pressure, cholesterol and blood sugar markers

Shown to lower blood pressure about 3/2 mmHg and LDL cholesterol and HbA1c slightly in adults, at low to moderate certainty.

Compared with no or minimal intervention, and another dietary intervention the abstract does not name [2]; a low-fat diet, a low-carbohydrate diet and the American Diabetes Association diet [7]; control diets the source does not describe [10]

The numbers and sources

Effect: Against no or minimal intervention, systolic blood pressure fell 2.99 mmHg (95% CI -3.45 to -2.53) and diastolic 2.0 mmHg (95% CI -2.29 to -1.71) at moderate quality, with little or no effect on LDL or HDL. Against other diets, LDL fell 0.15 mmol/L (95% CI -0.27 to -0.02) and triglycerides 0.09 mmol/L, with little or no effect on blood pressure [2]. In 9 RCTs (1,337 people at raised risk of diabetes or heart disease), HbA1c fell 0.18 points (95% CI -0.35 to -0.01), LDL 0.10 mmol/L and triglycerides 0.20 mmol/L. Fasting glucose, insulin, insulin resistance and HDL did not change [10]. Over 12 months or more, blood pressure and lipids changed about as much as on comparator diets [7].
Population: Healthy adults and adults at high cardiovascular risk [2]; adults at increased risk of type 2 diabetes and heart disease [10]; adults with overweight or obesity [7].
Sources: [2], [7], [10]
Caveats: The Cochrane review rated the lipid evidence low or very low quality for most comparisons [2]. The changes are small next to drug effects.

Nordic dietModerate

Blood pressure

Shown to lower systolic blood pressure about 4 mmHg against usual diets in adults, though one trial found no change.

Compared with control groups the source does not describe [1]; a usual or habitual Nordic diet, a usual Western diet or general healthy eating advice [2]; the participants' usual Western diet, in Swedish adults [3]; an average Nordic diet with the same calories [4]; an average Danish diet, with all food supplied free [5]

The numbers and sources

Effect: In 4 RCTs, systolic blood pressure fell 3.97 mmHg (95% CI -6.40 to -1.54) and diastolic 2.08 mmHg (95% CI -3.43 to -0.72) [1]. A second meta-analysis reported a small systolic reduction without a figure in its abstract [2]. In a 26-week Danish trial, systolic fell 5.1 mmHg (95% CI -8.2 to -2.1) and diastolic 3.2 mmHg more than on an average Danish diet [5]. In NORDIET, systolic fell 6.6 mmHg, which was no longer significant after adjusting for weight loss [3]. SYSDIET found no blood pressure change [4].
Population: Adults with cardiometabolic risk factors [1][2]; 181 Danish adults with central obesity [5]; adults with mild high cholesterol [3] or metabolic syndrome [4].
Sources: [1], [2], [3], [4], [5]
Caveats: Few trials. Part of the effect may come from weight loss [3].

High-protein dietModerate

Blood pressure and triglycerides

Mixed results: triglycerides fell more on higher-protein diets in two meta-analyses, while blood pressure fell slightly in one and did not differ in another.

Compared with diets with less protein

The numbers and sources

Effect: Across 74 trials, higher-protein diets lowered systolic (SMD -0.21, 95% CI -0.32 to -0.09) and diastolic blood pressure (SMD -0.18, 95% CI -0.29 to -0.06) and triglycerides (SMD -0.51, 95% CI -0.78 to -0.24). Total and LDL cholesterol, C-reactive protein and HbA1c did not change [6]. In energy-restricted trials, triglycerides fell 0.23 mmol/L more (95% CI 0.12 to 0.33), and blood pressure and total, LDL and HDL cholesterol did not differ [1].
Population: Adults in 74 trials [6] and in 24 weight-loss trials [1].
Sources: [1], [6]
Caveats: Restricting to studies at lower risk of bias reduced the triglyceride effect [6]. In type 2 diabetes, blood pressure changes were not significant [8].

