Objective data trials are neededBMJ 1996; 312 doi: https://doi.org/10.1136/bmj.312.7044.1479c (Published 08 June 1996) Cite this as: BMJ 1996;312:1479
- George Davey Smith,
- Yoav Ben-Shlomo
- Professor of clinical epidemiology Senior lecturer in epidemiology Department of Social Medicine, University of Bristol, Bristol BS8 2PR
EDITOR,—Paul Knekt and colleagues' study showing an apparent protective effect of consumption of flavonoids on the risk of coronary death illustrates the pitfalls of such investigations.1 Intake of flavonoids shows the same association with both non-coronary and coronary mortality. It either reduces biological susceptibility to many diseases or it serves as a marker for a complex of sociobehavioural factors associated with the risk of death.
Observational studies of diet and health have two main problems. Firstly, dietary intakes are socially and culturally determined …