Siervo 2013 (J Nutr) is a systematic review and meta-analysis of 16 randomized clinical trials (254 participants, 7-30 per study) of inorganic nitrate (sodium/potassium nitrate salt) or beetroot juice supplementation vs placebo, on blood pressure. All 16 were crossover, double-blind, placebo-controlled; conducted 2006-2012; intervention duration 2 h to 15 d (washout 6-28 d). This is a distinct exposure from the electrolyte-BP pages (Sodium Intake and Blood Pressure, Potassium Intake and Blood Pressure) and from the whole-pattern DASH Diet and Blood Pressure — the lever here is the inorganic nitrate anion, acting through a nitric-oxide (NO) pathway, not sodium, potassium, or a dietary pattern. (Siervo et al., 2013)
Second source (2026-08-31): Norouzzadeh 2025 (Nutrition Journal) supersedes Siervo on scale and adds the dose-response shape + vascular-marker channel. A GRADE-assessed one-stage dose-response meta-analysis of 75 RCTs (1823 participants), searched to Feb 2024, of dietary NO3 (beetroot n=70, spinach, lettuce, NO3-rich diet) on plasma NO3/NO2, BP, and three vascular-stiffness/endothelial markers (PWV, FMD, AI). (Norouzzadeh et al., 2025) Because Norouzzadeh’s 75 RCTs (through 2024) subsume Siervo’s 16 (2006-2012), the two are not independent backings — this is Siervo’s claim refined and upgraded by a larger, GRADE-rated, dose-response-shaped re-analysis (type-F, not type-E); the teams differ (Siervo/Mathers UK vs Norouzzadeh/Mirmiran Iran) but the overlapping trial base defeats independence. (inferred from Norouzzadeh et al., 2025; Siervo et al., 2013)
A number-decoding note. The source PDF’s OCR renders minus signs as a leading
2(e.g. it prints the pooled SBP effect as24.4 mm Hg (95% CI: 25.9, 22.8)). Every effect below is the decoded value (-4.4 mmHg (95% CI: -5.9, -2.8)), read against the sign-consistent abstract, results table and forest-plot text. Quoted prose spans are OCR-clean and reproduce verbatim; the garbled numeric strings are deliberately NOT quoted verbatim.
The pooled effect — SBP falls ~4.4 mmHg; DBP does not move
Random-effects pooled mean differences, supplementation vs placebo (Siervo et al., 2013):
| Outcome | Effect (95% CI) | P | Studies | State |
|---|---|---|---|---|
| Systolic BP | -4.4 mmHg (-5.9, -2.8) | <0.001 | 16 | benefit (I2=66%, high heterogeneity) |
| Diastolic BP | -1.1 mmHg (-2.2, +0.1) | 0.06 | 13 | no meaningful effect (CI crosses 0) |
| SBP — beetroot juice | -4.5 mmHg (-6.4, -2.5) | <0.001 | 12 | benefit |
| SBP — inorganic nitrate salt | -4.2 mmHg (-6.1, -2.2) | <0.001 | 4 | benefit |
«Inorganic nitrate and beetroot juice supplementation was associated with a significant reduction in systolic BP. These findings need to be tested in long-term trials and in individuals at greater cardiovascular risk.» (Siervo et al., 2013)
The headline is the SBP effect. DBP did not reach significance (-1.1 mmHg, CI crosses zero) — so this is a systolic-selective lever in the pooled data. Beetroot juice and nitrate salt give essentially the same SBP effect (-4.5 vs -4.2 mmHg), which the authors attribute to comparable delivered nitrate doses (beetroot 12.0 +/- 13.2 vs salt 15.5 +/- 9.2 mmol/dose) rather than to beetroot’s other bioactives — the nitrate anion, not the beetroot matrix, is doing the work. (Siervo et al., 2013) Norouzzadeh 2025 corroborates this on the larger base: the BP/serum-NO3 slope tracked NO3 dose more tightly than beetroot dose, so «the observed effects on BP reduction were primarily attributable to the NO3 content rather than other bioactive compounds present in the entire food matrix». (Norouzzadeh et al., 2025) (This is corroboration on an overlapping trial base — type-F reinforcement, not independent type-E.)
