← All chapters·Chapter 03 · free opening

What the evidence does not support

Twenty-six widely repeated claims, each graded, each with the source that settles it — and for each one, why it was reasonable to believe.

This is the largest chapter and the reason the publication exists. Twenty-six claims that circulate constantly, each checked against the primary literature and graded.

The tone is deliberate. Every entry states where the grain of truth is and why the claim was easy to believe, because most of these started from a real mechanism and took one step too far. Three of the twenty-six came back with a verdict opposite to the one expected, and they were kept in with the corrected verdict rather than dropped. That difference — between an audit and a debunking list — is the whole method.

Each entry also states which kind of failure it is: tested, no effect; untested; overstated; or category error. Chapter 1 explains why collapsing those four into "no evidence" is the same mistake this chapter is about.

The first six are open. They are not the mildest ones.

1. "The Nordic diet is anti-inflammatory"

B∅ Tested, no effect. Randomised-trial meta-analyses found no effect on CRP (n = 3), IL-6 (n = 1) or TNF-α (n = 1), at moderate certainty — not "insufficient data", but moderate-certainty evidence of absence (Reyneke 2026, PMID 40657695). Independently confirmed: no inverse association with CRP (Tran 2024, PMID 39364652).

Why it matters more than any other entry here. The Nordic pattern is built from almost the same foods as the Mediterranean one — fish, berries, whole grains, rapeseed oil. Same ingredient list, opposite result. That single comparison is the strongest available argument against the entire logic of superfood lists, and it comes from inside the evidence rather than from scepticism about it.

Verdict: The Nordic diet contains almost the same foods as the Mediterranean diet and, with moderate certainty, does nothing to CRP, IL-6 or TNF-α.

2. "Green tea reduces inflammation"

B∅ Tested, no effect. A GRADE-assessed dose-response meta-analysis of 38 randomised trials found CRP WMD +0.01 mg/L (95% CI −0.14 … 0.15), IL-6 −0.34 pg/mL (95% CI −0.94 … 0.26) and TNF-α −0.07 pg/mL (95% CI −0.42 … 0.28) — none significant (Dehzad 2025, PMID 40160899).

The grain of truth, and it is a big one. In the same 38 trials every oxidative stress marker improved: MDA, TAC, SOD, GPX. Green tea does something measurable. It is not this. Antioxidant and anti-inflammatory are not synonyms, and green tea is the cleanest available demonstration that they come apart.

Verdict: Across 38 randomised trials green tea improved every oxidative stress marker and moved CRP by +0.01 mg/L.

3. "Seed oils (omega-6) drive inflammation"

D Overstated in the wrong direction. In the meta-analysis that separated the two fatty-acid families, the omega-6 subgroup lowered IL-1β more than omega-3 did (MD −0.05, p = 0.03) — the opposite of what the hypothesis requires (Huang 2026, PMID 42163966). The intervention that consistently moves markers is an added EPA+DHA dose of 1–3 g/day, across 96 trials (Khabir 2026, PMID 41568426).

Where it came from. 1990s biochemistry: linoleic acid → arachidonic acid → pro-inflammatory eicosanoids. The pathway is real on paper. In humans the conversion step is limited and regulated, and eating more linoleic acid does not raise tissue arachidonic acid proportionally.

Verdict: Seed oils are the most-attacked food of 2026, and what moves the markers is the omega-3 that gets added, not the omega-6 that gets removed.

4. "Nightshades trigger joint inflammation"

D Untested. A PubMed search on 2026-09-02 across several phrasings found not one human clinical trial, systematic review or meta-analysis testing nightshade removal against inflammatory markers or joint symptoms. What exists is animal and in-vitro work on individual solanaceae compounds, some of it pointing the other way (for example PMID 42652140 on potato metabolites in inflammation modulation and wound healing).

Where it came from. Solanine is a real alkaloid and is genuinely toxic in quantity — in sprouted potatoes. The claim jumped from there to "edible vegetables cause joint inflammation" with no intermediate study at any point.

The grain of truth: individuals can be intolerant of specific foods. That is individual, and it has nothing to do with a botanical family.

Verdict: There is not one human trial testing whether removing nightshades reduces joint pain; the claim rests entirely on a toxin in sprouted potatoes.

5. "Leaky gut causes systemic inflammation"

C for the mechanism in diagnosed disease · D for "leaky gut syndrome" in otherwise healthy people · D for the test sold to measure it. Category error — three separate things fused into one.

Intestinal permeability is a real, measurable phenomenon. Increased permeability in coeliac disease, inflammatory bowel disease and cirrhosis is documented. That a non-specific symptom cluster in healthy people is caused by it is not established.

And the measurement does not hold up. Zonulin is the marker most often sold for this. A 2018 study using mass spectrometry and Western blot in a 376-person cohort showed the widely used commercial ELISA does not detect pre-haptoglobin-2 — the molecule it claims to measure — and instead binds other proteins, probably properdin (Scheffler 2018, DOI 10.3389/fendo.2018.00022, PMC5807381). A 2019 replication reached the same conclusion (DOI 10.1371/journal.pone.0210728).

Verdict: Intestinal permeability is real, and the commercial test most often sold to measure it was shown in 2018 not to detect the molecule it names.

6. "Lower inflammation in 7 days"

D Untested, and structurally impossible to test at that length. One night of sleep deprivation moves no inflammatory marker; three or more nights are needed before anything shifts (PMID 40474574). Omega-3 interventions of 10–12 weeks failed to change IL-6 and TNF-α (PMID 42163966). Combine that with the measurement problem in chapter 4 — a hs-CRP change has to be roughly 55% down or 123% up before it can be distinguished from ordinary biological variation — and a seven-day protocol can neither produce nor detect an interpretable change.

The grain of truth, and it is why the claim sells. People genuinely do feel better after a week. Sleep, alcohol, calorie intake and attention all change at once. That is real. It is not a marker change, and the two are being sold as the same thing.

Verdict: A single night without sleep does not move inflammatory markers, so neither does a single week of clean eating.