Summary
Greger’s iron episode argues a single thesis: most people carry more iron than is healthy, the “sweet spot” is a ferritin of 15–50, vegetarians sit closer to it, and excess iron drives diabetes and cancer via free-radical damage. He distinguishes iron stores (ferritin) from frank anemia, and closes with practical repletion advice for those who are actually deficient. The framing is consistently directional toward “less meat, more plants,” and the strength of evidence behind individual claims varies widely — the weakest claims carry the most rhetorical weight.
Key points as presented
- Ferritin measures stores; “normal” 30–300, but healthy ≈ 15–50; >50 starts toxicity risk, <12–15 too low.
- Iron is a pro-oxidant; we have no excretion mechanism, so stores rise with age.
- Higher iron → higher type 2 diabetes risk across the normal range (+20% at average ferritin, +43% at high), supported by Mendelian randomization and a small phlebotomy trial.
- Cancer as a “ferrotoxic disease”; a blood-donation RCT showed 60% lower cancer death.
- Vegetarians have lower ferritin but are not meaningfully more anemic (vegans 8% vs meat-eaters 9% in the British cohort).
- Iron deficiency without anemia has no demonstrated harms (one study, even showed “better memory”).
- Repletion advice: alternate-day dosing, vitamin C, bisglycinate, empty stomach; diet over supplements long-term.
What holds up
No iron excretion mechanism / stores accumulate with age — Correct. Regulation is absorption-only; men rise from adulthood, women post-menopause. (High)
Heme iron is less down-regulated than non-heme — Accurate, and it’s the mechanistic basis for heme iron tracking with cardiometabolic risk. (High)
Higher iron status associates with T2D, with some causal support — The observational association is robust (multiple meta-analyses), and MR provides some causal signal. Magnitudes (~20%/43%) are plausible. (Medium) — see caveats below.
Alternate-day dosing, vitamin C timing, bisglycinate tolerability — Aligns with current absorption/hepcidin science (Stoffel et al.). Bisglycinate’s ~2× bioavailability is reasonably supported. (Medium-high) — relevant to your own bisglycinate use.
~75% GI side effects / ~1-in-4 discontinuation on ferrous salts — In range for full-dose daily ferrous sulfate. (Medium-high)
Vegans not meaningfully more anemic than meat-eaters — Consistent with the EPIC-Oxford anemia data he cites; the anemia/low-stores distinction is correctly drawn. (Medium) — caveat: iron deficiency without anemia is genuinely more common in vegans, which he soft-pedals.
Menorrhagia as leading IDA cause in menstruating women — Correct for that population. (High)
What doesn’t hold up (or is overstated)
The “60% lower cancer death from donating blood” claim — This is the load-bearing overstatement. The source is the FeAST trial (Zacharski). Reality: 1,277 symptomatic peripheral-arterial-disease patients, mean age 67, 98.5% male — not healthy general-population volunteers, and “calibrated phlebotomy,” not voluntary blood donation. Critically, no difference was documented between treatment groups in all-cause mortality and the secondary outcome of death plus nonfatal MI and stroke — the trial’s primary endpoint was null. Cancer was a secondary substudy: reduced cancer incidence (HR 0.65, 6.0% vs 9.3%) and the cancer-mortality HR ≈0.39 that yields his “60%.” Single trial, unreplicated, secondary endpoint, narrow population. Presenting it as “a thousand cancer-free individuals randomized to donate blood” is misleading framing. Holds weakly; the framing does not hold. (Low) ScienceDirect + 3
Ferritin as a clean iron marker for the diabetes/cancer claims — Major omission: ferritin is an acute-phase reactant. It rises with inflammation, insulin resistance, fatty liver, and alcohol. Much of the “elevated ferritin → diabetes” association is confounded/reverse-causal — insulin resistance raises ferritin. MR partly addresses this, but MR results for iron→T2D are mixed across studies, not the settled “genetic evidence supports a causal link” he asserts. (Medium that some causal effect exists; Low for the clean magnitude he implies)
The n=6 phlebotomy/insulin-sensitivity study — Real (Fernández-Real), but six men. The ~40% effect is directionally consistent with MR but the study itself is far too small to “demonstrate” anything. He presents underpowered evidence as confirmatory. (Low as standalone proof)
“Iron deficiency without anemia is harmless” — Overstated via cherry-pick. The single study he leans on (the “better memory” finding is almost certainly a multiple-comparisons artifact) is contradicted by several RCTs (Verdon 2003, Krayenbuehl 2011, Vaucher 2012) showing iron repletion reduces fatigue in non-anemic iron-deficient women. The honest statement is “evidence is mixed,” not “no conclusive evidence of harm.” (Does not hold as stated)
The 15–50 “sweet spot” / “>50 = toxicity” — A defensible minority position presented as established. The optimal ferritin range is genuinely contested; mainstream deficiency cutoffs (<30, sometimes <45 with inflammation) and the upper bound are debated. Treating 50 as a toxicity threshold is the aggressive end of the literature. (Low-medium)
Bottom line
The mechanistic spine (no excretion, heme vs non-heme, iron as pro-oxidant) and the practical dosing advice are sound. The disease-prevention thesis rests disproportionately on one null-primary-endpoint trial’s secondary analysis and on a biomarker (ferritin) confounded by inflammation — neither acknowledged. It’s a reasonable hypothesis dressed as a settled conclusion. Net: directionally plausible, evidentially oversold. (Overall confidence in his strong causal framing: Low-medium.)