Dose-response — SBP tracks nitrate dose, not duration or plasma nitrite
«The meta-regression showed that mean differences in systolic BP were directly associated with the daily dose of inorganic nitrate (P < 0.05) … but not with the study duration (P = 0.67) or plasma nitrite concentrations (P = 0.40).» (Siervo et al., 2013)
A larger daily nitrate dose predicts a larger SBP fall (meta-regression slope ~ -0.12 mmHg per mmol; P<0.05). No knee or plateau is located — this is a monotone slope over the studied dose range only (nitrate salt 2.5-24 mmol/dose [157-1488 mg]; beetroot 5.1-45 mmol/dose [321-2790 mg]). Do not extrapolate above ~45 mmol/dose, where there is no data, and read the slope as dose-associated, not as a titratable target — it is a cross-trial meta-regression, the design most prone to ecological confounding. Duration did not modify the effect (SBP <3 d -4.9 mmHg vs >=3 d -3.8 mmHg, both significant), consistent with an acute pharmacodynamic action that does not accumulate over the (short) studied windows. (Siervo et al., 2013)
Dose-response, sharpened — Norouzzadeh’s per-mmol slopes (the matched-quantity comparison)
Parameter table — Siervo vs Norouzzadeh, matched quantities only (the BLOCKING guard: the naive contrast is Siervo’s fixed-dose pooled SBP effect vs Norouzzadeh’s per-mmol slope — different quantities, not comparable). Norouzzadeh frames its entire BP analysis as a per-mmol dose-response (restricted cubic splines, one-stage) and reports no overall fixed-dose pooled SBP effect, so the only same-quantity cell is slope-vs-slope:
| Parameter | Siervo 2013 | Norouzzadeh 2025 | Same quantity? |
|---|---|---|---|
| SBP change per mmol NO3 | ~-0.12 mmHg/mmol (cross-trial meta-regression, P<0.05) | acute -0.28 (CI -0.40, -0.17); short -0.24 (-0.40, -0.07); medium -0.48 (-0.71, -0.25) mmHg/mmol (one-stage DR) | YES — mmHg per mmol; method differs (ecological meta-regression vs one-stage spline) |
| SBP fixed-dose pooled effect | -4.4 mmHg (-5.9, -2.8) at ~12-15 mmol median dose | not reported (dose-response framing only) | NO — no matched cell; do not contrast -4.4 vs -0.48 |
| DBP | -1.1 mmHg NS (CI crosses 0) | -0.12 mmHg/mmol significant, acute (CI -0.21, -0.03) | NO — pooled-NS vs per-mmol-significant; different constructs |
| Trials / participants | 16 / 254 | 75 / 1823 | superset (Siervo ⊂ Norouzzadeh) |
«For each mmol increase in NO3 intake (Supplementary Fig. 6), there was a decrease in SBP levels in the acute (WMD: -0.28 mmHg; 95% CI: -0.40, -0.17; I2: 31.0), short-term (WMD: -0.24 mmHg;» … «(WMD: -0.48 mmHg; 95% CI: -0.71, -0.25; I2: 53.6) periods.» (Norouzzadeh et al., 2025)
The two sources reconcile as a type-F refinement, not a conflict. Norouzzadeh’s per-mmol slope integrated over a typical delivered dose lands on Siervo’s fixed-dose pool: acute -0.28 mmHg/mmol x ~15 mmol (Siervo’s median dose) ~= -4.2 mmHg, essentially Siervo’s -4.4. So Norouzzadeh supplies the shape whose integral reproduces the incumbent point estimate — the composite (dose-response curve
- its fixed-dose confirmation) beats either alone. (inferred from Norouzzadeh et al., 2025; Siervo et al., 2013)
DBP now moves — a genuine amplify/refine of Siervo’s null. Siervo’s pooled DBP was NS (-1.1 mmHg, CI crosses 0); Norouzzadeh’s larger base resolves a significant acute-term per-mmol DBP slope (-0.12 mmHg/mmol, CI -0.21, -0.03). Small, and acute-only, but no longer null — the added power converted a Siervo “no meaningful effect” into a detectable gradient. (Norouzzadeh et al., 2025)
Shape is outcome-specific — SBP is linear, MAP has a knee. SBP is linear across all three horizons (more mmol, more effect, no plateau located up to the 8-mmol clinical threshold below); but medium-term MAP is non-linear, greatest at 3 mmol/day and not significant above 5 mmol/day:
«There was a non-linear dose-response relationship between dietary NO3 with medium-term MAP, that the greatest effect observed at a dose of 3 mmol dietary NO3 per day (WMD: -4.43 mmHg; 95% CI: -7.84, -1.03). However, there was no significant decrease in medium-term MAP at doses higher than 5 mmol dietary NO3 per day.» (Norouzzadeh et al., 2025)
