Whole Grains Refined Grains and Pulses

Two-source page, confidence: low. The body below is SACN 2015; NNR - Nordic Nutrition Recommendations 2023 was added 2026-07-28 and is the second guidance family this page AWAITS-ed. It does not raise confidence, and the section at the bottom explains why: NNR read SACN. The questions this page answers are ones the maintainer holds a stated prior on (SC-11) — a reason for more scrutiny of agreement, not less. Still AWAITS Reynolds - Carbohydrate Quality and Human Health 2019 (staged; note Reynolds shares an author with Te Morenga, so it cannot corroborate independently — and NNR leans on Reynolds for its whole-grain dose-response, so that dependency now runs through this page too).

The headline: refined grains are NOT found harmful

Exposure -> outcome	Estimate	SACN grade
Refined grains -> cardiovascular events	RR 1.00 (95% CI 0.98, 1.01) per half serving/day, p=0.5	No association · Moderate
Refined grains -> type 2 diabetes	RR 1.00 (95% CI 0.98, 1.01), p=0.7	No association · Moderate

[EXTRACTED (SACN - Carbohydrates and Health 2015) 7.20-7.25]

And the randomised comparisons are null across the board. SACN states the comparator explicitly — «All trials compared whole grain diets to refined grain control diets.» — then reports no effect on systolic BP (0.2 mmHg, 95% CI -1.6 to 2.0; p=0.85), diastolic, total cholesterol (0.04 mmol/L, 95% CI -0.12 to 0.20; p=0.49), triacylglycerol, fasting glucose (-0.05 mmol/L, 95% CI -0.12 to 0.02; p=0.14), fasting insulin and insulin sensitivity — several at Adequate evidence. [EXTRACTED (SACN - Carbohydrates and Health 2015) 8.126-8.149]

The one non-null randomised outcome favours whole grains, on an intermediate: «An effect is demonstrated for higher whole grains consumption on reducing energy intake (-360 kJ, 95% CI -642, -79; p=0.01)» — graded Effect · Limited evidence, and SACN notes one trial contributes 75% of the pooled estimate. [EXTRACTED (SACN - Carbohydrates and Health 2015) 8.126-8.149]

The claim that survives is narrower than the popular one

Whole-grain benefit is cohort-only and mostly Limited — CVD RR 0.95, stroke RR 0.96, colo-rectal RR 0.97, the last graded No association. And SACN attaches its own attribution caveat:

«Any associations indicated for whole grain may be related to its cereal fibre component.» [EXTRACTED (SACN - Carbohydrates and Health 2015) chunk 05]

So the supported claim is whole grains are associated with lower risk in cohorts, possibly via fibre — NOT refined grain causes harm. Those are different claims and SACN separates them.

Decision consequence. Replacing refined with whole grain is supported as a plausible small benefit routed through fibre, on cohort evidence. Avoiding refined grain as a harm-avoidance move is not supported by this source — the harm estimate is a flat null at Moderate, SACN’s middle strength grade.

Self-critique correction (2026-07-27): the first draft called Moderate “the strongest evidence-state SACN issues for a no-association verdict.” False — SACN does issue No association · Adequate elsewhere [searched: "no association"+"adequate", both orders, across all 13 SACN chunks]. So the refined-grain null sits one grade below SACN’s ceiling, not at it. The conclusion is unchanged; its stated strength was inflated. -> [[Is the Food Category Doing Any Work]] — “grain” may be the wrong unit; the fibre caveat says so in SACN’s own voice.

Pulses — SACN largely did not look, and that is the finding

Where measured, null: legume fibre -> T2DM RR 1.01 (0.98, 1.04) per 1 g/day; legume fibre -> colo-rectal cancer RR 0.98 (0.94, 1.02); non-soy legume -> CVD RR 0.96 (0.90, 1.03). The one positive is faecal weight, which SACN itself bounds:

«The biological relevance is unclear due to the size of supplements, and it is unclear whether this finding is applicable to all legume fibres.» [EXTRACTED (SACN - Carbohydrates and Health 2015) chunk 08]

Two reasons those nulls are weak evidence of absence, not evidence of no effect.

