* This week's academic paper generated global news coverage. The headlines were "An avocado a week can cut heart disease risk by a fifth."
* The paper used data from the two US population studies that are used incessantly to produce epidemiological papers – the Nurses’ Health Study (women) and the Health Professionals Follow-up study (men). Data for 68,786 women and 41,701 men were used in this study, which had 30 years of follow-up.
* The headline claim came from comparing non-consumers with those who consumed one avocado or more each week. This was for men and women pooled together. However, two tables in the paper confirmed that there were no significant results for women. Avocado intake made no difference for women. This should have been made clear. There were many other subgroups for which the headline claim did not hold. These should also have been made clear.
* The three standard flaws of all epidemiological studies applied to this study. i) this paper could only establish association and yet the headlines more than implied causation. ii) the headline “fifth” was relative risk; the absolute risk was tiny. iii) the healthy person confounder is significant with an expensive food like avocado. Affluence was not adjusted for.
* All epidemiological papers suffer from these limitations. This paper suffered even more limitations than usual.
The rest of this article is available to site members, who get access to all articles plus a weekly newsletter.
To continue reading, please login below or sign up for a membership. Thank you.