Blood Pressure

Lower pressure prevents strokes and heart attacks at every starting level tested; how much you gain depends on your overall risk

Lowering systolic pressure prevents strokes, heart attacks and heart failure, in proportion to the drop. Drug trials show a similar relative benefit for each millimetre across every starting pressure they tested, from below 120 to 170 mmHg and above. The benefit holds in people with or without existing heart disease. (Blood Pressure Lowering Treatment Trialists Collaboration, 2021) In placebo-controlled trials it also brought modestly fewer cases of dementia (see A second outcome: dementia). (inferred from Peters et al., 2022)

How much that is worth to you depends on your overall cardiovascular risk, not on the reading alone. The trialists say treatment decisions should turn on a person’s overall risk of future cardiovascular events, not on the reading alone. (Blood Pressure Lowering Treatment Trialists Collaboration, 2021) So the same drop buys a person at high risk far more than it buys a person at low risk. (Blood Pressure Lowering and Cardiovascular Events)

Drug trials carry the proof of fewer events; lifestyle changes mostly carry proof of a lower reading. For most lifestyle changes, the payoff in events is assumed: it follows only if the drug-trial rate applies to pressure lowered by other means. Among the diet changes appraised here for their blood-pressure effect, only salt substitution has a trial that counted fewer events alongside a measured drop in pressure (Neal et al., 2021); the Mediterranean diet’s event result from PREDIMED is not tied to blood pressure. For cutting heavy drinking, a risk model projects the events; no trial counted them (Roerecke et al., 2017).

Among weight-loss routes, one drug has measured an event benefit, and only in people who already had cardiovascular disease. In people with type 2 diabetes, Look AHEAD’s lifestyle weight loss initially moved blood pressure and other markers but did not detectably reduce events (Semaglutide for Cardiovascular Risk in Obesity, Does Weight Loss Reduce Cardiovascular Events).

Structured exercise and, in heavy drinkers, cutting back on alcohol give the largest well-measured lifestyle drops in people with high blood pressure. Weight loss and DASH-style eating with less salt lower pressure too, but the held lifestyle weight-loss figures are modest (DiRECT -3.4 mmHg (Lean et al., 2019); the best named diets, Atkins and DASH, about 5 mmHg at 6 months, largely gone by 12 (Ge et al., 2020)), and the larger weight-loss drops come from drugs.

The alcohol effect appears only in heavier drinkers; people at two drinks a day or fewer saw no significant change (Roerecke et al., 2017). Exercise lowers pressure by an amount similar to drugs in hypertensive trials. But no trial has compared the two directly, and no exercise trial has counted events (Naci et al., 2018). Most exercise trials were at high risk of bias and the comparison is indirect, so the certainty is low to moderate .

Do not add the drops together. Where it was tested (DASH plus less salt), two good changes together lowered pressure by less than their sum, and for other pairs nobody has measured the combination.

Garlic supplements, beetroot nitrate, magnesium and relaxation practices such as mindfulness and yoga report drops that are either small or rest on weak trials. The biggest headline numbers in this group either shrink under stricter tests (yoga, slow breathing, and mindfulness in unmedicated patients) or rest on weak trials (garlic). None has event data. Here, the bigger the reported drop, the weaker the evidence behind it tends to be. (Chaddha et al., 2019; Chen et al., 2024; Dibaba et al., 2017; Geiger et al., 2025; Ma et al., 2025; Norouzzadeh et al., 2025; inferred from Ried et al., 2008; Zhang et al., 2016)

Who should act depends on who you are. For someone with high pressure and high overall risk, lowering it is one of the largest health levers available; drugs carry the proof of fewer events, and for other routes the event payoff is assumed. Lifestyle changes still lower the reading on top of drugs, by less in the DASH trials, and no held trial measures how much the two add together for events. For a low-risk person with normal pressure, the same millimetres buy little absolute benefit. That is a finding in its own right: a reason to stop optimizing, not a sign that blood pressure does not matter. (Layer 1 - Ranking Interventions for a Stratum)

This appraisal covers diet, weight, physical activity, alcohol, specific nutrients and supplements, and stress management, with blood-pressure drugs as the realistic alternative. Choosing, dosing and adjusting a drug is a prescriber’s job and is not covered. Salt and potassium appear here only in outline; their detail lives in Sodium.

All of this rests on one step being well proven: that a lower reading actually turns into fewer events.

A 5 mmHg drop cuts major events by about a tenth, at every starting pressure tested

Two steps, two kinds of evidence

Any blood-pressure lever has to clear two separate steps. First it must lower the reading; then the lower reading must prevent strokes, heart attacks and heart failure. A trial that measures only the first step is measuring a surrogate, and GRADE treats that as a real cost: relying on a surrogate outcome generally means rating the evidence down by one or two levels. (Schünemann et al., n.d.) A drop in pressure counts only once the second step is shown (Surrogate Outcomes).

The evidence on blood pressure is strongest where it matters least. A BP response can be randomised and measured in weeks in small trials; strokes and deaths cannot. So the reading is usually the better-graded quantity. WHO’s sodium guideline shows the pattern: high certainty for the effect on blood pressure, lower (mostly low or very low) for the effect on hard outcomes. (inferred from World Health Organization, 2012) A reader who acts on the strongest evidence is pulled toward the reading by the grading itself. For blood pressure, unusually, the second step has also been measured directly, in drug trials.

What the drug trials show

The Blood Pressure Lowering Treatment Trialists’ Collaboration (BPLTTC) pooled individual data from 344,716 people in 48 randomised trials. (Blood Pressure Lowering Treatment Trialists Collaboration, 2021) Its headline: «a 5 mm Hg reduction of systolic blood pressure reduced the risk of major cardiovascular events by about 10%, irrespective of previous diagnoses of cardiovascular disease, and even at normal or high–normal blood pressure values.» (Blood Pressure Lowering Treatment Trialists Collaboration, 2021)

• Without prior cardiovascular disease: HR 0.91 (95% CI 0.89-0.94) per 5 mmHg lower SBP.
• With prior cardiovascular disease: HR 0.89 (0.86-0.92).
• By component, per 5 mmHg: stroke down 13%, heart failure 13%, ischaemic heart disease 8%, cardiovascular death 5%.

(Blood Pressure Lowering Treatment Trialists Collaboration, 2021)

The effect did not fade at lower starting pressures. Across seven baseline-SBP bands, from below 120 to 170 mmHg and above, the trialists found «no reliable evidence of heterogeneity», and in no band did the hazard ratio rise above 1.0. (Blood Pressure Lowering Treatment Trialists Collaboration, 2021) That is a statement about the relative effect of each millimetre, not about how many millimetres a given lever delivers. Exercise lowered pressure less in trials with normal starting readings; for DASH the held evidence is mixed. The section Overall cardiovascular risk decides who gains most; starting pressure, age and kidney function come next takes that up.

The proof comes from drugs; for lifestyle it is mostly assumed

Every one of these event figures comes from drug-lowered blood pressure. For a lifestyle lever, the event payoff is assumed to follow at the same rate per millimetre, not observed. Blood Pressure Lowering and Cardiovascular Events states the limit plainly: the magnitude applies to pharmacological lowering and should not be transferred silently to another route. The sections that follow therefore present each lever on that condition: if the drug-trial rate applies.

One diet change has counted events alongside a measured drop in pressure: salt substitution. In SSaSS, 20,995 high-risk adults in rural China given a potassium-enriched salt substitute had SBP 3.34 mmHg lower and fewer strokes (RR 0.86, 95% CI 0.77 to 0.96). (Neal et al., 2021) At BPLTTC’s 13% per 5 mmHg for stroke, that pressure drop predicts an RR near 0.91. The observed point estimate is larger than the blood-pressure channel predicts, but the prediction sits inside the confidence interval, so the excess may be chance. If real, it would fit an effect of the electrolytes beyond blood pressure, or a sustained BP fall larger than the one measured. (inferred from Neal et al., 2021) The derivation, and the sodium-potassium detail, are in Sodium.

Over a mean 4.74 years the substitute also cut major cardiovascular events (rate ratio 0.87, 0.80 to 0.94) and deaths (0.88, 0.82 to 0.95); 72.6% of participants had already had a stroke (Neal et al., 2021). The result carries two limits. The substitute lowers sodium and raises potassium at once, so the benefit belongs to the combined switch, not to either electrolyte alone. And with a population this sick, the absolute benefit will be much smaller at lower risk. (inferred from Neal et al., 2021)

The relative effect is constant; your absolute gain is not

The absolute gain changes from person to person, and it scales with overall cardiovascular risk. GRADE notes that the relative effect is usually similar across baseline risks while the case for treatment differs by them (Schünemann et al., n.d.). The absolute reduction equals baseline risk times (1 minus the hazard ratio). Applied to a 5 mmHg drop, using the primary-prevention HR of 0.91:

• At 10% ten-year risk: about 9 fewer major events per 1,000 people over 10 years (CI roughly 6 to 11).
• At 2.5% ten-year risk: about 2 fewer per 1,000 (roughly 1.5 to 3).

(inferred from Blood Pressure Lowering Treatment Trialists Collaboration, 2021; European Society of Cardiology, 2021) The two baselines are illustrations; where they fall on ESC’s age-banded thresholds is in Overall cardiovascular risk decides who gains most; starting pressure, age and kidney function come next (SCORE2 Baseline Risk and the ESC Treatment Thresholds).

Three assumptions sit inside those numbers, and the reader can reject any of them. The hazard ratio is assumed to hold constant over a full 10 years. SCORE2’s endpoint (fatal and non-fatal myocardial infarction and stroke) (European Society of Cardiology, 2021) omits the heart failure that BPLTTC’s major-event composite counts (Blood Pressure Lowering Treatment Trialists Collaboration, 2021), so the hazard ratio is applied to a different composite. And the BP drop is assumed to come from a drug, because that is where the rate was measured. (inferred from European Society of Cardiology, 2021)

The trialists themselves warn against reading their absolute numbers as policy. They «caution against using the reported absolute risk differences from clinical trials for making policy decisions», because randomised trials are strongest on relative effects. (Blood Pressure Lowering Treatment Trialists Collaboration, 2021) Their trial populations ran at about 32 major events per 1,000 person-years in the control arms without prior CVD, well above a low-risk person’s rate. (Blood Pressure Lowering Treatment Trialists Collaboration, 2021) The same paper says a uniform relative effect does not mean every patient is worth treating; the decision should rest on overall risk, side effects, cost and preference (Baseline Risk and the Relative-Absolute Split). (Blood Pressure Lowering Treatment Trialists Collaboration, 2021)

At low risk, a proven effect still buys little

Of the four interventions the wiki compares for low-risk people, blood-pressure lowering is the clearest test, because its relative benefit on hard events in primary prevention is proven. Semaglutide and lifestyle weight loss are unproven in that group, and statins, whose relative benefit is real, fall below action thresholds at low risk. Even so, a proven 10% per 5 mmHg on a low baseline is a small absolute gain (Cardiometabolic Interventions and Hard CV Outcomes in Low-Risk People).

preDIVA shows how small. Its 6-year nurse-led vascular-care programme (median follow-up 6.7 years) in 3,526 unselected adults aged 70-78 lowered SBP by only 2.06 mmHg and found no effect on cardiovascular disease (HR 1.06, 0.86-1.31). (van Charante et al., 2016) Through the BPLTTC rate, a 2 mmHg contrast predicts an HR of about 0.96, far too small for a trial of this size to detect. So preDIVA’s null says more about how little pressure it moved than about who was treated. (inferred from van Charante et al., 2016)

A second outcome: dementia

Lowering blood pressure probably lowers the risk of dementia modestly. An individual-participant meta-analysis of five double-blind placebo-controlled drug trials (28,008 people) found an adjusted odds ratio of 0.87 (95% CI 0.75-0.99) for incident dementia, with a mean BP reduction of 10/4 mmHg. Dementia occurred in 2.9% of the treated and 3.3% of the placebo groups. (Peters et al., 2022) That is about 4 fewer cases per 1,000 over the trials’ follow-up, likely an underestimate because the trials stopped early on their cardiovascular endpoints and dementia accrues more slowly. (inferred from Peters et al., 2022) In observational data, people with untreated hypertension had a 42% higher dementia risk than healthy controls (HR 1.42; 1.15-1.76); among treated people the excess was attenuated or lost. (Livingston et al., 2024)

These five trials sit inside the BPLTTC pool, so the dementia result extends the outcome list; it is not independent support for the per-millimetre effect (Dementia Prevention and Modifiable Risk Factors). (Blood Pressure Lowering Treatment Trialists Collaboration, 2021; inferred from Peters et al., 2022) As with heart attacks and strokes, the dementia benefit is shown for drug-lowered pressure; that a lifestyle-lowered drop reaches dementia is assumed. (inferred from Peters et al., 2022)

If each millimetre is worth roughly the same, the practical question becomes which changes deliver the most millimetres, and how sure we are that they do.