1. Roughly fourteen legume outcome cells sit in the insufficient evidence tables against that single conclusion. Silence here is un-studied, not un-associated — the expectancy test applies.
2. The nulls are low-power by construction. SACN standardises increments to ~1 SD of UK intake, and legume fibre’s SD is 1 g/day against 7 g/day for total fibre. A null estimated across a distribution one-seventh as wide, in a low-consumption population, is close to uninformative about what happens at high intake. [INFERRED (SACN - Carbohydrates and Health 2015) — the increment widths are SACN's; the power inference is the wiki's]

Pulses appear in exactly one SACN recommendation, and it is about weaning — “from about six months of age gradual diversification of the diet to provide increasing amounts of whole grains, pulses, fruits and vegetables is encouraged.”

This is a G gap, stated as one: the wiki cannot currently answer whether pulses help, harm, or do nothing at the intakes someone considering them would actually eat. -> SC-11

What this page does NOT establish

• Nothing about preparation — now held on a sibling page. Soaking, cooking, fermenting and germinating are untouched by SACN; Petroski - Antinutrients Narrative Review 2020 was ingested 2026-07-29 and addresses them directly -> Antinutrients in Plant Foods (preparation defuses the antinutrient concern for prepared plant staples; raw high-lectin legumes are the exception that requires a full boil).
• Nothing about carbohydrate load or bulk tolerance — partly held on a sibling page. SACN excluded gastrointestinal tolerance from its remit. The related protein question — that hitting a protein target from plants costs more food mass because plant DIAAS is low (peas 64, wheat 40 vs milk 122) — is now held at Protein Quality and the DIAAS Score; the exact carb/bulk arithmetic remains a G gap (needs a food-composition dataset).
• Nothing about ultra-processing. “Refined” here is a grain-milling property, not a processing classification; Hall - Ultra-Processed Diets Inpatient RCT 2019 is the staged source for that and is a different question.

Self-critique [run 2026-07-27, before commit]

• Over-claim: FOUND and fixed. Moderate was described as SACN’s strongest no-association grade; it is not — No association · Adequate occurs in the report. Corrected in place with the search scope recorded. This error inflated a finding that runs AGAINST the maintainer’s stated prior (SC-11, bias against processed grains), so it is not flattery — it is the ordinary one-notch-too-strong failure, and it appeared even on a claim there was no motive to strengthen.
• Absence claims: three appear in What this page does NOT establish (preparation, tolerance, ultra-processing). The tolerance one is sourced — SACN states the exclusion. The other two are scoped to this source, not to the literature. Held.
• Laundered independence: none claimed. The page flags in advance that Reynolds shares an author with Te Morenga and cannot corroborate independently. Clean.
• Residual risk: single-source, confidence: low, and the pulses section leans on an [INFERRED] power argument that is the wiki’s, not SACN’s. Marked as such.

Fibre ISOLATES are better evidenced than fibre-bearing foods — and SACN bounds why [chunk 09]

This cuts against the usual intuition that whole food beats supplement, and SACN’s own grades say so:

«• Effect • Adequate evidence • The direction of the effect demonstrates higher consumption of fibre isolates and gum supplements is beneficial to health • The effect is biologically relevant, but demonstrated at intakes achieved through supplementation» [EXTRACTED (SACN - Carbohydrates and Health 2015) chunk 09]

«• Effect • Moderate evidence • The direction of the effect demonstrates consumption of cellulose is potentially beneficial to health • The effect is potentially biologically relevant, but demonstrated at concentration of intake achieved through supplementation» [EXTRACTED (SACN - Carbohydrates and Health 2015) chunk 09]

Form	SACN’s strongest grade for it	Design
Fibre isolates + gum supplements	Effect · Adequate — SACN’s top strength grade	supplement trials
Cellulose	Effect · Moderate	supplement trials
Psyllium	Effect · Moderate and Effect · Limited — the grade varies by outcome	supplement trials
Whole grains	mostly Limited; colo-rectal graded No association	cohort only

