Last week I wrote about the latest Eggs Will Kill You! study conducted by Dr. David Spence, who seems to be making a career of anti-cholesterol and anti-egg hysteria. (I’m sure the fact that he’s funded by statin-makers has nothing to do with that.) When I wrote about the study, I only had access to the abstract and some media articles. A reader in the research community later sent me a copy of the full study. There are some interesting bits in there. Let’s start with a paragraph from the introduction:
The underpinning of what used to be the step 2 diet and later became the diet recommended for CHD risk reduction by NCEP ATP III was a diet low in saturated fat (<7%) and dietary cholesterol (<200 mg). This diet if strictly applied tended to drive the consumer towards a more plant based diet with other potential advantages in terms of CHD risk reduction. In addition to saturated fat in meat (especially red meat) and full fat dairy products, eggs were also restricted due to their significant cholesterol content.
Right away we’ve got bias creeping into the text. Notice how “tended to drive the consumer towards a more plant based diet” is simply assumed to be a potential advantage in terms of CHD risk reduction.
Currently, however, serious doubts have been expressed over the relevance of these dietary components to cardiovascular disease.
That’s because they’re not relevant.
In the case of cholesterol much of the debate has been focused on the lack of clear consensus on whether egg consumption consistently raises serum cholesterol or impacts negatively on postprandial events, including vascular reactivity. Most importantly the association of egg consumption with CHD events in cohort studies has been inconsistent.
Well, there’s a reason the association of egg consumption with CHD has been inconsistent: eggs don’t cause heart disease. If they did, the association would be consistent. If the evidence supporting a hypothesis isn’t consistent, a good scientist assumes there’s something wrong with the hypothesis.
Here’s the description of how the data was collected:
In earlier years, data on smoking and egg consumption were recorded by patients into a lifestyle questionnaire at the time of referral. Since 2000, when our referrals were scheduled on an urgent basis soon after transient ischeamic attacks or strokes, a more limited set of lifestyle questions were asked at the time the history was obtained. These data were entered, along with the history, medications, physical examination and recommendations into fields in the database, from which clinic notes were generated.
Hmmm … so the study participants were people who’d been urgently referred to a clinic after suffering a heart attack or stroke. Not exactly what I’d call a random sample of the population, or even a random sample of the elderly population. And if egg yolks cause cardiovascular disease, why were the study participants who consumed less than one egg per week referred to a clinic for people who’ve had a heart attack or stroke? Seems to me they should have been out playing golf and enjoying their plaque-free health, not seeking an urgent referral.
The responses for smoking and egg yolk consumption were used to compute pack-years of smoking (number of packs per day of cigarettes times the number of years of smoking) and egg-yolk years (number of egg yolks per week times number of years consumed). This was not done for alcohol consumption, licorice intake or exercise, because the textual responses were mainly not quantifiable (e.g. “quit drinking six years ago”, “plays golf twice a week”).
I guess I’ll have to ask my Canadian pals: Why is licorice intake considered a potential confounding variable in Canada? Do you eat enough of it to skew health outcomes? Granted, it’s been a long time since I did standup comedy tours in Canada, but I don’t remember noticing a lot of people up there chewing on licorice. Nobody ever walked up to me after a set and said, “Great show, eh! Can I buy you a licorice?”
As for not being able to figure alcohol intake or exercise into the data, I’d consider that a serious matter. Later in the study, we also learn that waist circumference wasn’t included in the calculations. In other words, we have no idea if we’re looking at more plaque in people who simply eat more eggs, or in fat people who never exercise and also eat more eggs – the folks Dr. Mike Eades calls non-adherers and I call people who don’t give a @#$%.
After a Results section describing the association between egg-yolks years and plaque I covered in the previous post, there’s a Discussion section calling for (surprise) a reassessment of the possible role of eggs yolks in heart disease. Sure, the association is inconsistent, as we admitted earlier in the study, but since we managed to find one, it’s time to scare people away from eggs all over again.
My favorite paragraph in the study was this one:
The study weakness includes its observational nature, the lack of data on exercise, waist circumference and dietary intake of saturated fat and sources of cholesterol other than eggs, and the dependence on self-reporting of egg consumption and smoking history, common to many dietary studies.
Yes, those are weaknesses. Big ones … even bigger than not accounting for licorice intake. But after noting in the full text that the observational nature of this study is a weakness, Dr. Spence ran out and told media reporters he’d demonstrated that egg yolks make plaque build up faster — cause and effect.
This wouldn’t be such a lousy study if the lousy scientist who conducted it didn’t jump to lousy conclusions and share them with media reporters who wrote lousy articles about it.