Exercise and cutting heavy drinking give the largest well-measured lifestyle drops; weight loss is best measured with drugs

Structured exercise and, in heavy drinkers, cutting back on alcohol have the best-measured lifestyle effects on the reading; weight loss lowers it too, with the larger figures coming from drug trials. Each lowers systolic pressure by a few to about 9 mmHg in the people studied. None of the lifestyle versions has trial evidence of fewer events of its own; each figure below is a BP drop (see The proof comes from drugs; for lifestyle it is mostly assumed).

Exercise

In hypertensive trial participants, structured exercise lowered systolic pressure about as much as drug treatment did, judged indirectly. The evidence is one network meta-analysis of 391 RCTs: 197 trials of exercise (10,461 people) and 194 of antihypertensive drugs (29,281 people). Endurance and dynamic resistance training were among the exercise types tested. No trial compared exercise with a drug directly, so every exercise-versus-drug figure is inferred across separate sets of trials. (Naci et al., 2018)

The headline numbers mislead unless you match the starting pressure. Across all populations, the drug trials’ systolic change was -8.80 mmHg (95% CrI -9.58 to -8.02) and the exercise trials’ -4.84 (-5.55 to -4.13), a -3.96 mmHg (-5.02 to -2.91) advantage to drugs. But the exercise trials began at a mean of 132 mmHg, and the drug trials started above 150. Restricted to people starting at 140 or higher, exercise changed systolic pressure by -8.96 mmHg (-10.27 to -7.64), and the gap to drugs was 0.18 (-1.35 to 1.68), no difference. That matched estimate rests on 56 exercise trials and 3,508 people. (Naci et al., 2018)

Exercise’s drop grew with the starting pressure, so the unmatched contrast compared two different quantities. This is a between-trial subgroup comparison, not an interaction test (Baseline Risk and the Relative-Absolute Split). (inferred from Naci et al., 2018)

The authors hedge the match twice, and the hedges stand. Their conclusion holds only «Assuming equally reliable estimates» (Naci et al., 2018). The estimates are not equally reliable: 17 of 20 sampled exercise trials were at high risk of bias from lack of blinding, and the authors call the network’s linking assumption tenuous because the two sets of trials started at different pressures. (Naci et al., 2018) The authors also note a bias running the other way: they may have underestimated exercise, because they excluded ineffective or suboptimal drug doses but no exercise trials on that basis (Naci et al., 2018). Exercise cannot be blinded, so that weakness is built in.

Single exercise types lowered pressure about equally; combining endurance and resistance training did best. Naci also compared exercise types with one another: «Overall, different types of structured exercise interventions achieved similar reductions from baseline (table 2).» «One excep- tion was the combination of endurance and resistance training, which was more effective in reducing baseline SBP than dynamic resistance alone (−2.98, 95% CrI −5.04 to −0.93).» (Naci et al., 2018) In trials of people starting at 140 mmHg or higher, the combination beat endurance, resistance and isometric training alike, and lowered systolic pressure by 13.51 mmHg (95% CrI -16.55 to -10.45) against control. (Naci et al., 2018)

Three cautions keep that edge provisional. Only 30 of the 197 exercise trials compared one type with another, so most type contrasts are network estimates inferred across trials. The type groups started at different pressures, a mean 125 mmHg in resistance trials and 135 in combination trials (Naci et al., 2018). And Naci itself flags how thin the combination evidence is: «For example, only a small subset of studies identified in our review tested the effectiveness of the combi- nation of endurance and dynamic resistance training. Although these types of interventions were often effective, especially in hypertensive populations, their frequency and duration were not consistently reported to determine the optimal formulation and dose of physical activity to maximise its benefit.» (Naci et al., 2018)

Since a higher start brings a larger drop, part of the all-populations edge may be baseline; the edge held in the 140-plus trials, which partly controls for that. Part may also be more total training rather than the mix itself. Among single types, the one a person will keep doing is a reasonable choice; for a hypertensive person who can do both, indirect evidence leans toward the combination (Is the Exercise Modality Doing Any Work).

The fair summary: a similar systolic drop in hypertensive trials, at low-to-moderate certainty, with no event data for exercise. Naci measured no strokes or heart attacks. Exercise also improves fitness, glucose control and weight, benefits a BP-only comparison leaves out (Physical Activity Dose and Mortality, Cardiorespiratory Fitness and Mortality). (inferred from Naci et al., 2018)

Alcohol

Cutting back lowers blood pressure only in people drinking three or more drinks a day. A 36-trial meta-analysis (2,865 participants) found that people at two drinks a day or fewer saw no significant BP change when they cut to near abstinence. Above that, the drop grew with the amount drunk. (Roerecke et al., 2017)

Usual intake before cutting	Systolic change (mmHg, 95% CI)
2 drinks/day or fewer	-0.18 (-1.02 to 0.66), not significant
3 drinks/day	-1.18 (-2.32 to -0.04)
4-5 drinks/day	-3.00 (-3.98 to -2.03)
6 or more drinks/day	-5.50 (-6.70 to -4.30)

(Roerecke et al., 2017)

Across all trials the systolic change averaged -3.13 mmHg (-3.93 to -2.32), with high heterogeneity that starting intake almost entirely explained. The trial strata averaged 15, 30, 49 and 76 g of alcohol a day, so the curve was measured across real drinking levels, not extrapolated. The authors rate the certainty of the dose-response as high. (Roerecke et al., 2017)

Genetic evidence agrees that alcohol raises pressure. A Mendelian randomization study uses gene variants that change how much people drink as a natural experiment. In Chinese men, systolic pressure rose about 4.3 mmHg (3.7 to 4.9) per 280 g a week; in women the variants did not raise intake. The same study found alcohol causally raises stroke risk. (Millwood et al., 2019)

The event payoff is modelled, not measured. Using a risk-assessment model, Roerecke projected more than 7,000 UK hospital admissions and 678 cardiovascular deaths avoided each year. The model assumed people kept the lower intake and that the benefit arrived without delay. (Roerecke et al., 2017) At the drug-trial rate set out in A 5 mmHg drop cuts major events by about a tenth, at every starting pressure tested, a heavy drinker’s 5.5 mmHg drop would predict roughly 10% fewer major cardiovascular events. That figure is the wiki’s arithmetic. It holds only if pressure lowered by drinking less transmits like drug-lowered pressure. (Blood Pressure Lowering Treatment Trialists Collaboration, 2021; inferred from Roerecke et al., 2017)

This BP lever is separate from the old claim that light drinking protects the heart. That protective dip in mortality curves disappears once researchers stop counting people who quit because they were already ill as abstainers, and genetic analyses do not reproduce it (Alcohol and Mortality and Vascular Disease).

Weight loss

No source held here gives a figure for mmHg per kilogram lost, so the evidence comes as separate results from drug trials and lifestyle trials. Weight loss also shifts diet, so its BP effect overlaps with the diet changes in DASH and less salt each lower pressure, and together less than their sum, which explains why the drops should not be added. (Does Weight Loss Reduce Cardiovascular Events)

The drug route

The weight-loss drugs lowered systolic pressure by roughly 3 to 8 mmHg, and only one trial has shown fewer events. In STEP-1, semaglutide in adults with obesity, without diabetes and mostly without heart disease lowered systolic pressure by 5.10 mmHg relative to placebo at 68 weeks. (Wilding et al., 2021) In SELECT, the same drug lowered systolic pressure by 3.31 mmHg in people with established cardiovascular disease, and cut major cardiovascular events by 20% (HR 0.80, 0.72 to 0.90). The benefit appeared early after treatment started, and the authors suggest changes beyond weight loss may explain part of it. They offer that mechanism as speculation, so no single marker, BP included, is known to carry the benefit. (Lincoff et al., 2023) (Semaglutide for Cardiovascular Risk in Obesity)

Tirzepatide, in two 52-week trials of people with obesity and sleep apnoea (469 participants, manufacturer-funded), changed systolic pressure at week 48 by -7.6 mmHg (-10.5 to -4.8) in one trial and -3.7 (-6.8 to -0.7) in the other. These trials had no cardiovascular endpoint; the authors say the ongoing SURMOUNT-MMO outcome trial may supply event data. (Malhotra et al., 2024)

The pressure drop lasts only as long as the drug. When STEP-1 participants stopped semaglutide and the lifestyle programme at week 68, they regained two-thirds of the weight they had lost within a year. By week 120 systolic and diastolic pressure were back to baseline. The trialists concluded that the findings «suggest ongoing treatment is required to maintain improvements in weight and health». (Wilding et al., 2022) For a drug, adherence is part of the effect.

The lifestyle route

Look AHEAD shows that markers can improve while events do not. In 5,145 adults with type 2 diabetes and overweight or obesity, an intensive diet-and-activity programme improved HbA1c and produced greater initial improvements in fitness and every cardiovascular risk factor except LDL cholesterol, blood pressure included. After a median 9.6 years, cardiovascular events were no lower (HR 0.95, 0.83 to 1.09). (Look AHEAD Research Group, 2013) Read Look AHEAD as a caution about trusting a bundle of surrogates, not as a refutation of BP transfer. The gap in weight between the arms shrank to 2.5 percentage points by the end, and no source held here gives a figure for how far the arms’ blood pressure actually differed. (Surrogate Outcomes)

A meta-analysis of 54 weight-loss RCTs (30,206 adults with obesity, almost all on low-fat weight-reducing diets) found the same on the heart: in the 24 trials (15,176 participants) that reported new cardiovascular events, the risk ratio was 0.93 (0.83 to 1.04), high-quality evidence (C. Ma et al., 2017). Look AHEAD is among Ma’s trials, so this generalizes its null rather than independently confirming it (Does Weight Loss Reduce Cardiovascular Events).

In DiRECT, a total-diet-replacement programme for people with short-duration type 2 diabetes lowered systolic pressure at two years (difference -3.43 mmHg, 95% CI -6.70 to -0.16), with fewer people on blood-pressure drugs (47.3% versus 60.1%). (Lean et al., 2019) (Total Diet Replacement and Type 2 Diabetes Remission)

Sleep apnoea: the device showed no detectable drop in pressure

Treating sleep apnoea with CPAP did not detectably lower systolic pressure in pooled trials: -0.20 mmHg (-2.29 to 1.89). (Yu et al., 2017) Tirzepatide’s 7.6 mmHg drop came in the same kind of patient, people with sleep apnoea. The weight-loss drug moved the reading; in these trials the device did not detectably. The tirzepatide trials enrolled heavier people (mean BMI about 39 versus 30), so the contrast is suggestive, not a head-to-head result (Sleep Apnea Treatment and Cardiovascular Risk). (Malhotra et al., 2024; inferred from Yu et al., 2017)

DASH and less salt each lower pressure, and together less than their sum

Food changes move blood pressure even when body weight stays put. In the DASH-Sodium feeding trial, investigators held every participant’s weight constant and still saw pressure fall on both diet and salt (Sacks et al., 2001). DASH is among the best-studied patterns for blood pressure, and it is the one lever where a trial tested what happens when two good changes are combined (DASH Diet and Blood Pressure).