Read the MAP knee cautiously: it is one outcome, medium-term only, on a subset, with a wide CI — weak evidence of a true plateau, not a located optimum (the same discipline the fabric applies to any knee claim). The decision-bearing shape is the linear SBP slope, which yields Norouzzadeh’s own clinical-dose reading:
«Given that a reduction of at least 2 mmHg in BP is typically considered a clinically significant unin tended BP-lowering effect [17], our findings suggest that a daily NO3 dose of 8 mmol is required to achieve a» (Norouzzadeh et al., 2025)
i.e. a daily 8 mmol NO3 (~500 mg; roughly one large beetroot-juice shot or a big serving of leafy greens) to reach a >=2 mmHg SBP fall — the dose that operationalizes the effect as decision-relevant. Plasma NO3 itself rises linearly only above ~3 mmol/day (no significant change below 2 mmol), setting a floor on the exposure. (Norouzzadeh et al., 2025)
Hypertensive stratum — a steeper slope (route-(b) effect-modification signal)
The BP effect is more pronounced in hypertensives. In the HTN dose-response subgroup, the acute SBP slope is ~35% steeper than in the overall population:
«For each mmol increment in dietary NO3 dosage, we observed a 0.38 mmHg reduction in acute SBP within the hyperten sive subgroup, contrasting with a 0.28 mmHg reduction in the overall population analysis. This finding implies that hypertensive populations may experience more significant advantages from dietary NO3 supplementa tion, possibly owing to their impaired NO bioavailability.» (Norouzzadeh et al., 2025)
This is a positive effect-modification signal (route (b)) with a stated mechanism (HTN involves endothelial dysfunction / diminished NO synthesis, which exogenous NO3 directly opposes) — the stratum where the lever would matter most shows the largest per-mmol effect, and the subgroup slope has I2=0.00 (homogeneous). Treat it as a candidate effect-modifier, not a settled one: it is a subgroup dose-response (the false-positive-prone route), acute-term, and still on the BP surrogate. It is nonetheless the more decision-relevant reading than Siervo’s, whose ambulatory-BP nulls fell in exactly this higher-risk stratum — the two sources leave the hypertensive question open in opposite directions (Siervo: no ambulatory confirmation in older/obese/T2D; Norouzzadeh: steeper office-BP slope in HTN). (inferred from Norouzzadeh et al., 2025)
Vascular-stiffness and endothelial markers — a new surrogate channel (still a surrogate)
Beyond BP, Norouzzadeh adds a per-mmol dose-response on three vascular markers — arterial stiffness (PWV, AI) and endothelial function (FMD):
| Marker (medium-term, per mmol NO3) | Effect (95% CI) | Direction |
|---|---|---|
| Pulse wave velocity (PWV) | -0.07 m/s (-0.11, -0.03) | stiffness down (benefit) |
| Flow-mediated dilation (FMD) | +0.30% (0.15, 0.46) | endothelial function up (benefit) |
| Augmentation index (AI) | -0.57% (-0.98, -0.15) | wave reflection down (benefit) |
All three move favourably and coherently with the BP finding, which raises mechanistic confidence that the NO pathway is doing real vascular work — but these are surrogates too, not events. Norouzzadeh bridges FMD to hard outcomes by the same modelled-projection move as Siervo’s JNC-7 step, and it must be read the same way:
«A meta-analysis showed that each 1% increase in FMD caused a 13% decrease in the risk of cardiovas cular events [8]. Therefore, a 1.85% enhancement in FMD due to dietary NO3 supplementation [22] could poten tially lead to a 24% decrease in cardiovascular events.» (Norouzzadeh et al., 2025)
The 24% figure is a projection off two borrowed references (the 13%/1% FMD-CVD elasticity from [8], the 1.85% FMD gain from [22] — not this MA’s own per-mmol slope), chained onto a surrogate. It is a plausible direction, not measured event evidence — exactly the transmission-untested status of Siervo’s stroke/CVD projection. -> Surrogate Outcomes. (inferred from Norouzzadeh et al., 2025)
