Is Eating Eggs as Bad as Smoking? – Statistics Sometimes Lie

There’s a whole lot of buzz on the internet lately about the dangers of egg yolks. The headlines, mostly variations of “are egg yolks as bad as smoking”, are based on a Canadian  study published a couple weeks ago in Atherosclerosis, an academic journal. The study uses survey data to look at correlations between smoking, eating eggs, and total plaque area (a measure of arterial disease), concluding that “regular consumption of egg yolk should be avoided by persons at risk of cardiovascular disease”.

Unlike other criticisms*, I’m not concerned about the sample size of the study (1,262) or the population of people surveyed, most of whom had previously experienced a stroke or mini-stroke. The main problem with the study is a poor execution of the statistical analysis. (Warning: The following discussion is fairly wonky.)

In order to measure the true relationship between eating egg yolks and total plaque area (TPA), the number of egg yolks that you eat must be uncorrelated with other unobserved factors that affect TPA. Otherwise, you’re going to be picking up the effect of those things as well. Having no medical training myself, it’s not immediately apparent to me that this is problematic; however, a quick glance the main table from the study reveals severe statistical bias.

There are two proven risk factors (according to NIH) for Atherosclerosis included in this analysis – BMI and Total Cholesterol. The coefficients on both of these variables show up as negative and significant. This implies that having high cholesterol and being over-weight are good for your arteries – the exact opposite of what countless previous studies have shown. Having some knowledge of regression analysis, my first instinct is that by including egg-yolks in this analysis is confounding or picking up the effect of BMI and cholesterol. However, the authors interpret these results as if they were statistically unbiased.

Moreover they single out egg yolks (a relatively high-cholesterol food) as the culprit rather than the whites, stating in the conclusion that “meals high in cholesterol should not be consumed regularly by those at risk for cardiovascular diseases, as dietary cholesterol itself is harmful, and potentiates the effect of saturated fats”. However, their own analysis indicates that blood cholesterol is negatively correlated with TPA.

In conclusion, the jury is still out on the relationship between dietary cholesterol and serum cholesterol.  In an attempt to weigh in on the “dietary cholesterol is bad and should be avoided” side of the debate, the authors of this study neglected to preform a thorough statistical analysis, drawing conclusions that are not representative of their results. By all means enjoy your eggs, yolks and all.

*Mark @ Mark’s Daily Apple also notes that the number of egg yolks eaten is positively correlated with age, further confounding the analysis.