DASH lowers systolic pressure by about 3 to 5 mmHg, depending on how you pool the trials

DASH means more fruit, vegetables, low-fat dairy and whole grains, and less red meat, sweets and saturated fat. Two meta-analyses of DASH against a control diet give two different systolic estimates, and the held evidence cannot say which is closer to the truth.

• Siervo pooled 20 RCTs (1917 people, 2-24 weeks): SBP -5.2 mmHg (-7.0 to -3.4), DBP -2.6 (-3.5 to -1.7), with high heterogeneity (I2 76%), a small LDL fall (-0.10 mmol/l) and no significant effect on glucose or HDL. (Siervo et al., 2014)
• Filippou pooled 30 RCTs (5545 people, mean follow-up about 15 weeks, average starting pressure 134/85 mmHg): SBP -3.2 mmHg (-4.2 to -2.3), DBP -2.5 (-3.5 to -1.5). Some trials lacked blinding or complete outcome data, so certainty is moderate: DASH probably lowers pressure by about this much. (Filippou et al., 2020)

The two figures are not the same quantity. Filippou compared pressure at the end of the trial; Siervo mostly compared change from baseline. Only 15 of Filippou’s 30 trials appear in Siervo. Nothing held separates the effect of the metric from the effect of the different trial sets, so this guide quotes both and pools neither. Filippou’s pool also admitted trials that paired DASH with salt restriction, weight loss or exercise, so part of its -3.2 may come from those add-ons. (inferred from Filippou et al., 2020; Siervo et al., 2014) Durability is unknown past roughly 15 weeks: Filippou states the short follow-up means the effect cannot be assumed to last longer. (Filippou et al., 2020)

The trials test the whole pattern, so no single food or nutrient can be credited. Siervo’s meta-regression found that the incidental sodium difference between the DASH and control arms did not predict the BP change (Siervo et al., 2014); a null meta-regression has little power, so this bounds sodium’s share rather than excluding it.

Do not expect a bigger DASH drop just because your pressure starts higher. The two pooled analyses disagree on whether the drop grows with baseline pressure (figures in Overall cardiovascular risk decides who gains most; starting pressure, age and kidney function come next). (inferred from Filippou et al., 2020; Siervo et al., 2014)

No trial has counted heart attacks or strokes on DASH. Cohorts link closer DASH-style eating to lower incident CVD (RR 0.80, 0.76-0.85) and stroke (RR 0.81, 0.72-0.92), in an umbrella review whose cohort layer spans 15 cohorts and 942,140 people; the umbrella review rates both associations low certainty. (Chiavaroli et al., 2019) The often-quoted figure of about 13% lower 10-year risk is a Framingham risk-score projection from the marker changes, not counted events (Siervo et al., 2014). A DASH score in a cohort is also likely to track other healthy habits.

The umbrella review’s own blood-pressure figure is Siervo’s estimate re-imported, so it is the same number, not a second confirmation of it. (inferred from Chiavaroli et al., 2019; Siervo et al., 2014) For DASH, the step from pressure to events is assumed, not measured. If the drug-trial rate from A 5 mmHg drop cuts major events by about a tenth, at every starting pressure tested applies to diet-lowered pressure, DASH’s few mmHg would buy a proportionate cut in events.

DASH plus less salt beats either alone, but falls short of the sum

DASH-Sodium randomized 412 adults with systolic pressure of 120-159 mmHg (41% with stage-1 hypertension, none on BP drugs) to DASH or a typical US diet, then fed each person high, intermediate and low sodium for 30 days each. All food was provided. Achieved urinary sodium was about 142, 107 and 65 mmol/d. (Sacks et al., 2001) The trial sits inside both the Siervo and Filippou pools, so it refines them rather than adding a second witness to the pooled DASH effect. (Filippou et al., 2020; inferred from Sacks et al., 2001; Siervo et al., 2014)

Each lever did less when the other was already in place:

• DASH lowered SBP by 5.9 mmHg (-8.0 to -3.7) against the control diet at high sodium, but only 2.2 (-4.4 to -0.1) at low sodium. (Sacks et al., 2001)
• Cutting sodium from high to low lowered SBP by 6.7 mmHg (-5.4 to -8.0) on the control diet, but only 3.0 (-1.7 to -4.3) on DASH. (Sacks et al., 2001)
• DASH with low sodium, against the control diet with high sodium, lowered SBP by 8.9 mmHg (-6.7 to -11.1) and DBP by 4.5 (-3.1 to -5.9). (Sacks et al., 2001)

Adding the two single effects from the same starting point (-5.9 and -6.7) would predict about -12.6 mmHg; the trial measured -8.9. (inferred from Sacks et al., 2001) The investigators put it directly: «The reductions in blood pressure caused by the combination of dietary interventions were smaller than they would have been if the effects of each dietary intervention were strictly additive (P<0.001 for the interaction).» (Sacks et al., 2001)

So estimate a combined DASH-and-salt effect from the -8.9 cell, never by adding two separate numbers. Sub-additivity is shown here only for DASH and salt, in one 30-day feeding trial; until other pairs are tested, assume their drops may not add either. No held trial tests weight loss, DASH and lower sodium together, so the drops in this section and the weight-loss figures should not be summed. (inferred from Sacks et al., 2001)

This is efficacy under controlled feeding, on blood pressure only, over 30-day periods. The authors note that «long-term health benefits remain to be demonstrated and will depend on the ability of people to make long-lasting dietary changes, including the consistent choice of lower-sodium foods.» (Sacks et al., 2001)

Sodium and potassium in one line each

• Cutting sodium lowers resting SBP by 3.39 mmHg (2.46 to 4.31) at high certainty, across 36 RCTs (World Health Organization, 2012), with a larger fall in people with hypertension (He et al., 2013). Direct evidence on heart attacks and strokes from cohorts is very low certainty (World Health Organization, 2012).
• More potassium lowered SBP by 5.32 mmHg (3.43 to 7.20) in 16 trials of people with hypertension, against 0.09 (-0.77 to 0.95) in 3 shorter trials of people without it; pooled across all adults the fall was 3.49 mmHg (1.82 to 5.15), at high certainty. (Aburto et al., 2013)
• Who should cut: for a person with hypertension eating a lot of salt, both sides of the sodium-policy dispute agree to reduce it. For a person with normal pressure at moderate intake (3-6 g of sodium a day), the evidence is insufficient either way -> Should Sodium Reduction Be Population-Wide or Targeted.

The figures, dose-response, and the kidney and drug contraindications live in the Sodium deliverable (Sodium Intake and Blood Pressure, Potassium Intake and Blood Pressure).

Other diet patterns: small drops, fading drops, or weak evidence

• Mediterranean diet: about -3/-2 mmHg against minimal intervention, at moderate certainty but from 2 RCTs and 269 people; against another healthy diet, an imprecise -1.5 mmHg at low certainty. (Rees et al., 2019) Its event evidence comes from PREDIMED, not from pressure (Mediterranean Diet and Cardiovascular Events).
• Named diet programmes: low-carbohydrate and low-fat diets each lowered SBP by about 5 mmHg against usual diet at 6 months (5.14 moderate certainty, 5.05 low), above the review’s 3 mmHg importance bar. Unlike the DASH-Sodium drops, these came with weight loss, of 4.63 and 4.37 kg. And «Macronutrient diet related improvements in both blood pressure and blood lipids disappeared almost completely at 12 months.» (Ge et al., 2020) Differences between named diets were trivial (Named Diet Programs Compared).
• Vegan diets: SBP -2.56 mmHg (-4.66 to -0.45) against omnivorous diets, from observational data at very low certainty. (Landry et al., 2024) (Vegetarian Dietary Patterns and Mortality)
• Coffee: no significant effect on blood pressure in trials; the pooled estimate was marginally favourable but did not reach significance. (Poole et al., 2017) A possible signal in slow caffeine metabolisers belongs to Overall cardiovascular risk decides who gains most; starting pressure, age and kidney function come next (Coffee Consumption and Health).
• Time-restricted eating: one controlled-feeding trial of an early eating window in 8 prediabetic men, weight held stable, reported SBP/DBP -11/-10 mmHg (p=.03) (Sutton et al., 2018), and its authors warn that morning-only readings may overstate it. A free-living trial of a late window found no between-group BP difference (Lowe et al., 2020). These are different exposures, so neither result is a claim about eating windows in general (Time-Restricted Eating).

Fibre, whole grains and sweeteners barely move the reading

Extra fibre lowers systolic pressure by about 1 mmHg; whole grains and non-sugar sweeteners show no clear effect. These come from general-population trials. Reynolds admitted people with mild to moderate hypertension but excluded those on medications affecting the outcomes, and gives no hypertensive-only estimate (Reynolds et al., 2019), so the figures are not directly comparable with the hypertensive figures for exercise or garlic.

Trial contrast	Systolic change (mmHg, 95% CI)	Certainty
More vs less dietary fibre, general population (15 trials)	-1.27 (-2.50 to -0.04)	Moderate
More vs less fibre, people with diabetes or prediabetes (12 trials)	-1.86 (-4.85 to 1.12), not significant	—
More vs less whole grain (8 trials)	-1.01 (-2.46 to 0.44)	Moderate
Whole-grain vs refined-grain diets (SACN)	0.2 (-1.6 to 2.0), p=0.85	Moderate (SACN: no effect)
Non-sugar sweeteners (14 trials)	-1.33 (-2.71 to 0.06)	Moderate

(Reynolds et al., 2019) (A. N. Reynolds et al., 2020) (Scientific Advisory Committee on Nutrition, 2015) (World Health Organization, 2023)

SACN, the UK’s advisory committee, put fibre and blood pressure in its table of insufficient trial evidence, which it keeps separate from its table of inconsistent evidence (Scientific Advisory Committee on Nutrition, 2015). SACN normally files evidence as insufficient when fewer than three studies of the same type qualify, and sometimes when more exist but are of poor quality (Scientific Advisory Committee on Nutrition, 2015). So the fibre entry means too few or too weak qualifying trials, not conflicting ones. Its chapter summary states: «8.161 Randomised controlled trials indicate there is no effect of total or mixed dietary fibre intake on the cardiovascular or type 2 diabetes mellitus risk factors considered here.» (Scientific Advisory Committee on Nutrition, 2015) With blood pressure filed as insufficient, that null need not cover it.

Splitting fibre by type, SACN found one effect: «8.79 An effect is demonstrated for higher oat bran and β-glucan consumption on reducing systolic blood pressure (-2.86 mmHg, 95% CI -4.87, -0.85; p<0.01).» (Scientific Advisory Committee on Nutrition, 2015) The gap with Reynolds may come from that split and different inclusion rules rather than from reading the same trials differently.

SACN’s whole-grain null rests mostly on one trial, which contributed 92% of the systolic estimate, raised whole-grain intake by less than a serving a day, and also produced weight differences between groups (Scientific Advisory Committee on Nutrition, 2015).