The asymmetry is a DESIGN artifact, not a verdict on food versus supplement. Isolates can be randomised at a chosen dose; whole-grain intake mostly cannot, so it is observed. The better grade tracks the better design, not the better food — and SACN attaches the transportability caveat itself, twice, in the same clause as the finding: the effect is demonstrated at supplement-level intakes, which is not a claim about eating more of a fibre-bearing food. [INFERRED (SACN - Carbohydrates and Health 2015) — the design explanation is the wiki's; SACN states the supplementation bound but does not draw the design contrast] -> [[Is the Food Category Doing Any Work]], -> [[Upgrading Observational Evidence]]

The 30 g target against what people actually eat [chunk 08/10]

«Setting the dietary reference value for AOAC fibre at 30g/day for adults means that current mean intakes would be 10-11g below the dietary reference value for men and 13g below for women.» [EXTRACTED (SACN - Carbohydrates and Health 2015) chunk 08]

A ~40% shortfall against the target, population-wide. Two consequences:

• A “get to 30 g” recommendation is a large behaviour change, not a tweak — and the telos requires judging interventions on adherence, not on the ideal.
• The evidence for benefit is strongest in the range people are already leaving, which is where the dose-response is estimated -> Measurement Error in Dietary Assessment.

Do not compare this figure to the NSP tables. SACN’s annex reports non-starch polysaccharide intakes of ~11-15 g/day by age and sex (Table 3.12) — NSP is a narrower measure than AOAC fibre and the two are not interchangeable. SACN’s own AOAC-terms statement above is the comparison to use; reading the NSP number against the 30 g AOAC target would overstate the shortfall.

Self-critique of the chunk-09/10 additions [run 2026-07-27, before commit]

• Grade error: FOUND and fixed. The isolate table listed psyllium as Effect · Limited only. SACN issues psyllium conclusions at both Moderate and Limited, varying by outcome [searched: "(adequate|moderate|limited) evidence ... psyllium" across chunk 09]. Corrected. A single grade per substance is the wrong shape — SACN grades exposure-outcome pairs, so any table keyed on substance alone will misrepresent it.
• Same-quantity error: AVOIDED, and recorded as a near miss. The NSP intake figure (~11-15 g/day) was about to be compared against the 30 g AOAC reference value. They are different measures. SACN supplies the correct AOAC-terms comparison itself; the near miss is documented on Measurement Error in Dietary Assessment rather than quietly dropped.
• Over-claim check on the isolate asymmetry. The claim that better grades track better design rather than better food is tagged [INFERRED] — SACN states the supplementation bound but does not draw the design contrast. Held, and correctly attributed to the wiki.
• Adjudication check (the new rule): no SC-NN row was scored and no challenge adjudicated in this pass. The fibre-isolate material bears on SC-07 and SC-18; the handles are noted in the log line only, and the scoring is left to a separate operation.
• Residual: still a single-source page at confidence: low. The isolate-vs-food asymmetry is the kind of finding a second source could easily overturn.

The second guidance family arrives — and it is not a second witness [2026-07-28]

NNR - Nordic Nutrition Recommendations 2023 cashes the AWAITS above. The first thing it establishes is that it cannot corroborate this page, and NNR says so itself by listing its inputs:

«Seven qSRs are available on the role of cereals (grains) and health outcomes (Fogelholm et al., 2012; Hauner et al., 2012; Reynolds et al., 2019; SACN, 2015; WCRF/AICR, 2018b, j; Åkesson et al., 2013).» [EXTRACTED (NNR - Nordic Nutrition Recommendations 2023) chunk 06]

SACN 2015 is one of the seven. So NNR agreeing with SACN on grains is a body agreeing with a review it read — not independent backing, and no [E-independent] is claimable. The surface markers again point the wrong way (different country, different decade, different committee, convergent conclusion), which is exactly the trap the laundering guard exists to catch.

