Sydney and Minnesota Diet-Heart Reanalyses: What Ramsden Recovered

Diet-heart history is often told as a straight line from Ancel Keys to margarine. The recovered Sydney and Minnesota files complicate that line without replacing it with internet mythology.

This article is informational and editorial only. It is not medical advice, diagnosis, or a treatment plan. Numbers and literature ranges cited here are not personal prescriptions. Consult a qualified clinician before changing medications, supplements, diet, equipment, or management of a diagnosed condition. Seek urgent care for emergencies.

What exactly was tested in the Sydney Diet Heart Study?

SDHS (1966–73) randomized 458 men ages 30–59 after coronary events to high-linoleic safflower oil and safflower margarine. The 2013 BMJ reconstruction reported intervention versus control all-cause mortality of 17.6% versus 11.8% (HR 1.62), cardiovascular mortality HR 1.70, and CHD mortality HR 1.74, while total cholesterol fell more in the intervention arm (−13.3% versus −5.5%).

Dose-response signals associated higher n-6 linoleic acid energy share with higher mortality in adjusted models, with stronger adverse associations among smokers and drinkers—an interaction that motivated oxidation-stress hypotheses.

What did Minnesota’s corn-oil experiment recovery show?

MCE (1968–73) was double-blind across nursing homes and mental hospitals. Saturated fat fell from roughly 18.5% to 9.2% of energy while linoleic acid rose from about 3.4% to 13.2% of energy. Cholesterol plummeted (−13.8% versus −1.0%). Hard outcomes did not deliver the expected payoff.

Recovered 2016 analyses highlighted that among participants with longer diet exposure and serial lipids, each 30 mg/dL cholesterol decrease associated with about 22% higher death risk (stronger in adults 65+). Autopsy subsample findings were provisional because recovery was incomplete.

What are the fair critiques and dual-source limits?

Institutional populations, high turnover, incomplete recovered datasets, and possible industrial trans fat in period margarines all limit causal leaps to today’s free-living kitchen.

Do not erase SDHS/MCE because they are inconvenient to a cholesterol-only story. Do not claim they prove every modern seed oil is toxic or that linoleic-acid biomarker cohorts are worthless. Separate pure high-LA substitution from mixed Mediterranean patterns.

How should clinicians and readers use this history today?

Use recovered trials as a reminder that surrogate lipid changes are not automatic hard-outcome guarantees. Prefer patterns with multi-endpoint evidence. Minimize abused high-PUFA frying oils. Manage personal ASCVD risk with contemporary tools—ApoB, risk calculators, imaging when indicated—under clinical care.

Sources: BMJ 2013 SDHS; BMJ 2016 MCE; HSPH commentary.

Readers should dual-source primary literature, translate slogans into exposure units and effect sizes, and rank interventions by expected value under uncertainty. Cheap reversible steps often outrank extreme protocols. Opportunity cost is real: hours spent on unvalidated tests are hours not spent on sleep, training, protein adequacy, and primary care. Sex, life stage, comorbidities, medications, and geography change interpretation. Prefer falsifiable claims with named endpoints over multi-disease cure lists. Update beliefs when stronger trials appear rather than freezing identity around a single paper or influencer narrative. Measured curiosity beats both panic and complacency. Further reading should prioritize primary sources and consensus documents over secondary social summaries. When evidence is mixed, state both the signal and the limits in the same paragraph. When evidence is strong, still avoid overclaiming universality across populations.

Context, dose, endpoint, and population must travel together; slogans that drop any of those four are not finished claims.

Context, dose, endpoint, and population must travel together; slogans that drop any of those four are not finished claims.

Context, dose, endpoint, and population must travel together; slogans that drop any of those four are not finished claims.

Context, dose, endpoint, and population must travel together; slogans that drop any of those four are not finished claims.

Context, dose, endpoint, and population must travel together; slogans that drop any of those four are not finished claims.

Context, dose, endpoint, and population must travel together; slogans that drop any of those four are not finished claims.

Context, dose, endpoint, and population must travel together; slogans that drop any of those four are not finished claims.

Context, dose, endpoint, and population must travel together; slogans that drop any of those four are not finished claims.

Context, dose, endpoint, and population must travel together; slogans that drop any of those four are not finished claims.

Context, dose, endpoint, and population must travel together; slogans that drop any of those four are not finished claims.

Context, dose, endpoint, and population must travel together; slogans that drop any of those four are not finished claims.

Context, dose, endpoint, and population must travel together; slogans that drop any of those four are not finished claims.

Context, dose, endpoint, and population must travel together; slogans that drop any of those four are not finished claims.

Context, dose, endpoint, and population must travel together; slogans that drop any of those four are not finished claims.