Sweeteners split by study design. In WHO’s review, cohorts linked sweetener use to hypertension (HR 1.13, 95% CI 1.09 to 1.17, low certainty), while the trials showed no rise in systolic pressure (World Health Organization, 2023). WHO names one explanation: «Reverse causation suggests that those already at elevated risk of disease initiated or increased use of NSS because of their risk status, rather than NSS leading to increased risk in otherwise healthy or low-risk individuals.» (World Health Organization, 2023) It then limits that reading: «However, efforts taken by the authors to address reverse causation and confounding in most studies suggest that these phenomena are not the sole causes of observed associations and may not even play a significant role in many of the studies.» (World Health Organization, 2023)

In its body-weight discussion WHO notes that most sweetener trials lasted 3 months or less (World Health Organization, 2023). If the blood-pressure trials were as short, they could not detect a harm that takes years to show up as hypertension, so the short-term question has a moderate-certainty null and the long-term one stays open. None of the three exposures moves pressure enough to rank as a blood-pressure lever. At the drug-trial rate, fibre’s -1.27 mmHg would predict roughly 2.5% lower relative risk of cardiovascular events, and only if the drop is sustained and diet-lowered pressure transmits like drug-lowered pressure. Their value, where they have any, runs through other outcomes (Dietary Fibre and Health, Whole Grains Refined Grains and Pulses, Non-Sugar Sweeteners). (Blood Pressure Lowering Treatment Trialists Collaboration, 2021; inferred from A. Reynolds et al., 2019)

Supplements and relaxation practices report big drops that shrink under stricter tests

Beyond whole diets, several single foods, supplements and relaxation practices report systolic drops as large as a drug’s, on much weaker evidence. Among the non-drug levers, the biggest headline drops come from the least-tested levers. Garlic, mindfulness courses and yoga post some of the biggest office-BP numbers, yet none has been tested on strokes or heart attacks. Each mind-body drop fails a stricter test somewhere: against an active control (yoga), in low-risk-of-bias trials (slow breathing) or in unmedicated patients (mindfulness). (Chaddha et al., 2019; Geiger et al., 2025; X. Ma et al., 2025; inferred from Ried et al., 2008)

The only non-drug lever with a trial that counted fewer events alongside a measured blood-pressure drop, salt substitution, lowered systolic pressure by a modest 3.34 mmHg (95% CI -4.51 to -2.18) in SSaSS (Neal et al., 2021), so the pattern rests on that one contrast. (PREDIMED’s Mediterranean diet cut events too, but its benefit is not tied to a blood-pressure drop.) Drugs sit outside the pattern: they lowered systolic pressure by 8.80 mmHg (95% CrI -9.58 to -8.02) against control across Naci’s drug trials, which started from higher pressures (class means of roughly 156-160 mmHg, against 125-135 in its exercise trials) (Naci et al., 2018), and their effect on events is proven (Blood Pressure Lowering and Cardiovascular Events). (Blood Pressure Lowering Treatment Trialists Collaboration, 2021; inferred from Neal et al., 2021)

So rank these levers on certainty and event proof, not on the headline mmHg. No trial of any lever in this section counted cardiovascular events; the event payoff is assumed, as set out in The proof comes from drugs; for lifestyle it is mostly assumed (Surrogate Outcomes).

Dietary nitrate (beetroot juice)

Nitrate supplements, given as beetroot juice or nitrate salts, lower systolic pressure by about 4.4 mmHg (95% CI -5.9 to -2.8) against placebo, pooled from 16 small crossover trials (254 people). Diastolic pressure did not move significantly (-1.1 mmHg, CI crossing zero). The trials lasted between 2 hours and 15 days. Most enrolled young, healthy participants, mostly men, and seven were exercise-performance studies; removing those did not change the systolic result. Beetroot juice and nitrate salt gave about the same drop, so the nitrate itself does the work. (Siervo et al., 2013)

A larger, GRADE-rated dose-response analysis of 75 trials (1,823 people) puts the dose needed for at least a 2 mmHg systolic fall at about 8 mmol of nitrate a day. It rated the acute and short-term systolic effects moderate certainty at best, medium-term systolic effects low, and risk of bias serious for every BP outcome. The slope looked steeper in people with hypertension, but that comes from a subgroup analysis and stays a candidate finding. (Norouzzadeh et al., 2025) The second analysis includes the first one’s trials, so it refines the estimate rather than confirming it independently. (inferred from Norouzzadeh et al., 2025; Siervo et al., 2013)

The drop was not confirmed on 24-hour ambulatory monitoring in the three trials that used it, which included older, overweight and obese people with type 2 diabetes, the group a BP lever would most need to help. (Siervo et al., 2013) Ambulatory monitoring is the more reliable measure.

The proposed mechanism needs bacteria: nitrate «may be reduced by facultative anaerobic bacteria on the dorsal surface of the tongue to nitrite», which is further reduced to nitric oxide (Siervo et al., 2013). If that route carries the effect, daily antibacterial mouthwash or antibiotics would be expected to blunt it; no trial in the pool tested that.

To link the drop to events, the authors borrow a Joint National Committee estimate for a systolic fall «of at least 5 mm Hg (as observed here)» (Siervo et al., 2013). The pooled fall was 4.4 mmHg, below that threshold, and the source does not say what kind of pressure difference the Joint National Committee estimate rests on.

Overall: a real short-term effect on resting pressure, at low-to-moderate certainty, with no event data (Dietary Nitrate and Blood Pressure).

Garlic supplements

In people with hypertension, garlic supplements lowered systolic pressure by about 8 mmHg against placebo: -8.38 mmHg (-11.13 to -5.62) across 4 trials of 12-23 weeks (Ried et al., 2008), and -8.12 mmHg (-10.95 to -5.28) in a 2025 update of 12 all-hypertensive trials (X. Ma et al., 2025). The two figures are not independent confirmations. The later analysis re-pools the older trial base, and a third of its trials come from the earlier review’s first author. (X. Ma et al., 2025)

In people with normal pressure the drop was small and not significant (-2.28 mmHg, -4.61 to 0.05). Almost all trials tested standardised dried garlic powder at 600-900 mg a day, not garlic eaten as food, and within that narrow 600-900 mg range dose was not associated with the size of the effect. (Ried et al., 2008) The trials measured office BP only, and none recorded a cardiovascular event. Publication bias remains unresolved: the 2025 paper’s results and limitations sections contradict each other on it (Garlic and Blood Pressure). (X. Ma et al., 2025)

Magnesium supplements

Magnesium supplements lower systolic pressure by about 2 mmHg against placebo. A pool of 34 double-blind trials (2,028 adults, median 368 mg a day for 3 months) found -2.00 mmHg (-3.58 to -0.43) systolic and -1.78 mmHg diastolic. (Zhang et al., 2016) A second analysis, restricted to people with insulin resistance, diabetes or heart disease, found -2.22 mmHg between groups. (Dibaba et al., 2017) That second analysis shares 8 of its 11 trials with the first, so it is the same evidence sliced by stratum, not a replication. Its larger, often-quoted figure is the before-and-after change in the supplemented arm alone, not a placebo-controlled effect. Compared like with like, the two strata respond about the same. That is a between-pool comparison. (Dibaba et al., 2017; inferred from Zhang et al., 2016)

Zhang also ran formal interaction tests within its own pool. It compared subgroups of its trials on the same placebo-controlled quantity, a firmer check than comparing two pools. The effect did not differ between people with normal and raised pressure (P for interaction 0.79 systolic, 0.59 diastolic). Point estimates were larger with low baseline magnesium and in people on medication, but neither interaction was significant (systolic P 0.21 and 0.11) (Zhang et al., 2016). So a person with hypertension should not expect a bigger drop. Correcting a deficit is the likeliest place for a larger effect, and it is unproven.

In cohorts, each extra 100 mg a day of dietary magnesium tracked lower stroke (RR 0.93, 0.89-0.97) and type 2 diabetes risk (RR 0.81, 0.77-0.86), but not total cardiovascular disease (RR 0.99, 0.88-1.10) or coronary heart disease (RR 0.92, 0.85-1.01) (Fang et al., 2016). The authors «cannot exclude the possibility that other nutrients and/or dietary components correlated with dietary magnesium may have been responsible, either partially or entirely, for the observed associations» (Fang et al., 2016), so eating magnesium-rich foods is already covered by the DASH advice. Magnesium is a small refinement, not a main lever, with no event data (Dietary Magnesium and Cardiometabolic Outcomes).

Mindfulness courses, yoga and slow breathing

Yoga and slow breathing lose their effect under a stricter test; mindfulness holds against active controls but shows no drop in unmedicated patients (Stress Management and Cardiometabolic Health). (Chaddha et al., 2019; inferred from Chen et al., 2024; Geiger et al., 2025)

• Mindfulness courses (MBSR / MBCT, 6-8 weeks): across all 12 trials (715 people) systolic pressure fell -9.12 mmHg (-12.18 to -6.05), with extreme heterogeneity (I2 92%), mostly unblinded trials and moderate to high risk of bias. The drop held against active controls (-8.78, -15.98 to -1.57) but was not significant against a waitlist. It was largest in trials of medicated patients (-15.30, -17.85 to -12.74), and in the 2 trials of unmedicated patients there was none (+0.53, -1.89 to 2.95). The authors suggest the course may work partly by improving drug adherence, so unmedicated patients might benefit less. (Chen et al., 2024)
• Yoga: against an active control the systolic drop was -4.16 mmHg (-10.76 to 2.44), not significant, and on 24-hour ambulatory monitoring the systolic effect disappeared, though only 4 trials used it and the authors note this limits its power. Against a waitlist the headline is -7.95 mmHg (-10.24 to -5.66; 26 trials, 2,007 people), at very low GRADE certainty. (Geiger et al., 2025)
• Slow breathing (mostly device-guided): in the low-risk-of-bias trials the systolic drop was -2.14 mmHg (-4.76 to 0.48), not significant. The headline across 17 trials (1,017 people) is -5.62 mmHg (-7.86 to -3.38), on office readings over 4-8 weeks, with high heterogeneity. (Chaddha et al., 2019)

Treat these practices as an adjunct at most. They cannot replace medication or salt reduction.

Urate-lowering drugs

In people with raised uric acid, urate-lowering drugs (mostly allopurinol, 15 trials) lowered systolic pressure by a standardized mean difference of 0.321 (0.145 to 0.497). That is a standardized effect, not mmHg, and it shrank to 0.211 after correcting for publication bias. The often-cited about 3 mmHg is borrowed from an older analysis of 10 longitudinal studies, so it should be discounted. (Qu et al., 2016)

The trials also cannot say whether lower urate does the work. Allopurinol blocks an enzyme that may affect blood vessels directly (Qu et al., 2016); a trial of a drug that lowers urate by another route would settle it (Uric Acid Lowering and Blood Pressure). It is a drug, so whether to use it is a prescriber’s decision, outside this appraisal.

The dietary link sits upstream. With high certainty, sugar-sweetened drinks raise uric acid even at equal calories (Ayoub-Charette et al., 2021) (Free Sugars Intake). But no study has followed the whole chain from cutting sugary drinks to lower urate, lower pressure and fewer events in people with normal uric acid.

Put every lever, and the drugs, on the same three axes: size of drop, certainty, and proof of fewer events. The ranking then changes with the person.

Ranked side by side, drugs and the big lifestyle levers overlap; supplements trail

Lined up on the same three axes, the levers sort into rough tiers rather than a strict order. A blood-pressure drug, structured exercise, cutting heavy drinking, a weight-loss drug, and DASH eating with less salt each lower systolic pressure by a few to about 9 mmHg in the people studied. They differ in how sure we are of the drop and in whether anyone has counted the events it prevents. Only blood-pressure drugs have measured fewer events at every starting pressure tested. Among the other levers, only a salt substitute and, in people who already had heart disease, semaglutide have counted events alongside a measured BP drop; PREDIMED’s Mediterranean-diet events are not tied to pressure. Supplements and relaxation practices report drops of the same headline size on far weaker evidence. None has been tested on events.

The table ranks on effect size x certainty x proof of fewer events, never on the headline mmHg alone. Among the non-drug levers, the biggest surrogate drops come with the weakest event evidence (the pattern set out in Supplements and relaxation practices report big drops that shrink under stricter tests) -> Blood Pressure Lowering and Cardiovascular Events. Every figure is a drop in systolic pressure against a control group, and each is detailed in the section named in the notes.

