Can you say eggs are good for you?

[andyswarbs]

There was a recent (2018) study of 500,000 people in china and their consumption of eggs. It resulted in a headline in the BMJ "Daily egg consumption may reduce cardiovascular disease" as indicated at eurekalert.org/pub_releases...

So are eggs good you? Does this study show that? Is the evidence from this study clear evidence? Was the BMJ headline an accurate reflection of the study?

Rather than taking my analysis let me quote from critiques of what went on by people more qualified than myself. My sources are healthnewsreview.org/2018/0... and sciencemediacentre.org/expe...

Let me start with the main thrust of the critique against the BMJ. "The news release leads with a headline that clearly makes a cause-and-effect claim about eggs reducing risk, and waits until the 16th paragraph to provide the all-important qualifier: “This was an observational study, so no firm conclusions can be drawn about cause and effect.”

It is important because all exciting headlines by a publication such as the BMJ get picked up and go viral across newsagencies globally in a matter of minutes. So this was picked up by the BBC, CNN, TIME and many more. All of these went with similar headlines as per the BMJ. These headlines then get repeated on forums such as this without analysing them further. The next think you know people are eating more eggs...

Now let us look at a few comments on the study itself.

1. "In a Western context, if you eat eggs with lots of refined white bread, processed meats like bacon and sausages and sugar-rich ketchup, that is materially different to eating an egg with whole-grain bread and vegetables for instance." Prof Nita Forouhi, MRC Epidemiology Unit, University of Cambridge.

2. "But an important limitation of this present study is that the people who consumed eggs regularly were much more affluent than those who avoided them." Prof Tom Sanders, Professor emeritus of Nutrition and Dietetics at King’s College London.

3. "To say that eating eggs is good (or bad) for you based on a study like this would be foolish as diet is much more complicated than picking on one foodstuff like eggs." Dr Gavin Sandercock, Reader in Clinical Physiology (Cardiology) and Director of Research at the University of Essex.

Finally if you want to interpret this study as good news that eggs are good for you, based on this study check out today's headline from the inquirer "an Egg a Day may keep the doctor away" cebudailynews.inquirer.net/...

[Hidden]

Hi Andy, to me its interesting when one particular food or food group is chosen for a study as its all about eating a healthy balanced diet for 'our' needs and comment 1 says that perfectly regarding eggs:

1. "In a Western context, if you eat eggs with lots of refined white bread, processed meats like bacon and sausages and sugar-rich ketchup, that is materially different to eating an egg with whole-grain bread and vegetables for instance." Prof Nita Forouhi, MRC Epidemiology Unit, University of Cambridge.

Statistics are just that and have to be kept in context...😊

[TheAwfulToad in reply to Hidden]

Even if you assume there is a cause/effect thing going on here, the effect is small. That suggests to me that the presence of eggs in one's diet is neither here nor there. One of the main problems with dietary studies these days is that researchers pick up on tiny effects (10% here, 20% there) which for all practical purposes are meaningless. If you observe a factor of two or three increase (or decrease) in risk, you can be pretty sure you're onto something, but it's rare that a study produces a result of that magnitude.

IMO they make this mistake because they misunderstand the underlying meaning of the p-value associated with their statistics. A low p-value simply means that your results were probably not due to chance; the smaller the number, the less likely it is that your outcome is due to random variation. 0.05 is typically taken as the threshold for 'significant', but that's really just a number plucked out of the air. If 'p' meets the threshold, all it's telling you is that your outcome would be observed purely by chance in 1 out of 20 experimental re-runs.

Some scientists overinterpret this, imagining that p<0.05 actually means their result has practical, real-world significance. It doesn't. It tells you merely that you've measured something with a reasonable degree of confidence. It doesn't confirm that you've measured what you think you're measuring, or whether that measurement is useful in any way.

It's also worth asking the question: is there any physiological reason why eggs should be bad for you? As far as we know the answer is "no", because we've tested a few suspicions (eg., cholesterol) and found they they produce a null result. Eggs are a common food for many different animals (including primates, if they can get them). Of course there could be something we've missed, but until an exact mechanism is elucidated, statistical associations don't mean much. In some places you can correlate live births with stork sightings, but not many people suggest there's a cause and effect.