GRADE certainty — Moderate at best, and only for the acute BP outcomes
Norouzzadeh is GRADE-rated (the upgrade over Siervo’s un-graded estimate). The certainty ceiling is Moderate, reached by only a few outcomes; most sit lower:
«(acute and short-term) and DBP (acute-term) were rated as Moderate due to their significant and homogeneous results. Also, acute plasma NO2 received a Low CoE due to suspected publication bias, while medium-term SBP was rated Low due to heterogeneity in results.» (Norouzzadeh et al., 2025)
So acute + short-term SBP and acute DBP are Moderate; medium-term SBP is Low (heterogeneity);
acute plasma NO3 and medium-term PWV are Very Low; and — a load-bearing caveat — risk of bias
was serious for all outcomes except acute and medium-term FMD. No outcome reaches High. The
dose-response gradient is a GRADE upgrade factor, but publication bias, heterogeneity and serious
risk-of-bias hold the ceiling at Moderate. This is why the page moves to confidence: medium and no
higher. (Norouzzadeh et al., 2025)
Safety — no excess adverse events or withdrawals
«The safety analysis results (Supplementary Figs. 12, 13) revealed that dietary NO3 supplementation did not lead to an increase in adverse events (Risk ratio: 1.00, 95% CI 0.35, 2.82 I2 = 0.00%) or withdrawal from the study (Risk ratio: 0.42, 95% CI 0.07, 2.68; I2 = 0.00%).» (Norouzzadeh et al., 2025)
Adverse-event RR is a flat 1.00 and withdrawals if anything lower on NO3 (RR 0.42, NS) — no safety signal in these short-to-medium trials (beeturia, the red urine/stool discoloration, is noted as a benign non-cause of withdrawal). This is a short-horizon safety read only; it says nothing about chronic high-dose intake, and Norouzzadeh separately flags that the nitrate acceptable daily intake set for cured meats and water may not transfer to vegetable NO3 and needs reassessment — a regulatory open question, not a demonstrated harm. (Norouzzadeh et al., 2025)
Layer-1 sizing — the BP-drug comparator shrinks the rock for the outcome it covers
Norouzzadeh itself supplies the substitution comparator that sizes this lever:
«A meta-analysis of 68 clinical trials revealed that a single antihypertensive medication can lower SBP and DBP by 12 and 7 mmHg, respectively [17], while a combination therapy can result in a reduction of up to 18.9 mmHg in SBP [157].» (Norouzzadeh et al., 2025)
A single mature antihypertensive lowers SBP ~12 mmHg vs the ~2-5 mmHg realistically reachable from dietary NO3 at tolerable doses — so for the BP-lowering outcome specifically, a low-harm generic BP drug captures most of the benefit and the nitrate rock is small at the margin. Two guards keep this from over-shrinking the lever: substitution is outcome-specific — if the vascular-marker / endothelial channel (FMD, PWV) carries value the single-channel BP drug does not, the drug does not shrink that part; and net of the drug’s own costs (lifelong dependency, side-effects) a food lever that a person will actually sustain retains structural value. The choice stays the person’s (Layer 3); Layer 1 only notes that on BP alone this is a small marginal rock behind a mature drug. (inferred from Norouzzadeh et al., 2025)
Mechanism — the enterosalivary nitrate-nitrite-NO pathway
«Dietary and endogenous inorganic nitrate molecules may be reduced by facultative anaerobic bacteria on the dorsal surface of the tongue to nitrite, which can be chemically (low pH) and enzymatically (xanthine oxidoreductase, myoglobin, cytochrome P450, complexes of the mitochondrial electron transport chain) further reduced to NO … This pathway may represent a plausible mechanism to explain the beneficial effects of dietary inorganic nitrate intake on BP.» (Siervo et al., 2013)
This is a human-corroborated physiological pathway (nitrate -> nitrite -> NO -> vasodilation), distinct from the classical L-arginine/NO-synthase route, and it depends on oral commensal bacteria as an obligatory reduction step — a support factor the mechanism needs. The dependence is a transportability caveat with a concrete failure mode: antibacterial mouthwash or antibiotics that suppress the tongue microbiome would be expected to break the pathway (not tested in this MA, a named gap). (inferred from Siervo et al., 2013)