Lever (instead of…)	Typical SBP drop (who was studied)	BP certainty	Proof of fewer events	What it takes
BP drug (vs none)	-8.80 mmHg (drug trials, baseline >150)	High	Measured	Daily and usually lasting; side effects, cost
Structured exercise (vs none)	-8.96 at baseline >=140; -4.84 across all trials	Low-moderate	Not measured; assumed	Has to be kept up; also lifts fitness, glucose, weight
Cut heavy drinking	No significant drop at <=2 drinks/day; -5.50 at >=6/day when intake roughly halved	High	Modelled, not measured	Applies only at >=3 drinks/day
DASH plus low sodium (vs usual diet, high sodium)	-8.9 combined (one 30-day feeding trial)	One trial, n=412	Not measured	Less than the two drops added together
DASH diet (vs usual diet)	About 3-5 (two pooled estimates)	Moderate	Cohort associations only, low	Trials averaged ~15 weeks; durability unknown
Semaglutide (weight-loss drug)	-5.10 (STEP-1, no CVD); -3.31 (SELECT, CVD)	Moderate-high	Measured in established CVD only	GI side effects; BP gain reverses on stopping
Lifestyle weight loss	Initially moved BP and other markers	—	Not reduced in Look AHEAD	See Weight loss
Less sodium; more potassium	See Sodium	High (BP)	Salt substitute: measured	Potassium: shown in hypertensive trials only (between-trial contrast)
Garlic, magnesium, nitrate, mind-body	Yoga null vs active control; garlic -8.38 (hypertensives, office BP)	Low	None trialled	Adjunct at most

Notes, by row. Drug: -8.80 (-9.58 to -8.02) from trials whose participants started above 150 mmHg (Naci et al., 2018); the per-mmHg event reduction is set out in A 5 mmHg drop cuts major events by about a tenth, at every starting pressure tested (Blood Pressure Lowering Treatment Trialists Collaboration, 2021). Exercise: indirect network estimate, no head-to-head trial, mostly unblinded trials; figures in Exercise and cutting heavy drinking give the largest well-measured lifestyle drops; weight loss is best measured with drugs (Naci et al., 2018). Alcohol: (Roerecke et al., 2017).

DASH: Filippou -3.2 and Siervo -5.2, not pooled into one number (Filippou et al., 2020) (Siervo et al., 2014); the -8.9 factorial result is from DASH-Sodium (Sacks et al., 2001); detail in DASH and less salt each lower pressure, and together less than their sum.

Semaglutide: (Wilding et al., 2021) (Lincoff et al., 2023). Supplements and mind-body: (Geiger et al., 2025) (Ried et al., 2008); detail in Supplements and relaxation practices report big drops that shrink under stricter tests.

Lifestyle weight loss: Look AHEAD initially moved blood pressure and other markers without reducing events -> Does Weight Loss Reduce Cardiovascular Events; no pooled mmHg figure for it is held here. Sodium and potassium: the salt substitute in SSaSS cut events (Neal et al., 2021); magnitudes and the kidney-disease caution live in Sodium.

Read the event column with one rule. For every lifestyle row, a fewer-events figure is what you get if the drug-trial rate per mmHg applies to a drop reached by other means. And the same mmHg is worth more at higher overall risk (see The relative effect is constant; your absolute gain is not) -> Layer 1 - Ranking Interventions for a Stratum.

A drug can stand in for the blood-pressure part of a lever, not the whole lever

When a safe, effective drug can deliver the same blood-pressure drop, a lifestyle lever’s extra value for blood pressure shrinks. Its value on other outcomes does not. Exercise and weight loss work on more than one channel. Exercise also improves fitness, glucose control and weight, which a BP-only comparison ignores -> Physical Activity Dose and Mortality. Diet and weight change also act on what drives the pressure up in the first place, while a drug manages the reading. The comparison also has to subtract the drug’s own costs: tolerability, lifelong use, and leaving the underlying driver in place -> Layer 1 - Ranking Interventions for a Stratum.

The trialists behind the per-mmHg evidence list the same costs. Treating raised pressure, they write, requires weighing «an individual’s overall risk of future cardiovascular events, potential risk of adverse effects, the cost of treatment, and patient preferences» (Blood Pressure Lowering Treatment Trialists Collaboration, 2021).

The weight-loss drug carries its own costs. In SELECT, adverse events led 16.6% of patients on semaglutide to stop it permanently, against 8.2% on placebo, including gastrointestinal disorders in 10.0% against 2.0% of patients (Lincoff et al., 2023). And the blood-pressure gain is a state you maintain, not one you bank. When people in STEP-1 stopped the drug and the lifestyle programme, blood pressure returned to baseline within a year (figures under Weight loss). The smaller weight loss in SELECT than in STEP-1 reflects older, sicker participants and a different time point, not a weaker drug -> Semaglutide for Cardiovascular Risk in Obesity.

Drugs and lifestyle stack, and the choice between them is yours

Lifestyle changes still lower pressure in people already taking blood-pressure drugs. In 7 DASH trials of treated hypertensives, systolic pressure fell -2.1 mmHg (-2.5 to -1.8), against -5.9 (-9.9 to -1.8) in 8 trials of untreated hypertensives; the difference was not significant (P=0.07) (Filippou et al., 2020). The point estimates suggest the add-on is smaller than the effect in untreated people, but the trials do not show it. The overlap of trial lists across Filippou’s low-effect subgroups makes the cause unclear -> DASH Diet and Blood Pressure.

For a person with hypertension, structured exercise is a credible companion to a drug, and a candidate alternative where a drug is declined or not yet indicated -> Blood Pressure Lowering and Cardiovascular Events. Which route to take, or how to combine them, is a decision for the person and their prescriber. The figures here size the options; they do not choose among them. Picking an agent or a dose is a prescriber’s job and outside this appraisal.

A population’s top killer is not automatically your top lever

High systolic pressure was the leading risk factor for deaths worldwide in 2019, «accounting for 10·8 million (9·51–12·1) deaths» (19.2% of all deaths that year), ahead of tobacco (GBD 2019 Risk Factors Collaborators, 2020). That ranking counts deaths attributed across a whole population, and it rises with how common the exposure is. It does not tell any one person how much they would gain by lowering their own pressure. That depends on their own overall risk -> Layer 1 - Ranking Interventions for a Stratum.

Who gains most depends first on overall risk, and only then on how high the pressure starts.

Overall cardiovascular risk decides who gains most; starting pressure, age and kidney function come next

Three different reasons can make a recommendation differ between people, and they need different evidence. Under route (a), baseline risk, the benefit per mmHg is the same for everyone, and the absolute gain grows with each person’s risk. That needs only a risk estimate, with no claim that the treatment works differently in anyone (Schünemann et al., n.d.). Route (b), effect modification, says a lever genuinely works better in some people. That needs positive interaction evidence; a plausible mechanism is not enough, and a contrast between separate sets of trials is not an interaction test. Route (c), contraindication, covers a harm that fires only in one group -> Baseline Risk and the Relative-Absolute Split.

Overall cardiovascular risk (route a)

Because the per-mmHg event reduction does not change with starting pressure, 10-year cardiovascular risk is the main stratifier. It needs no subgroup claim; the per-1,000 arithmetic is in A 5 mmHg drop cuts major events by about a tenth, at every starting pressure tested. The European Society of Cardiology (ESC) turns a SCORE2 10-year risk into an action with thresholds that rise with age:

10-year SCORE2 risk	under 50	50-69	70 and over
Low-to-moderate: treatment of risk factors generally not recommended	under 2.5%	under 5%	under 7.5%
High: treatment should be considered	2.5 to under 7.5%	5 to under 10%	7.5 to under 15%
Very high: treatment generally recommended	7.5% or more	10% or more	15% or more

(European Society of Cardiology, 2021)

So the same 8% means very high at 45 and high at 72. ESC-EAS’s 2025 lipid guidance uses different, non-age-banded cutoffs (very high at 20% or more, high at 10 to under 20%, moderate at 2 to under 10%). Under those, 8% is moderate: a risk category is specific to the guideline that assigns it (Mach et al., 2025). SCORE2 and its older-age version cover people aged 40 to 89 without known cardiovascular disease, and not people already on lipid-lowering drugs (Mach et al., 2025). The base model also excludes diabetes; SCORE2-Diabetes fills that gap (SCORE2-Diabetes Working Group and ESC Cardiovascular Risk Collaboration, 2023).

A person’s category flips on inputs that age and sex alone do not supply. Here is ESC’s chart for men aged 60-64 in its moderate-risk region (which includes Germany, Italy and Sweden), with non-HDL cholesterol of 4.0-5.9 mmol/L:

Assumed inputs	10-year risk read off the chart	ESC category (age 50-69)
non-smoker, SBP 140-159	10-11%	very high: treatment generally recommended
non-smoker, SBP 120-139	8-9%	high: treatment should be considered
smoker, SBP 140-159	15-17%	very high

(European Society of Cardiology, 2021)

Smoking, the blood-pressure band, cholesterol and region each move the answer, so no single row speaks for a stratum. ESC attaches two conditions. The charts apply only where risk factors are untreated or have been stable for several years, and risk should be adjusted upward as a person nears the next age band (European Society of Cardiology, 2021) -> SCORE2 Baseline Risk and the ESC Treatment Thresholds.

Normal versus raised pressure: two different questions

The relative benefit of each mmHg holds at normal pressure (see A 5 mmHg drop cuts major events by about a tenth, at every starting pressure tested).

The number of mmHg a lever delivers often looks smaller at normal pressure, but the evidence for that comes from comparing separate sets of trials. In Naci’s network analysis, exercise trials started from lower pressures than the drug trials. Restricting them to hypertensive participants nearly doubled the estimated drop (figures under Exercise) (Naci et al., 2018). A higher start may leave room for a larger fall, so two levers can only be compared at matched baselines (inferred from Naci et al., 2018)

Garlic’s drop is not significant in normotensives (Garlic supplements; Garlic and Blood Pressure). Sodium reduction lowers pressure more in hypertensives. Potassium lowered it in 16 hypertensive trials but not in 3 short normotensive trials (figures under Sodium and potassium in one line each). Each of these is a between-trial subgroup comparison with no interaction test, so each is a route (b) candidate, not a finding; details in Sodium. Magnesium, like DASH below, has a formal within-pool interaction test by blood-pressure status, and it found no significant interaction (all P > 0.05) (Zhang et al., 2016) (Magnesium supplements).

DASH is the exception. Filippou pooled 30 trials and found the same SBP fall in hypertensive and normotensive trials, -3.9 mmHg in each (P=0.96), and no difference between baselines of 140 and above versus below (P=0.70) (Filippou et al., 2020). Siervo’s earlier pooling found a larger fall at higher baseline pressure (Siervo et al., 2014). But that finding came from a between-trial meta-regression, a design prone to ecological confounding and, on change-from-baseline metrics, to regression to the mean (inferred from Siervo et al., 2014).

DASH-Sodium’s larger combined fall in its hypertensive participants is a subgroup comparison, not an interaction test (inferred from Sacks et al., 2001). The safe default is route (a): do not promise someone a bigger DASH drop because their pressure is higher (inferred from Filippou et al., 2020; Siervo et al., 2014) -> DASH Diet and Blood Pressure.

One argument for restricting sodium advice to hypertensives rests on a premise from Mente: that lowering pressure helps only above 140 mmHg systolic. Mente draws it from BP-lowering meta-analyses in people with diabetes, HOPE-3 and secondary-prevention trials (Mente et al., 2016). BPLTTC, a later and larger analysis, contradicts that premise. It does not settle the separate dispute over very low sodium intake (Blood Pressure Lowering Treatment Trialists Collaboration, 2021; inferred from Mente et al., 2016) -> Should Sodium Reduction Be Population-Wide or Targeted.

Age

For younger adults, a low 10-year risk understates the case. ESC notes that women under 50 and men under 40 are almost invariably at low 10-year risk, even when their risk factors sharply raise longer-term risk. It offers a lifetime-benefit frame, including a grid for a 10 mmHg SBP reduction, as a route into treatment from the lowest band (European Society of Cardiology, 2021).