Refined grains: the two bodies describe the evidence base differently

Parameter table (op-weave 2a):

Parameter	SACN 2015	NNR 2023	Same quantity?
Exposure	«refined grains», per half serving/day	«refined grains (flour) specifically»	NO — NNR’s parenthetical narrows to flour
Outcome	cardiovascular events; type 2 diabetes	health outcomes, unspecified in the adverse-effects box	NO — SACN names outcomes, NNR does not
Estimate	RR 1.00 (0.98, 1.01), both outcomes	none given	NO — one is a pooled estimate, the other a statement about literature volume
Evidence-state verdict	No association · Moderate	«Few or no studies on refined grains (flour) specifically»	NO — measured-null vs unstudied
Benefit claim	not asserted	«available evidence does not indicate similar beneficial associations compared with whole grains»	comparable in direction only

The fourth column is NO on every substantive row, so the honest artifact is a distinction, not a tension — and the distinction is one the telos names explicitly.

SACN and NNR occupy two of the four evidence states, on what looks like one question. SACN pooled cohort data to a flat null and graded it Moderate — no meaningful effect. NNR reports «Few or no studies on refined grains (flour) specifically» and lists the sparsity in its data-gaps block — insufficient evidence. These are not the same verdict, and the telos is explicit that they must not be collapsed.

Before reading that as a contradiction, note what dissolves most of it. The exposures are not the same: SACN estimated refined grains as a food category from cohorts; NNR’s sparsity claim is scoped by its own parenthetical to flour specifically. A category can be well-estimated in aggregate while its individual constituents are not — which is Is the Food Category Doing Any Work running in the opposite direction to its usual use. What survives is narrower and still useful: two Tier-A bodies looking at overlapping literature reached “measured and null” and “barely studied”, and a reader told only one of the two would form a different picture of how settled this is.

What NNR adds that SACN does not have

• An explicit carve-out for refined grains, on an energy-adequacy warrant. NNR’s science advice based on health outcomes ends: «At high energy requirements refined grains also have a role. This justifies allowing some refined cereals in the diet.» SACN issues no equivalent — it reports a null and stops. This is a decision-change for one identifiable stratum (high energy requirement: athletes, heavy manual work, high-growth adolescence), and it converts a null into a permission. [EXTRACTED (NNR - Nordic Nutrition Recommendations 2023) chunk 06]

• A quantified whole-grain target with a stated direction beyond it: «It is recommended to consume at least 90 g/day (dry weight) of whole grains (including whole grains in products), with likely further benefits of higher intakes. Such further intakes have no adverse effects». Compare the units before comparing the numbers — NNR’s 90 g/day is dry weight of whole grains; SACN’s 30 g/day is AOAC fibre. Different substances, different measures, not comparable, and the page’s existing NSP-vs-AOAC warning applies with more force here. [EXTRACTED (NNR - Nordic Nutrition Recommendations 2023) chunk 06]

• A dose-response statement that may or may not describe a knee — recorded as [AMBIGUOUS].

  «Dose-response curves show that risk reduction for all-cause mortality is observed for intakes up to 50-60 g/day of whole grains. Higher intakes (i.e. 90 g/day) confer even greater risk reduction for coronary heart disease, type 2 diabetes and colorectal and breast cancer (Reynolds et al., 2019).» [EXTRACTED (NNR - Nordic Nutrition Recommendations 2023) chunk 06]

  «observed for intakes up to 50-60 g/day» is genuinely two-way. It reads either as benefit accrues up to 50-60 g/d and then flattens (a plateau on all-cause mortality, with other outcomes still climbing at 90 g/d — which would be the corpus’s first outcome-specific knee) or as the studied range only extends that far (no shape claim at all). NNR does not disambiguate it, and the underlying source — Reynolds 2019 — is staged but not ingested, so the wiki cannot settle it here. Recorded as an open question rather than scored against the dose-response prior, because a knee claimed on an ambiguous sentence is precisely the artifact that prior exists to guard against. [AMBIGUOUS (NNR - Nordic Nutrition Recommendations 2023) chunk 06] AWAITS Reynolds - Carbohydrate Quality and Human Health 2019 — it would resolve the shape directly, and NNR attributes the sentence to it.