The load-bearing caveats — surrogate, short-term, and disconfirmed on ambulatory BP
- Every endpoint is a SURROGATE — office/resting BP, not events. No mortality, stroke, MI, or incident CVD is measured; the trials run 2 h to 15 d. -> Surrogate Outcomes.
- The resting-BP benefit was NOT confirmed on ambulatory 24-h monitoring — the more valid BP
measure. This is a within-source disconfirmation, not a footnote:
«The beneficial effects of beetroot juice supplementation on BP were not confirmed when ambulatory 24-h BP monitoring was used to measure changes in BP (39-41), including a group of older (>60 y) overweight and obese participants with type 2 diabetes.» (Siervo et al., 2013) The three ambulatory studies are also the ones in higher-risk participants (older, obese, T2D) — exactly the stratum where a BP lever would matter most — so the null there is doubly cautionary.
- The tested population is the wrong one for the decision. Most participants were young,
non-smoking, normal-weight, healthy men recruited for exercise-performance studies. The authors:
«we advocate for a careful interpretation of the results in consideration of the small number of participants, the short duration of the interventions, and the over-representativeness of healthy participants.» (Siervo et al., 2013)
- High heterogeneity on the primary outcome (SBP I2=66%), driven by shorter-duration and beetroot-juice studies. (Siervo et al., 2013)
Surrogate-to-outcome — a modelled projection on a borrowed evidence base
The authors translate the surrogate to hard outcomes via a JNC-7 projection, not measured events:
«a systolic BP reduction of at least 5 mm Hg (as observed here) could decrease the risk of mortality due to stroke by 14% and mortality from cardiovascular diseases by 9%.» (Siervo et al., 2013)
Read this as a plausible direction, not measured evidence: it applies a projection derived from pharmacological BP-lowering (the real trial transmission lives on Blood Pressure Lowering and Cardiovascular Events) to a dietary intervention, and the observed SBP fall (-4.4 mmHg) is in fact below the >=5 mmHg the projection assumes. Whether the surrogate transmits at all here is untested — the acute, short-term, ambulatory-disconfirmed BP signal is a weaker basis for the outcome bridge than a sustained clinic-BP reduction would be. The absolute benefit, if the bridge holds, still scales with the person’s baseline CV risk (Baseline Risk and the Relative-Absolute Split). (inferred from Siervo et al., 2013)
Relation to DASH — this is the head-on test of DASH’s proposed nitrate component
DASH Diet and Blood Pressure (Siervo 2015, the same research group) flags the pattern’s high inorganic-nitrate content (~1200 mg/d from leafy/root vegetables, feeding NO generation) as one candidate mechanism for DASH’s -5.2 mmHg SBP effect, but cannot decompose the pattern. This MA supplies the isolated estimate: dietary nitrate alone (157-2790 mg/dose) lowers SBP ~4.4 mmHg — which bounds how much of DASH’s BP effect the nitrate component could plausibly carry. Because the two MAs share lead authors (Siervo, Lara, Mathers), this is the same group’s refinement of its own claim (type-F), not independent corroboration (not type-E) — the convergence is self-stated. It does not license summing DASH and nitrate as independent additive levers (overlapping NO mechanism). (inferred from Siervo et al., 2013, 2014)
Decision relevance
- Dietary nitrate / beetroot is a real but MODEST, LOW-CERTAINTY BP lever — a systolic-selective ~4.4 mmHg fall on resting BP, from short (hours-to-days) trials in mostly healthy young men, not confirmed on 24-h ambulatory monitoring and untested on hard outcomes. It is a peripheral supplement-class lever, not a big rock; a person whose large BP levers (weight, sodium/potassium balance, alcohol, activity) are unpulled should address those first.