ESC bounds that frame in the same passage: «CVD risk predictions, as well as predictions of lifetime benefit of risk factor treatment, are likely to be imprecise at very young age (<40 years). At that age, lipid-lowering and BP-lowering drug treatment are not usually considered, except for patients with FH or specific BP disorders. A healthy lifestyle that is maintained throughout life is more relevant for the very young.» (European Society of Cardiology, 2021) -> Lifetime Benefit - The Frame for Younger Adults.

At older ages the logic runs the other way. ESC sets higher risk thresholds from 70 because risk-factor gradients weaken with age and competing non-cardiovascular mortality rises (European Society of Cardiology, 2021). Lower pressure still pays at older ages, including for dementia (benefit under A second outcome: dementia; no U-shape at any age under Targeting systolic below 120 prevented more events in high-risk trial patients, with more side effects).

preDIVA is the trial in unselected 70-78-year-olds described under At low risk, a proven effect still buys little. Its authors read it as easing the worry that treating blood pressure in old age harms cognition: «Our results mitigate fears that antihypertensive treatment in older age groups evokes cognitive decline. Moreover, they show that such an intervention is safe, which is in accordance with findings from the recent SPRINT trial.» (van Charante et al., 2016) Because the trial moved systolic pressure by only about 2 mmHg, it tests that worry only weakly.

Its overall dementia result was null (HR 0.92, 0.71-1.19). But in a subgroup of adherent participants with untreated hypertension at baseline, dementia was lower (HR 0.54, 0.32-0.92) (van Charante et al., 2016). That is a hypothesis that the benefit lies where hypertension goes untreated, not a finding: adherence is measured after randomisation, so the subgroup is not a randomised comparison.

DASH’s apparent age effect is a hypothesis. Trials with a mean age under 50 showed a larger SBP fall (-4.9 vs -2.0 mmHg, P<0.001). Filippou calls this a hypothesis, warns that meta-regressions are cross-sectional tools, and notes it is unknown whether individual patients sat above or below the age cut-off (Filippou et al., 2020). Trial age is tangled up with sodium level and with one large trial that sits in every low-effect subgroup. So this does not license telling an older person that DASH works less for them (inferred from Filippou et al., 2020).

Kidney disease

Chronic kidney disease raises baseline risk (route a), so the same mmHg buys more. ESC’s 2025 lipid guidance places moderate CKD in its high-risk and severe CKD in its very-high-risk category, whatever the score (Mach et al., 2025). ESC’s 2021 prevention guideline grades CKD the same way (European Society of Cardiology, 2021). Preventing kidney disease itself is covered in Chronic Kidney Disease and Modifiable Exposures.

Kidney disease or potassium-sparing drugs

Potassium supplements and potassium-enriched salt substitutes are a route (c) exception. People who cannot excrete potassium normally risk hyperkalaemia, and they were left out of the potassium review (Aburto et al., 2013). The large salt-substitute trial found no excess hyperkalaemia, but it screened out known kidney disease and potassium-sparing or potassium-supplement use, so it confirms the caution rather than lifting it (Neal et al., 2021). Details: Sodium.

Genotype

Poole’s umbrella review cites one report, by Palatini and colleagues, in which the hypertension risk linked to coffee varied by CYP1A2 genotype: «Those with alleles for slow caffeine metabolism were at increased risk of hypertension compared with those with alleles for fast caffeine metabolism.» (Poole et al., 2017) This is a route (b) candidate, not yet a basis for advice -> Coffee Consumption and Health.

What guidance does with strata

WHO’s 2012 sodium guideline declined to stratify, citing both the high prevalence of hypertension and a smaller but still significant effect in normotensives (-1.38 mmHg, 95% CI -2.74 to -0.02) — partly a public-health reason, partly an evidential one (World Health Organization, 2012) -> Which Objective Moved This Recommendation.

For those already on treatment, the remaining question is how far down to go.

Targeting systolic below 120 prevented more events in high-risk trial patients, with more side effects

One large trial tested a lower target head-on. The randomised evidence shows no excess cardiovascular events or deaths at the lower pressures studied, though adverse effects rose. A prescriber chooses a target for a particular person, with their history and medicines in hand. What follows is what the trials found.

SPRINT: a target below 120 cut events and deaths in high-risk patients

SPRINT randomised 9,361 people aged 50 or older with a systolic pressure of 130-180 mmHg, high cardiovascular risk, no diabetes and no prior stroke to a systolic target below 120 mmHg or below 140 mmHg. Achieved pressures were about 121 and 136, and the trial stopped early, at a median of 3.26 years (SPRINT Research Group, 2015).

• The lower target prevented more major events: the composite of heart attack, acute coronary syndrome, stroke, heart failure and cardiovascular death fell with HR 0.75 (0.64-0.89) (SPRINT Research Group, 2015).
• It also lowered all-cause mortality: HR 0.73 (0.60-0.90) (SPRINT Research Group, 2015).
• In absolute terms, 61 people had to be treated to the lower target for 3.26 years to prevent one primary-outcome event (SPRINT Research Group, 2015).

The extra benefit came with extra harm. Serious hypotension, fainting (syncope), electrolyte abnormalities and acute kidney injury or failure were all more common on the lower target; injurious falls were not (SPRINT Research Group, 2015). So the target is a net-effect judgement, not a rule that lower is always better.

SPRINT’s result is what the per-mmHg drug-trial rate predicts. A gap of about 15 mmHg, at the rate from the pooled trials described in A 5 mmHg drop cuts major events by about a tenth, at every starting pressure tested, predicts a composite HR near 0.75, which is what SPRINT observed (Blood Pressure Lowering Treatment Trialists Collaboration, 2021; inferred from SPRINT Research Group, 2015). Whether this is an independent check is unclear. SPRINT’s investigators appear among BPLTTC’s collaborators, but the pooled trials were published between 1972 and 2013 and SPRINT in 2015, so whether SPRINT is inside the pool cannot be read from the held text -> Blood Pressure Lowering and Cardiovascular Events.

The result holds for the people SPRINT enrolled. They were high-risk, without diabetes, followed for about three years, and reached an average of about 121 mmHg; more than half of the intensive group stayed above 120. The authors themselves name the groups it does not cover: people with diabetes, prior stroke, or under 50, and it did not enrol nursing-home residents (SPRINT Research Group, 2015). It does not show what happens when pressure is held well below 120.

The low-pressure J-curve does not survive randomisation

Observational data once suggested that pushing pressure too low is harmful. A registry of people with ischaemic heart disease found the lowest event risk at around 130/75 mmHg, which was read as a reason not to lower pressure further (Blood Pressure Lowering Treatment Trialists Collaboration, 2021). BPLTTC notes that one randomised trial in people without cardiovascular disease and a systolic pressure below 140 came to a similar conclusion, and its pooled drug trials found no fading of benefit down to a starting systolic below 120 (see A 5 mmHg drop cuts major events by about a tenth, at every starting pressure tested) (Blood Pressure Lowering Treatment Trialists Collaboration, 2021).

The authors conclude that their findings «do not substantiate concerns about a J-shaped association between blood pressure and cardiovascular outcomes in observational studies» (Blood Pressure Lowering Treatment Trialists Collaboration, 2021).

Note the scope: these strata are starting pressures. People who began below 120 still benefited from a further drop; the strata do not by themselves show the effect of holding pressure at any given level .

Dementia shows the same pattern. Observational cohorts in older people show a U-shaped link between blood pressure and dementia. An individual-patient analysis of five placebo-controlled trials found «no evidence of a U-shaped relation of the effect at any age, nor any increase in risk of dementia with treatment in the oldest age» (Peters et al., 2022). Plotting achieved pressure across both arms, it found «a linear relationship between lower risk of dementia and lower BP, down to at least 100 mmHg systolic and 70 mmHg diastolic» (Peters et al., 2022).

That 100/70 is where the data run out, not an optimum: the relationship is monotone over the studied range, and the trials say nothing about pressures below it. The treatment effect on dementia itself is given in A 5 mmHg drop cuts major events by about a tenth, at every starting pressure tested -> Dementia Prevention and Modifiable Risk Factors.

Run through the U-curve diagnostics, the low arm fails as a treatment-effect concern. A harmful-looking low arm in observational data is suspect when sick people’s pressure falls for reasons of their own: reverse causation, frailty, and the falling pressure of the dementia prodrome. Covariate adjustment and dropping early follow-up are weak checks; a randomised or genetic test is the strong one. Randomised evidence removes the low arm as a treatment-effect concern down to a starting systolic below 120 for cardiovascular events, and shows no U in the treatment effect on dementia. The achieved-pressure J-curve is most plausibly an artifact, but no randomised test of holding pressure well below about 120 is held, and none of the named mechanisms was shown to operate in these data -> The U-Shaped Association Artifact. The same diagnostic removed alcohol’s protective arm on mortality -> Alcohol and Mortality and Vascular Disease.

This refutation does not carry over to salt. The J-curve for low sodium intake, seen in the PURE cohort, is a different exposure. Its low arm has so far faced only the weak checks and remains unadjudicated (inferred from Mente et al., 2016) -> Should Sodium Reduction Be Population-Wide or Targeted, Sodium.

Know your overall risk, pull the big levers, and treat the extras as extras

The decision comes down to a few steps. The numbers behind each are in the sections above; this list adds none.

• Start from your overall cardiovascular risk, not the reading alone. The same drop buys a large absolute gain at high risk and a small one at low risk (Baseline Risk and the Relative-Absolute Split, SCORE2 Baseline Risk and the ESC Treatment Thresholds).
• If you drink three or more drinks a day, cutting back is a well-measured way to lower your pressure. The drop grows from about 1 mmHg at 3 drinks a day to about 5.5 at 6 or more. At two or fewer, cutting back made no significant difference to the reading (Alcohol and Mortality and Vascular Disease).
• Regular exercise, cutting heavy drinking, weight loss (best measured with drugs), and DASH-style eating with less salt are the large lifestyle levers. They may overlap: DASH plus lower sodium lowered pressure by less than the sum of each, so do not add those two drops (DASH Diet and Blood Pressure, Sodium).
• At high risk, drug treatment is the route with measured event reductions; lifestyle still lowers the reading on top of drugs, by less in the DASH trials. Whether to use a lifestyle change, a drug, or both is your decision; discuss the drug side with a clinician (Blood Pressure Lowering and Cardiovascular Events, Layer 1 - Ranking Interventions for a Stratum).
• Treat garlic, magnesium, beetroot nitrate and relaxation practices as unproven extras. Their reported drops are either small or rest on weak trials, and none of these trials counted events .
• At low risk with normal pressure, the remaining gain is small. That is a reason to stop optimizing, not a sign that blood pressure does not matter (Cardiometabolic Interventions and Hard CV Outcomes in Low-Risk People).

One question stays open across all of this. The fewer-events rate comes from drug-lowered pressure, and for lifestyle levers it is assumed, not counted. If the drug-trial rate applies, a lifestyle drop should pay off the same way, but outside salt substitution no trial has shown it (Surrogate Outcomes).

Several questions that shape the decision are still open

The evidence here cannot yet answer several things. None points in a known direction; each could move a choice, and a specific study would settle it.

How big each lever is, and how levers combine

• No trial has put exercise head to head against a blood-pressure drug. Naci states it plainly: «No RCTs compared directly exercise against medications.» (Naci et al., 2018) Every exercise-versus-drug comparison is therefore indirect. A randomized comparison in people with hypertension would settle it. -> Blood Pressure Lowering and Cardiovascular Events

• No source held here says how many mmHg a kilogram of weight loss buys. The figures held come from drug trials and marker bundles, not a dose-response. A systematic review of weight change against blood-pressure change would supply it. -> Does Weight Loss Reduce Cardiovascular Events

• Nobody knows what weight loss, DASH-style eating and lower salt do together. The only measured combination is DASH with lower sodium, which came out smaller than the sum of its parts. A factorial trial adding weight loss would show how much of each drop is shared. -> DASH Diet and Blood Pressure

• Isometric training (handgrip, wall-sits) appears only as one arm inside Naci’s exercise-drug network. No dedicated review of it is held, so its size cannot be ranked with confidence.