Low-carb dietModerate

Heart risk factors

Mixed results: triglycerides and cholesterol ratio improved in some trials, while a Cochrane review found little to no difference in LDL or blood pressure over 1 to 2 years.

Compared with diets with more carbohydrate

The numbers and sources

Effect: Over 1 to 2 years, diastolic blood pressure and LDL cholesterol showed little to no difference from balanced-carbohydrate diets, with and without type 2 diabetes [1]. Over 2 years, the ratio of total to HDL cholesterol fell 20% on a low-carbohydrate diet and 12% on a low-fat diet [5]. In type 2 diabetes, triglycerides improved substantially at 6 months and less at 12 months. LDL cholesterol worsened at 12 months by an amount judged clinically important but not statistically significant [6].
Population: Adults with overweight or obesity [1][5]; adults with type 2 diabetes [6].
Sources: [1], [5], [6]
Caveats: Mean baseline LDL and diastolic pressure were in the normal range across the Cochrane trials [1]. No trial here measured heart attacks or strokes.

Vegetarian dietModerate

Blood pressure

Mixed results: one meta-analysis found systolic blood pressure about 5 mmHg lower, while trials in people with diabetes found no change.

Compared with control diets the source does not describe [2]; omnivorous diets, in 7 controlled trials and 32 observational studies [7]; omnivorous Adventists in the same cohort, white adults in the USA and Canada [8]

The numbers and sources

Effect: In 7 controlled trials (311 participants), vegetarian diets lowered systolic blood pressure by 4.8 mmHg (95% CI -6.6 to -3.1) and diastolic by 2.2 mmHg against omnivorous diets. In 32 observational studies, systolic was 6.9 mmHg lower [7]. In 9 RCTs in people with diabetes, blood pressure did not change [2]. In an Adventist cohort sample, lacto-ovo vegetarians had lower odds of hypertension than omnivores (OR 0.57, 95% CI 0.36 to 0.92) [8].
Population: Adults, mean age 44.5 in the trials [7]; adults with diabetes [2]; 500 white Adventists, observational [8].
Sources: [2], [7], [8]
Caveats: Few small trials. Heterogeneity was high in the observational studies (I² 91.4%) [7]. Part of the cohort difference reflected lower body weight [8].

Vegan dietModerate

Blood pressure

Not shown to lower blood pressure in trials of adults with overweight or type 2 diabetes. One cohort linked vegan diets to less hypertension.

Compared with the participants' habitual diet, unchanged, in 3 trials in US and New Zealand adults, and in 8 trials a National Cholesterol Education Program diet, American Diabetes Association and Korean Diabetes Association diets, a portion-controlled diet, a Mediterranean diet, a low-fat diet, a healthy omnivorous diet or a high-carbohydrate lacto-ovo vegetarian diet [1]; omnivorous Adventists in the same cohort, white adults in the USA and Canada [4]

The numbers and sources

Effect: Blood pressure did not change in 11 trials [1]. In an Adventist cohort sample, vegans had lower systolic (by 6.8 mmHg) and diastolic (by 6.9 mmHg) blood pressure than omnivores, and lower odds of hypertension (OR 0.37, 95% CI 0.19 to 0.74) [4].
Population: Adults with overweight or type 2 diabetes [1]; 500 white Adventists, observational [4].
Sources: [1], [4]
Caveats: The cohort link was partly explained by lower body weight [4].

Devices and wearables

Wrist heart rate and calorie estimatesStrong

Blood pressure, cholesterol and blood sugar

Mixed results: effects of activity trackers on blood pressure, cholesterol and HbA1c across systematic reviews were typically small and often not statistically significant.

The numbers and sources

Effect: Effects of activity trackers on blood pressure, cholesterol and glycosylated hemoglobin were typically small and often non-significant, as were effects on quality of life and pain [18].
Population: Participants in 39 systematic reviews and meta-analyses [18].
Sources: [18]
Caveats: The abstract gives no pooled figures for these outcomes [18].