- Beetroot juice and nitrate salt are interchangeable for this effect (dose-matched); the effect tracks nitrate dose within the studied range, with no located knee or plateau.
- The evidence gap that governs the recommendation is duration + population: whether the acute effect sustains, and whether it holds in the hypertensive / older / higher-CV-risk people who would actually benefit — the ambulatory-BP nulls in exactly those strata are the reason to hold this at low-to-moderate certainty and not recommend it as a standing antihypertensive strategy yet.
- Update (2026-08-31, Norouzzadeh 2025): the larger GRADE-rated dose-response upgrades confidence to medium (acute/short SBP + acute DBP are Moderate CoE), gives an operable dose (~8 mmol/day for a >=2 mmHg SBP fall), shows the effect is steeper in hypertensives (route-(b) candidate), and confirms short-term safety (adverse-event RR 1.00). It does not close the governing gap: still no hard outcomes, and medium-term SBP is only Low CoE. Net decision unchanged in kind — a real, modest, now-better-quantified peripheral lever — but with a concrete dose and a higher-yield stratum (hypertensives) named.
- Sizing against the drug (Layer 1): on BP alone a single generic antihypertensive (~12 mmHg SBP) dwarfs the ~2-5 mmHg from tolerable NO3 doses, so this is a small marginal rock behind a mature drug for BP; its non-substitutable residual is the food-sustainability and the vascular-marker / endothelial channel a single-target drug may not cover.
(Norouzzadeh et al., 2025; inferred from Siervo et al., 2013)
Limits
- Single gold-tier MA, short-duration surrogate evidence, ambulatory disconfirmation — held at
confidence: lowdespite the gold tier: the design (2 h-15 d, resting BP, healthy young men) is a weak basis for a chronic-BP recommendation. Named gap: a long-term nitrate/beetroot RCT with ambulatory BP in a hypertensive or higher-CV-risk population would move this (none held). - DBP effect null; dose-response is a cross-trial meta-regression (ecological, not within-person titration); plasma nitrite did not track BP, so the biomarker chain is not cleanly closed.
- No hard-outcome evidence — now confirmed across BOTH MAs (a stronger type-G gap). Every endpoint in Siervo and Norouzzadeh is a surrogate (office BP, plasma NO3/NO2, PWV/FMD/AI); neither measures events or mortality. Both bridge to hard outcomes only by modelled projection off borrowed references (Siervo’s JNC-7 stroke/CVD figures; Norouzzadeh’s FMD->CVD 24% chain), not measured events. The gap persists at n=75 RCTs / 1823 participants — a long-term NO3/beetroot RCT with a clinical endpoint in a hypertensive/higher-risk population would move this; none held (transmission via Blood Pressure Lowering and Cardiovascular Events).
- Norouzzadeh caveats (2026-08-31): GRADE ceiling is Moderate (no High outcome); risk of bias serious for all outcomes except FMD; medium-term SBP and plasma-NO3 outcomes are Low/Very-Low CoE; the HTN-steeper-slope is a subgroup dose-response (route-(b), false-positive-prone), acute-term. The MAP knee (~3 mmol, non-significant >5 mmol) is weak/single-outcome and not a located optimum.
(Norouzzadeh et al., 2025; inferred from Siervo et al., 2013)