• DASH’s effect has not been followed for long. Filippou’s trials averaged about 15 weeks, too short to say the drop lasts (see DASH lowers systolic pressure by about 3 to 5 mmHg, depending on how you pool the trials) (Filippou et al., 2020). The DASH-Sodium comparisons ran 30-day controlled feeding periods. Trials lasting a year or more, on self-chosen food, would show whether the drop holds.

Whether a lower reading means fewer events

• Only salt substitution has counted events after a lifestyle-lowered blood pressure. For every other lever, the event payoff assumes the drug-trial rate applies. Event trials of the large lifestyle levers would test that assumption. -> Surrogate Outcomes
• The Look AHEAD report held gives no between-arm difference in achieved blood pressure, so nobody can check whether its null on events was predictable from the blood-pressure channel. -> Does Weight Loss Reduce Cardiovascular Events
• No cardiovascular-event result for tirzepatide is held. The held trial reports point to SURMOUNT-MMO, its outcome trial, which was ongoing when they were published; SELECT’s benefit belongs to a different drug in a different population. -> Semaglutide for Cardiovascular Risk in Obesity
• Uric-acid-lowering drugs may lower blood pressure for a reason other than lower urate. Qu: «It also remains unclear if a decrease in BP does occur, if it can be attributed to allopurinol itself or a decrease in UA level.» (Qu et al., 2016) A trial of a uricosuric drug would separate the two. -> Uric Acid Lowering and Blood Pressure

Who benefits, and how low to go

• No measure of salt sensitivity is held, so the evidence here cannot pick out the people for whom cutting salt matters most. -> Should Sodium Reduction Be Population-Wide or Targeted

• No coffee source held here estimates caffeine’s own effect on blood pressure. It offers only a signal that coffee’s link to hypertension differs by CYP1A2 genotype. -> Coffee Consumption and Health

• The evidence for intensive targets comes from high-risk people without diabetes. How low to go with diabetes, or when older and frail, is not settled by what is held.

• The reasoning behind shifting hypertension thresholds is not held. Why guideline bodies drew the line where they did, and which objectives moved it, cannot be read from the sources here. -> Which Objective Moved This Recommendation

Evidence box

Question	’For an adult with elevated blood pressure or drifting toward it: what is the effect of each modifiable exposure (diet, weight, activity, alcohol, specific nutrients, stress, and drug therapy as the realistic alternative) on blood pressure and on downstream patient-important cardiovascular outcomes, how large and how certain is each, does a given blood-pressure change transmit to those outcomes, and how do the levers rank for a stratum?‘
Evidence included	52 sources — 31 gold, 18 high, 1 moderate, 1 weak, 1 method
Overall certainty	Medium (see Rating Certainty of Evidence)
Source-selection note	2 source(s) below the gold/high evidence bar feed this page: Sutton (RCT, weak); Lowe (RCT, moderate). Their roles and limitations are described in the text.
Last updated	2026-10-07 · Independently reviewed: No · Full edit history

References

Aburto, N. J., Hanson, S., Gutierrez, H., Hooper, L., Elliott, P., & Cappuccio, F. P. (2013). Effect of increased potassium intake on cardiovascular risk factors and disease: systematic review and meta-analyses. BMJ, 346(apr03 3), f1378–f1378. https://doi.org/10.1136/bmj.f1378

Ayoub-Charette, S., Chiavaroli, L., Liu, Q., Khan, T. A., Zurbau, A., Au-Yeung, F., Cheung, A., Ahmed, A., Lee, D., Choo, V. L., Blanco Mejia, S., de Souza, R. J., Wolever, T. M., Leiter, L. A., Kendall, C. W., Jenkins, D. J., & Sievenpiper, J. L. (2021). Different Food Sources of Fructose-Containing Sugars and Fasting Blood Uric Acid Levels: A Systematic Review and Meta-Analysis of Controlled Feeding Trials. The Journal of Nutrition, 151(8), 2409–2421. https://doi.org/10.1093/jn/nxab144

Blood Pressure Lowering Treatment Trialists Collaboration. (2021). Pharmacological blood pressure lowering for primary and secondary prevention of cardiovascular disease across different levels of blood pressure: an individual participant-level data meta-analysis. The Lancet, 397(10285), 1625–1636. https://doi.org/10.1016/s0140-6736(21)00590-0

Chaddha, A., Modaff, D., Hooper-Lane, C., & Feldstein, D. A. (2019). Device and non-device-guided slow breathing to reduce blood pressure: A systematic review and meta-analysis. Complementary Therapies in Medicine, 45, 179–184. https://doi.org/10.1016/j.ctim.2019.03.005

Chen, Q., Liu, H., & Du, S. (2024). Effect of mindfulness-based interventions on people with prehypertension or hypertension: a systematic review and meta-analysis of randomized controlled trials. BMC Cardiovascular Disorders, 24(1). https://doi.org/10.1186/s12872-024-03746-w

Chiavaroli, L., Viguiliouk, E., Nishi, S., Blanco Mejia, S., Rahelić, D., Kahleová, H., Salas-Salvadó, J., Kendall, C., & Sievenpiper, J. (2019). DASH Dietary Pattern and Cardiometabolic Outcomes: An Umbrella Review of Systematic Reviews and Meta-Analyses. Nutrients, 11(2), 338. https://doi.org/10.3390/nu11020338

Dibaba, D. T., Xun, P., Song, Y., Rosanoff, A., Shechter, M., & He, K. (2017). The effect of magnesium supplementation on blood pressure in individuals with insulin resistance, prediabetes, or noncommunicable chronic diseases: a meta-analysis of randomized controlled trials. The American Journal of Clinical Nutrition, 106(3), 921–929. https://doi.org/10.3945/ajcn.117.155291

European Society of Cardiology. (2021). 2021 ESC Guidelines on cardiovascular disease prevention in clinical practice. European Heart Journal, 42(34), 3227–3337. https://doi.org/10.1093/eurheartj/ehab484

Fang, X., Wang, K., Han, D., He, X., Wei, J., Zhao, L., Imam, M. U., Ping, Z., Li, Y., Xu, Y., Min, J., & Wang, F. (2016). Dietary magnesium intake and the risk of cardiovascular disease, type 2 diabetes, and all-cause mortality: a dose–response meta-analysis of prospective cohort studies. BMC Medicine, 14(1). https://doi.org/10.1186/s12916-016-0742-z

Filippou, C. D., Tsioufis, C. P., Thomopoulos, C. G., Mihas, C. C., Dimitriadis, K. S., Sotiropoulou, L. I., Chrysochoou, C. A., Nihoyannopoulos, P. I., & Tousoulis, D. M. (2020). Dietary Approaches to Stop Hypertension (DASH) Diet and Blood Pressure Reduction in Adults with and without Hypertension: A Systematic Review and Meta-Analysis of Randomized Controlled Trials. Advances in Nutrition, 11(5), 1150–1160. https://doi.org/10.1093/advances/nmaa041

GBD 2019 Risk Factors Collaborators. (2020). Global burden of 87 risk factors in 204 countries and territories, 1990–2019: a systematic analysis for the Global Burden of Disease Study 2019. The Lancet, 396(10258), 1223–1249. https://doi.org/10.1016/s0140-6736(20)30752-2

Ge, L., Sadeghirad, B., Ball, G. D. C., da Costa, B. R., Hitchcock, C. L., Svendrovski, A., Kiflen, R., Quadri, K., Kwon, H. Y., Karamouzian, M., Adams-Webber, T., Ahmed, W., Damanhoury, S., Zeraatkar, D., Nikolakopoulou, A., Tsuyuki, R. T., Tian, J., Yang, K., Guyatt, G. H., & Johnston, B. C. (2020). Comparison of dietary macronutrient patterns of 14 popular named dietary programmes for weight and cardiovascular risk factor reduction in adults: systematic review and network meta-analysis of randomised trials. BMJ, m696. https://doi.org/10.1136/bmj.m696

Geiger, C., Cramer, H., Anheyer, D., Dobos, G., & Kohl-Heckl, W. K. (2025). A systematic review and meta-analysis of yoga for arterial hypertension. PLOS One, 20(5), e0323268. https://doi.org/10.1371/journal.pone.0323268

He, F. J., Li, J., & MacGregor, G. A. (2013). Effect of longer term modest salt reduction on blood pressure: Cochrane systematic review and meta-analysis of randomised trials. BMJ, 346(apr03 3), f1325–f1325. https://doi.org/10.1136/bmj.f1325

Landry, M. J., Senkus, K. E., Mangels, A. R., Guest, N. S., Pawlak, R., Raj, S., Handu, D., & Rozga, M. (2024). Vegetarian dietary patterns and cardiovascular risk factors and disease prevention: An umbrella review of systematic reviews. American Journal of Preventive Cardiology, 20, 100868. https://doi.org/10.1016/j.ajpc.2024.100868

Lean, M. E. J., Leslie, W. S., Barnes, A. C., Brosnahan, N., Thom, G., McCombie, L., Peters, C., Zhyzhneuskaya, S., Al-Mrabeh, A., Hollingsworth, K. G., Rodrigues, A. M., Rehackova, L., Adamson, A. J., Sniehotta, F. F., Mathers, J. C., Ross, H. M., McIlvenna, Y., Welsh, P., Kean, S., … Taylor, R. (2019). Durability of a primary care-led weight-management intervention for remission of type 2 diabetes: 2-year results of the DiRECT open-label, cluster-randomised trial. The Lancet Diabetes &amp; Endocrinology, 7(5), 344–355. https://doi.org/10.1016/s2213-8587(19)30068-3

Lincoff, A. M., Brown-Frandsen, K., Colhoun, H. M., Deanfield, J., Emerson, S. S., Esbjerg, S., Hardt-Lindberg, S., Hovingh, G. K., Kahn, S. E., Kushner, R. F., Lingvay, I., Oral, T. K., Michelsen, M. M., Plutzky, J., Tornøe, C. W., & Ryan, D. H. (2023). Semaglutide and Cardiovascular Outcomes in Obesity without Diabetes. New England Journal of Medicine, 389(24), 2221–2232. https://doi.org/10.1056/nejmoa2307563

Livingston, G., Huntley, J., Liu, K. Y., Costafreda, S. G., Selbæk, G., Alladi, S., Ames, D., Banerjee, S., Burns, A., Brayne, C., Fox, N. C., Ferri, C. P., Gitlin, L. N., Howard, R., Kales, H. C., Kivimäki, M., Larson, E. B., Nakasujja, N., Rockwood, K., … Mukadam, N. (2024). Dementia prevention, intervention, and care: 2024 report of the Lancet standing Commission. The Lancet, 404(10452), 572–628. https://doi.org/10.1016/s0140-6736(24)01296-0

Look AHEAD Research Group. (2013). Cardiovascular Effects of Intensive Lifestyle Intervention in Type 2 Diabetes (N Engl J Med 2013;369(2):145-154). New England Journal of Medicine, 369(2), 145–154. https://doi.org/10.1056/nejmoa1212914

Lowe, D. A., Wu, N., Rohdin-Bibby, L., Moore, A. H., Kelly, N., Liu, Y. E., Philip, E., Vittinghoff, E., Heymsfield, S. B., Olgin, J. E., Shepherd, J. A., & Weiss, E. J. (2020). Effects of Time-Restricted Eating on Weight Loss and Other Metabolic Parameters in Women and Men With Overweight and Obesity: The TREAT Randomized Clinical Trial. JAMA Internal Medicine, 180(11), 1491. https://doi.org/10.1001/jamainternmed.2020.4153

Ma, C., Avenell, A., Bolland, M., Hudson, J., Stewart, F., Robertson, C., Sharma, P., Fraser, C., & MacLennan, G. (2017). Effects of weight loss interventions for adults who are obese on mortality, cardiovascular disease, and cancer: systematic review and meta-analysis. BMJ, j4849. https://doi.org/10.1136/bmj.j4849

