Don’t blame the eggs for what the cake did.
Last week there was a piece of tabloid science with a screaming headline about eggs being associated with an increased risk of diabetes. It went global and had huge pick up, especially in Australia.
I thought I’d wait for some of my favourite researchers to take a more detailed look at the paper (it’s still behind a paywall for me on research sites) and offer their own thoughts.
First up to the plate is Zoe Harcombe. Zoe is pretty famous for the way she can critically evaluate a piece of research and can often see findings that the researchers do not necessarily see and draw entirely different conclusions. She is quite brilliant at it and as a consequence is often viewed as a controversial figure.
The findings of the study were published in the British Journal of Nutrition and was titled “Higher egg consumption associated with increased risk of diabetes in Chinese adults”. The headline grabber from one of the findings said “PEOPLE WHO REGULARLY CONSUME ONE OR MORE EGGS PER DAY INCREASED THEIR RISK OF DIABETES BY 60%”. Diabetes can be described as the inability to handle glucose. Eggs contain very little carbohydrate so this is all very confusing.
Nutrition research is notoriously tricky, especially epidemiological studies as this one is. Amongst other things you need to take account of details such as: the limitations of whether people can remember what they ate when the study was being taken, what else they eat, their calorie intake, whether they smoke or drink, the health of the individuals involved, any medications involved, the type of health conditions they may have, where they live, their socio-economic status, ethnicity (Chinese), how many people involved (8,545), age (mean age was 51), gender etc.
In short, it is not a perfect data set. It means that researchers at best can only establish association and not causation, relative risk, not absolute risk and they also need to allow for health confounding factors – in this case females from disadvantaged backgrounds eating processed foods. Read on.
Zoe has picked apart the two sets of analyses in the paper. I have the detail if anyone wants it , but here are the main points:
The definition of eggs turned out to be important. In the study, the definition of eggs included “food containing eggs”. This is significant. What else were they eating?
In the first analysis, when the adults were split into quartile ranges by egg consumption and compared the quartile ranges with the first quartile (the best). Q3 was the ‘worst’, then Q2, then Q4. The highest egg intake was second best – didn’t make the headlines! Also an average egg is 50g and even people in the highest intake were not averaging one egg a day.
The second analysis looked at egg consumption between 1991 and 2009 by splitting people into three groups and is where the 60% claim came from – a statistical model that had not been fully adjusted. When Zoe fully adjusted the figure, it was not statistically different between the highest intake and the lowest intake. Harcombe states that the researchers would know that they should have fully adjusted the statistics, therefore the 60% is a lie.
When men and women were reviewed separately, there were no significant differences for for men for egg intake and diabetes.
For females – there was a processed food issue and the paper actually referred to it: “women prefer cakes, soft drinks and snacks - most of which are ultra-processed foods, increasing the risk of diabetes”.
That, according to Harcombe, should have been the headline – so don’t blame the eggs for what the cake did!
Often researchers are blinded by their own findings – especially when they are not what they expect them to be. Peer reviewers (people who check the research), often fall into the same trap.
Thank you Zoe for bringing sense to the discussion.
References:
https://pubmed.ncbi.nlm.nih.gov/33028452/
https://www.zoeharcombe.com/2020/11/eggs-diabetes-and-chinese-adults/