Low-Carb and Mortality: Analyzing the 2010 Fung Study
This page details Professor Bart Kay's critique of the paper "Low-Carbohydrate Diets and All-Cause and Cause-Specific Mortality," published in the Annals of Internal Medicine.
Study Overview
- Authors: Fung, Willett, et al.
- Population: 85,168 women and 44,548 men (Nurses and Health Professionals).
- Finding: Reported that animal-based low-carb diets were associated with a "modest increase" in overall mortality.
Key Criticisms
1. Reliance on Food Frequency Questionnaires (FFQs)
The study uses self-administered questionnaires to estimate diet over 26 years. Professor Kay argues these are invalid because they rely on memory and are subject to "social desirability bias," especially among health professionals.
2. The Calorie Confounder
Kay points out a significant unaddressed variable:
- In the women's cohort, the "High Low-Carb" group (Decile 10) ate roughly 200 fewer calories per day than the "Low Low-Carb" group (Decile 1).
- This makes it impossible to separate the effects of carbohydrate intake from the effects of caloric restriction.
3. Raw Data vs. Adjusted Results
The researchers reported a higher "Hazard Ratio" (relative risk) for low-carb eaters. However, the raw incidence data showed:
- Decile 1 (High Carb): ~9.8 deaths per 1,000 person-years.
- Decile 10 (Low Carb): ~9.2 deaths per 1,000 person-years.
The reported "increase" in risk only appeared after researchers applied multivariate regression to "adjust" the numbers, which Kay describes as a form of statistical fabrication.
4. Absolute Risk is Negligible
Even using the study's own numbers, the absolute difference in mortality was roughly 5 deaths per 10,000 person-years. For any single individual over 20 years, the probability change is statistically zero.
Summary of Findings
| Feature | Analysis |
|---|---|
| Study Type | Observational/Epidemiological (Not an experiment). |
| Causality | Cannot prove cause and effect. |
| Primary Flaw | Misreporting via FFQs and caloric intake differences. |
| Conclusion | No evidence of increased mortality for low-carb animal-based diets. |
"There is nothing here that would be of any utility to any given living human being... the findings are so close to zero as makes absolutely no difference."