Ma, X., Zhang, H., & Jia, J. (2025). The effect of garlic on the lowering of blood pressure in the patients with hypertension: an updated meta-analysis and trial sequential analysis. Asian Biomedicine, 19(3), 131–140. https://doi.org/10.2478/abm-2025-0016

Mach, F., Koskinas, K. C., Roeters van Lennep, J. E., Tokgözoğlu, L., Badimon, L., Baigent, C., Benn, M., Binder, C. J., Catapano, A. L., De Backer, G. G., Delgado, V., Fabin, N., Ference, B. A., Graham, I. M., Landmesser, U., Laufs, U., Mihaylova, B., Nordestgaard, B. G., Richter, D. J., … Shek, A. (2025). 2025 Focused Update of the 2019 ESC/EAS Guidelines for the management of dyslipidaemias. European Heart Journal, 46(42), 4359–4378. https://doi.org/10.1093/eurheartj/ehaf190

Malhotra, A., Grunstein, R. R., Fietze, I., Weaver, T. E., Redline, S., Azarbarzin, A., Sands, S. A., Schwab, R. J., Dunn, J. P., Chakladar, S., Bunck, M. C., & Bednarik, J. (2024). Tirzepatide for the Treatment of Obstructive Sleep Apnea and Obesity. New England Journal of Medicine, 391(13), 1193–1205. https://doi.org/10.1056/nejmoa2404881

Mente, A., O’Donnell, M., Rangarajan, S., Dagenais, G., Lear, S., McQueen, M., Diaz, R., Avezum, A., Lopez-Jaramillo, P., Lanas, F., Li, W., Lu, Y., Yi, S., Rensheng, L., Iqbal, R., Mony, P., Yusuf, R., Yusoff, K., Szuba, A., … Yusuf, S. (2016). Associations of urinary sodium excretion with cardiovascular events in individuals with and without hypertension: a pooled analysis of data from four studies. The Lancet, 388(10043), 465–475. https://doi.org/10.1016/s0140-6736(16)30467-6

Millwood, I. Y., Walters, R. G., Mei, X. W., Guo, Y., Yang, L., Bian, Z., Bennett, D. A., Chen, Y., Dong, C., Hu, R., Zhou, G., Yu, B., Jia, W., Parish, S., Clarke, R., Davey Smith, G., Collins, R., Holmes, M. V., Li, L., … Chen, Z. (2019). Conventional and genetic evidence on alcohol and vascular disease aetiology: a prospective study of 500 000 men and women in China. The Lancet, 393(10183), 1831–1842. https://doi.org/10.1016/s0140-6736(18)31772-0

Naci, H., Salcher-Konrad, M., Dias, S., Blum, M. R., Sahoo, S. A., Nunan, D., & Ioannidis, J. P. A. (2018). How does exercise treatment compare with antihypertensive medications? A network meta-analysis of 391 randomised controlled trials assessing exercise and medication effects on systolic blood pressure. British Journal of Sports Medicine, 53(14), 859–869. https://doi.org/10.1136/bjsports-2018-099921

Neal, B., Wu, Y., Feng, X., Zhang, R., Zhang, Y., Shi, J., Zhang, J., Tian, M., Huang, L., Li, Z., Yu, Y., Zhao, Y., Zhou, B., Sun, J., Liu, Y., Yin, X., Hao, Z., Yu, J., Li, K.-C., … Elliott, P. (2021). Effect of Salt Substitution on Cardiovascular Events and Death. New England Journal of Medicine, 385(12), 1067–1077. https://doi.org/10.1056/nejmoa2105675

Norouzzadeh, M., Hasan Rashedi, M., Ghaemi, S., Saber, N., Mirdar Harijani, A., Habibi, H., Mostafavi, S., Sarv, F., Farhadnejad, H., Teymoori, F., Khaleghian, M., & Mirmiran, P. (2025). Plasma nitrate, dietary nitrate, blood pressure, and vascular health biomarkers: a GRADE-Assessed systematic review and dose-response meta-analysis of randomized controlled trials. Nutrition Journal, 24(1). https://doi.org/10.1186/s12937-025-01114-8

Peters, R., Xu, Y., Fitzgerald, O., Aung, H. L., Beckett, N., Bulpitt, C., Chalmers, J., Forette, F., Gong, J., Harris, K., Humburg, P., Matthews, F. E., Staessen, J. A., Thijs, L., Tzourio, C., Warwick, J., Woodward, M., & Anderson, C. S. (2022). Blood pressure lowering and prevention of dementia: an individual patient data meta-analysis. European Heart Journal, 43(48), 4980–4990. https://doi.org/10.1093/eurheartj/ehac584

Poole, R., Kennedy, O. J., Roderick, P., Fallowfield, J. A., Hayes, P. C., & Parkes, J. (2017). Coffee consumption and health: umbrella review of meta-analyses of multiple health outcomes. BMJ, j5024. https://doi.org/10.1136/bmj.j5024

Qu, L., Jiang, H., & Chen, J. (2016). Effect of uric acid-lowering therapy on blood pressure: systematic review and meta-analysis. Annals of Medicine, 49(2), 142–156. https://doi.org/10.1080/07853890.2016.1243803

Rees, K., Takeda, A., Martin, N., Ellis, L., Wijesekara, D., Vepa, A., Das, A., Hartley, L., & Stranges, S. (2019). Mediterranean-style diet for the primary and secondary prevention of cardiovascular disease. Cochrane Database of Systematic Reviews, 2019(3). https://doi.org/10.1002/14651858.cd009825.pub3

Reynolds, A., Mann, J., Cummings, J., Winter, N., Mete, E., & Te Morenga, L. (2019). Carbohydrate quality and human health: a series of systematic reviews and meta-analyses. The Lancet, 393(10170), 434–445. https://doi.org/10.1016/s0140-6736(18)31809-9

Reynolds, A. N., Akerman, A. P., & Mann, J. (2020). Dietary fibre and whole grains in diabetes management: Systematic review and meta-analyses. PLOS Medicine, 17(3), e1003053. https://doi.org/10.1371/journal.pmed.1003053

Ried, K., Frank, O. R., Stocks, N. P., Fakler, P., & Sullivan, T. (2008). Effect of garlic on blood pressure: A systematic review and meta-analysis. BMC Cardiovascular Disorders, 8(1). https://doi.org/10.1186/1471-2261-8-13

Roerecke, M., Kaczorowski, J., Tobe, S. W., Gmel, G., Hasan, O. S. M., & Rehm, J. (2017). The effect of a reduction in alcohol consumption on blood pressure: a systematic review and meta-analysis. The Lancet Public Health, 2(2), e108–e120. https://doi.org/10.1016/s2468-2667(17)30003-8

Sacks, F. M., Svetkey, L. P., Vollmer, W. M., Appel, L. J., Bray, G. A., Harsha, D., Obarzanek, E., Conlin, P. R., Miller, E. R., Simons-Morton, D. G., Karanja, N., Lin, P.-H., Aickin, M., Most-Windhauser, M. M., Moore, T. J., Proschan, M. A., & Cutler, J. A. (2001). Effects on Blood Pressure of Reduced Dietary Sodium and the Dietary Approaches to Stop Hypertension (DASH) Diet. New England Journal of Medicine, 344(1), 3–10. https://doi.org/10.1056/nejm200101043440101

Schünemann, H., Brożek, J., Guyatt, G., & Oxman, A. (n.d.). GRADE Handbook: for grading quality of evidence and strength of recommendations. https://gradepro.org/handbook/

Scientific Advisory Committee on Nutrition. (2015). Carbohydrates and Health. https://www.gov.uk/government/publications/sacn-carbohydrates-and-health-report

SCORE2-Diabetes Working Group and ESC Cardiovascular Risk Collaboration. (2023). SCORE2-Diabetes: 10-year cardiovascular risk estimation in type 2 diabetes in Europe. European Heart Journal, 44(28), 2544–2556. https://doi.org/10.1093/eurheartj/ehad260

Siervo, M., Lara, J., Chowdhury, S., Ashor, A., Oggioni, C., & Mathers, J. C. (2014). Effects of the Dietary Approach to Stop Hypertension (DASH) diet on cardiovascular risk factors: a systematic review and meta-analysis. British Journal of Nutrition, 113(1), 1–15. https://doi.org/10.1017/s0007114514003341

Siervo, M., Lara, J., Ogbonmwan, I., & Mathers, J. C. (2013). Inorganic Nitrate and Beetroot Juice Supplementation Reduces Blood Pressure in Adults: A Systematic Review and Meta-Analysis. The Journal of Nutrition, 143(6), 818–826. https://doi.org/10.3945/jn.112.170233

SPRINT Research Group. (2015). A Randomized Trial of Intensive versus Standard Blood-Pressure Control (SPRINT Research Group; N Engl J Med 2015;373(22):2103-2116). New England Journal of Medicine, 373(22), 2103–2116. https://doi.org/10.1056/nejmoa1511939

Sutton, E. F., Beyl, R., Early, K. S., Cefalu, W. T., Ravussin, E., & Peterson, C. M. (2018). Early Time-Restricted Feeding Improves Insulin Sensitivity, Blood Pressure, and Oxidative Stress Even without Weight Loss in Men with Prediabetes. Cell Metabolism, 27(6), 1212-1221.e3. https://doi.org/10.1016/j.cmet.2018.04.010

van Charante, E. P. M., Richard, E., Eurelings, L. S., van Dalen, J.-W., Ligthart, S. A., van Bussel, E. F., Hoevenaar-Blom, M. P., Vermeulen, M., & van Gool, W. A. (2016). Effectiveness of a 6-year multidomain vascular care intervention to prevent dementia (preDIVA): a cluster-randomised controlled trial. The Lancet, 388(10046), 797–805. https://doi.org/10.1016/s0140-6736(16)30950-3

Wilding, J. P. H., Batterham, R. L., Calanna, S., Davies, M., Van Gaal, L. F., Lingvay, I., McGowan, B. M., Rosenstock, J., Tran, M. T. D., Wadden, T. A., Wharton, S., Yokote, K., Zeuthen, N., & Kushner, R. F. (2021). Once-Weekly Semaglutide in Adults with Overweight or Obesity. New England Journal of Medicine, 384(11), 989–1002. https://doi.org/10.1056/nejmoa2032183

Wilding, J. P. H., Batterham, R. L., Davies, M., Van Gaal, L. F., Kandler, K., Konakli, K., Lingvay, I., McGowan, B. M., Oral, T. K., Rosenstock, J., Wadden, T. A., Wharton, S., Yokote, K., & Kushner, R. F. (2022). Weight regain and cardiometabolic effects after withdrawal of semaglutide: The <scp>STEP</scp> 1 trial extension. Diabetes, Obesity and Metabolism, 24(8), 1553–1564. https://doi.org/10.1111/dom.14725

World Health Organization. (2012). Guideline: Sodium intake for adults and children. https://www.who.int/publications/i/item/9789241504836

World Health Organization. (2023). Use of non-sugar sweeteners: WHO guideline. https://www.who.int/publications/i/item/9789240073616

Yu, J., Zhou, Z., McEvoy, R. D., Anderson, C. S., Rodgers, A., Perkovic, V., & Neal, B. (2017). Association of Positive Airway Pressure With Cardiovascular Events and Death in Adults With Sleep Apnea: A Systematic Review and Meta-analysis. JAMA, 318(2), 156. https://doi.org/10.1001/jama.2017.7967

Zhang, X., Li, Y., Del Gobbo, L. C., Rosanoff, A., Wang, J., Zhang, W., & Song, Y. (2016). Effects of Magnesium Supplementation on Blood Pressure: A Meta-Analysis of Randomized Double-Blind Placebo-Controlled Trials. Hypertension, 68(2), 324–333. https://doi.org/10.1161/hypertensionaha.116.07664