If you're interested, the correlation between storks and babies is high enough to merit study funding (0.62) if you didn't already know what causes babies:

robertmatthews.org/wp-conte...

[andyswarbs in reply to TheAwfulToad]

I am put in mind of a point that Dr Michael Greger makes. There is a very high correlation between cancer and ash trays, because cigarrette smokers use ash trays, and largely no-one else does. But there is no way ash trays can ever be said to cause cancer, or indeed any other illness (apart perhaps from a headache if someone hits you with one!)

[NSNG-am in reply to andyswarbs]

Greger is completely biased an has an agenda. He said an egg is a 70 on the glycemic scale. How many other doctors or scientists support this claim?

[benwl in reply to NSNG-am]

Do you have a reference for Gregor saying that?

Is it possible that you are confusing glycemic index with insulin index?

[andyswarbs in reply to NSNG-am]

Greger website is entirely free with nothing to sell. Proceeds from his books go to charity. That kind of bias I like!

A reference to the egg=70 on glycemic index would be good. There is nothing, imo, like fact checking.

[NSNG-am in reply to andyswarbs]

Sure. Will provide reference tomorrow.

Heading up to Tahoe today to enjoy my Independence holiday.

[TheAwfulToad in reply to andyswarbs]

As usual, Greger is being intellectually dishonest. He's playing on the fact that his audience have no training in science or logic to bamboozle them.

Correlation might imply causation or it might not. You can take a guess at the direction of the causal arrow by looking at features of the real world. Do storks deliver babies? No, because even if we live in a primitive society that hasn't quite figured out what causes them, it's pretty obvious where they come from. Do ashtrays cause cancer? Of course they don't, because there is no feature of an ashtray that could conceivably cause cancer.

Hence my comment about referring back to physiological mechanisms before drawing conclusions.

[andyswarbs in reply to TheAwfulToad]

I think the "intellectually dishonest" phrase could bite you back. To state that his audience are all scientifically ignorant is patently false, just by looking at the statistics of website ranking. Alexa.com rates nutritionfacts.org at 26,000 globally and 8,500 in the USA. To imply all of those visitors are ignorant of science...

[TheAwfulToad in reply to andyswarbs]

If his audience - as a group - accept that stuff about ashtrays at face value then yes, the only rational conclusion is that they're either not engaging their critical faculties, or don't have any.

His website popularity is really not relevant to the validity (or otherwise) of his theories. Science is not a popularity contest. There are probably an equal number of people following websites about UFOs, telepathy, and the Illuminati. Especially in the USA

By way of contrast: I came across a video recently with Gary Taubes debating with Dean Ornish. I don't have anything good to say about Taubes there: he came across as supercilious and arrogant. Ornish, in contrast, was very good at presenting his views (he always has been). He gives the impression of having actually done his research, despite (IMO) coming to the wrong conclusions. He commands respect even if you don't agree with him. In short, he appears to be a man you could have an intelligent conversation with. Greger, on the other hand, irritates me because he uses tactics you'd expect from a snake-oil salesman, ie., making faulty associations that most people wouldn't spot, manipulating facts that most people wouldn't understand, glossing over the truth, or just lying.

[NSNG-am]

This study type sounds similar to the Adventist 2 study that is sometimes cited as definitive proof of a certain diet with health benefits. My understanding is these food questionnaire studies can only show an an association and not causation.

I fall on the eggs are good for you side.

[andyswarbs in reply to NSNG-am]

Since it is both very unethical and extremely expensive to run a double-blind all-singing-all-dancing trial we are left with an imperfect science. It is critical go with the science with a very closely eye on validity, and at the same time largely ignoring newspaper / magazine headlines.

Show me good quality keto/paleo research that definitively proves causation!

In the vegan foods for life group I posted healthunlocked.com/veganfoo... about vegans having hemorraghic strokes. That is not final proof research but it is important to raise the subject since if vegans know that just reducing salt it is likely to get rid of most of the risk, that's worth knowing.

[TheAwfulToad]

Nice succinct bit of common sense there!