Lab markers

HOMA-IRStrong

High blood pressure

Linked to more new high blood pressure, 43% more in the highest than the lowest HOMA-IR group across pooled cohorts of the general population.

The numbers and sources

Effect: Comparing the highest with the lowest category, the pooled adjusted risk ratio for developing hypertension was 1.43 (95% CI 1.27 to 1.62) for HOMA-IR and 1.54 (1.34 to 1.76) for fasting insulin [14]. A second meta-analysis found a hazard ratio for hypertension of 1.35 (95% CI 1.15 to 1.59) with a higher HOMA-IR [12].
Population: 55,059 participants with 10,230 cases of hypertension in 11 prospective studies [14]. 215,878 participants in 38 cohort studies across all outcomes [12].
Sources: [14], [12]
Caveats: Observational, so confounding by body weight and other factors cannot be ruled out. Each study defined its own highest and lowest groups.

Fasting insulin and HOMA-IRModerate

High blood pressure and stroke

Linked to more new high blood pressure, 63% more in the highest against the lowest fasting insulin group across 10 cohorts, and not linked to stroke.

The numbers and sources

Effect: Comparing the highest with the lowest group of fasting insulin, the pooled relative risk was 1.63 (95% CI 1.35 to 1.97) for hypertension and 1.18 (0.87 to 1.60) for stroke. Each 50 pmol/L higher fasting insulin went with a relative risk of 1.25 (1.14 to 1.36) for hypertension and 0.999 (0.99 to 1.01) for stroke [11].
Population: 36,617 people and 4,491 cases in 10 studies of hypertension, and 27,887 people and 1,550 cases in 7 studies of stroke [11].
Sources: [11]
Caveats: Observational, from one meta-analysis, so confounding by body weight and other factors cannot be ruled out.

Magnesium levelModerate

Blood pressure, in cohorts and in trials of magnesium supplements

Shown in trials: magnesium supplements that raised serum magnesium by 0.05 mmol/L lowered blood pressure by about 2 mm Hg over a median 3 months.

The numbers and sources

Effect: In a meta-analysis of randomized, double-blind, placebo-controlled trials, magnesium at a median 368 mg a day for a median 3 months lowered systolic blood pressure by 2.00 mm Hg (95% CI 0.43 to 3.58) and diastolic by 1.78 mm Hg (95% CI 0.73 to 2.82). Serum magnesium rose by 0.05 mmol/L (95% CI 0.03 to 0.07), about 0.12 mg/dL, compared with placebo [10]. In cohorts, each 0.1 mmol/L higher circulating magnesium was associated with a 4% lower incidence of hypertension (RR 0.96, 95% CI 0.94 to 0.99) [8].
Population: 2,028 adults with normal or high blood pressure in 34 trials [10]. 14,876 participants in 3 cohorts, 3,149 cases, average follow-up 6.7 years [8].
Sources: [10], [8]
Caveats: The trial authors note that residual heterogeneity may remain. The trials ran for months and measured blood pressure, not heart attack or stroke [10].

Uric acid (serum urate)Moderate

High blood pressure

Linked to more new high blood pressure, 13% higher risk per 1 mg/dL in a meta-analysis of 18 cohorts.

The numbers and sources

Effect: In a meta-analysis of 18 prospective cohorts, hyperuricemia went with more new hypertension (adjusted risk ratio 1.41, 95% CI 1.23 to 1.58), and each 1 mg/dL higher uric acid with a risk ratio of 1.13 (1.06 to 1.20). The association was larger in younger study populations and tended to be larger in women [14]. An umbrella review found the hypertension association in observational studies and in some trials of intermediate outcomes, and not statistically significant in Mendelian randomization studies [12].
Population: 55,607 participants in 18 cohorts [14].
Sources: [14], [12]
Caveats: Observational. Later and US-based studies reported smaller effects [14]. The absence of a genetic association leaves open whether uric acid causes the rise [12].