Self-critique of the NNR addition [run 2026-07-28, before commit]

• Fake-tension check: FIRED, and the tension was NOT filed. The attractive claim here was “two Tier-A bodies contradict each other on refined grains.” The parameter table returned NO on four of five rows — different exposure scope (flour vs category), different outcomes, estimate vs literature-volume statement. Filed as a distinction with the residual stated. This is the finding the guard was built for, and it would have read very well as a tension.
• Laundered independence: checked and denied, from NNR’s own text. NNR lists SACN 2015 among its seven cereal qSRs. Stated up front rather than buried, and the page header was rewritten so a reader does not take “two sources” as “two witnesses”.
• Over-claim check on the carve-out. «At high energy requirements refined grains also have a role» is quoted, not paraphrased, and the stratum named (athletes, manual work, adolescence) is the wiki’s illustration of “high energy requirements”, not NNR’s list. Flagged here rather than in the body.
• Same-quantity near miss: AVOIDED. NNR’s 90 g/day whole grains was about to sit near SACN’s 30 g/day fibre target. Dry weight of a food against an AOAC fibre measure — recorded in the body as a non-comparison.
• Dose-response: declined to score. The [AMBIGUOUS] sentence is the only candidate knee in the corpus and the temptation was to bank it against the falsified prior. Left open, source named.
• Residual: confidence stays low and the source count going from one to two changes nothing epistemically, which is the whole point of the section.

The pulses gap gets a first number — from a third guidance family [2026-07-28, ESC]

This page files pulses as a G gap, stated as one: the wiki could not say whether pulses help, harm or do nothing at realistic intakes. ESC - CVD Prevention Guidelines 2021 supplies the first quantified claim the corpus holds.

«A single portion of pulses (legumes) a day low- ers LDL-C by 0.2 mmol/L and is associated with a lower risk of CHD.» [EXTRACTED (ESC - CVD Prevention Guidelines 2021) chunk 04]

Read the sentence’s two halves separately — they are different evidence classes, and ESC runs them together:

Claim	Outcome type	Evidence form as stated
LDL-C -0.2 mmol/L per daily portion	surrogate	a magnitude, so presumably trial-derived
«associated with a lower risk of CHD»	hard outcome	associational, no magnitude, no interval

Only the surrogate limb carries a number. So the gap narrows rather than closes: the wiki can now say one portion a day moves LDL-C by a stated amount, and still cannot say what pulses do to events at that intake. -> Surrogate Outcomes

How this sits against SACN, which is the page’s incumbent. SACN found legume-fibre nulls (T2DM RR 1.01; colo-rectal RR 0.98) and this page already explains why they are weak evidence of absence — the increment was 1 g/day against 7 g/day for total fibre, a low-power slice. ESC’s exposure is a whole portion of the food, not a 1 g fibre increment, so the two are not the same quantity and ESC does not contradict SACN. The honest composite: a null across a narrow fibre slice, and a surrogate benefit at a food-level dose. Both can hold.

No [E-independent], and no sources: inflation of the claim’s strength — ESC gives one sentence with two references and no certainty rating, in a guideline whose subject is cardiovascular prevention rather than pulses.

Fibre: a fourth family, and the first one to give a RANGE [2026-07-28, ESC]

ESC’s Table 8 specifies «3045 g of fibre of per day, preferably from wholegrains» (i.e. 30-45 g; the hyphen is lost in the source’s OCR). [EXTRACTED (ESC - CVD Prevention Guidelines 2021) chunk 04, Table 8]

Body	Fibre target	Form	Method specified?
SACN	30 g/day	point	yes — AOAC
WHO	>=25 g/day	floor	not held
NNR	3 g/MJ primary; 25 g/d female, 35 g/d male	energy-scaled + sex-split	no
ESC	30-45 g/day	range	no

Four bodies, four constructs — and the wiki cannot rank them. A point, a floor, an energy-scaled value and a range are not four estimates of one quantity; and only SACN names the analytical method, which this page already records as the thing that makes 30 g and 25 g non-comparable. ESC’s upper bound of 45 g is the highest figure in the corpus and comes with no stated warrant, so it is recorded as a held position, not as evidence that more is better.