The decision (Challenge #12). Should low-direct-heat cooking (steaming, slow-cooking, sous-vide) be preferred over frying/roasting/grilling because it produces fewer advanced glycation end products (AGEs)? The efficacy question is whether eating fewer dietary AGEs moves a health outcome.
Efficacy verdict — some surrogate markers move, the headline cardiometabolic ones do not, no hard outcomes
Baye et al. 2017 (MA of 17 RCTs, 560 participants, low- vs high-AGE diets):
- What improved (all surrogates, and small): «Low AGE diets decreased insulin resistance (mean difference [MD] −1.3, 95% CI −2.3, −0.2), total cholesterol (MD −8.5 mg/dl, 95% CI −9.5, −7.4) and low-density lipoprotein (MD −2.4 mg/dl, 95% CI −3.4, −1.3).» (Baye et al., 2017) The LDL effect is −2.4 mg/dl ≈ −0.06 mmol/L — trivially small; the TC effect ~−0.22 mmol/L. Some inflammatory markers moved too (TNF-alpha −4.7; adiponectin +7.0), but CRP did not (−0.06 mg/dl, 95% CI −0.4, 0.2, NS) — so even the “inflammation” story is mixed.
- What did NOT move — the parameters a person actually cares about: «There were no changes in weight, fasting glucose, 2-h glucose and insulin, haemoglobin A1c, high-density lipoprotein or blood pressure.» (Baye et al., 2017) Weight, glycaemia (HbA1c −0.01%, essentially zero), and BP — all null.
- No hard outcomes. Every endpoint is a surrogate marker; there is no event or mortality trial of dietary-AGE reduction. So the causal transmission from fewer dietary AGEs to a patient-important outcome is unevidenced -> Surrogate Outcomes.
Symmetric-standards / halo flag: the senior author (Uribarri) leads the dietary-AGE research programme that advocates the hypothesis, and the abstract’s upbeat framing («may be an effective strategy») sits on a result where most headline cardiometabolic parameters were null and the movers were small-magnitude lipids and inflammatory markers of uncertain clinical meaning. Hold the reassurance to the same bar as any advocacy-adjacent finding.
The claim decomposed — the recommendation outruns the evidence
The challenge’s operative word was «should be preferred» — a recommendation. Held evidence supports only a weak version:
- Dietary AGEs -> surrogate markers: a real but small effect on some lipids/inflammation, null on the outcomes that matter, no hard-outcome data. This does not license «should be preferred» as a health recommendation for the general person.
- A separate, firmer axis for gentler cooking exists but is narrow: high-temperature cooking of red/processed meat forms heterocyclic amines / PAHs, a WCRF-graded carcinogen mechanism (Red and Processed Meat and Cancer) — a reason to avoid charring those foods, distinct from the AGE argument and not generalizable to all browning.
- A third, minor axis: high heat also degrades protein quality (Maillard/reactive-lysine loss -> Protein Quality and the DIAAS Score). Another small nudge toward gentler cooking, on a different axis again.
- Net effect, not intended effect: a recommendation must weigh the realistic alternative — charring and browning carry palatability and adherence value, and the wiki holds no evidence that the net of switching cooking methods improves any outcome.
(inferred from Baye et al., 2017)
Decision relevance
- The strong claim is not supported: low-heat cooking should be preferred as a general health recommendation outruns the evidence — the effect is on small surrogates, null on weight/glucose/BP, and untested on hard outcomes.
- A defensible weak version: for someone eating a lot of high-temperature-cooked red/processed meat, gentler cooking has a narrow, firmer rationale (the HCA/PAH carcinogen axis), not the AGE axis.
Limits
- Single MA,
confidence: low, surrogate-only, advocacy-adjacent authorship. A hard-outcome trial could revise this in either direction; the expectancy test says the field is under-studied on outcomes, not that AGEs are proven harmless or harmful.
Self-critique [run 2026-07-29, before commit]
- Halo/COI guard: FIRED. The Uribarri authorship and the upbeat abstract framing are flagged, and the verdict leans on the null headline parameters rather than the small movers.
- Over-claim check: the page refutes the recommendation (should-be-preferred) without asserting AGEs are harmless — insufficient-evidence, not no-effect.
- Cross-axis honesty: the HCA/PAH-meat axis and the protein-quality axis are kept separate from the AGE axis, not blended into a single gentle cooking is better halo.