Hi guys,
Over the last few days we have had a lot of our clients at the RISE Group and 1:1 Personal Training Studio in Buxton, Derbyshire ask us about the study released this week that supposedly showed the eating red meat is bad for you. The study, by Walter Willett colleagues at Harvard has been reported as showing that increasing red meat consumption increases all causes mortality by 20%.
It’s great that so many of you are asking about this and our opinions on it. We like when people question what they hear – whether it comes from us or other sources. What we ultimately aim for ourselves is to filter through the mountains of contradicting information out there, determine the validity and relevance and decide what we feel, on balance of evidence is the most likely to be correct. Having just read through the study, I have the following points to raise, which question it’s validity (which I will expand on):
- It’s an observational study
- It shows a very weak correlation, particularly for an observational study.
- It also showed that eating some red meat is better for you then eating none
Point 1: It’s an observational study
In this study researchers studied previous studies which tracked 37,698 men for 22 years and 83,644 women for 28 years. In these studies subjects had answered surveys about their eating habits every four years. Death from cancer, heart disease and all causes was then compared to the results of the questionnaires.
Supposedly those who ate a card-deck-sized serving of unprocessed red meat each day on average saw a 13 percent higher risk of dying than those who did not eat red meat as frequently.
Observational studies are to be taken with caution at the best of times. Good science will attempt to prove causation between variable A and outcome B. It will want to show that when you adjust A by a set amount, something predictable and consistent will happen to B. Good science will also look at every possible reason why something may be the case and it will vigorously attempt to disprove a hypothesis, and only when this doesn’t happen, will it suggest that the hypothesis is correct – that A is causing B.
Observational studies will usually succeed in finding a correlation (ie: two things that seem to go up or down or whatever together) but often make little attempt to prove causation. Even when they do take into account other factors, they will often attempt to mathematically remove the effect of these influences, but doing this with sufficient accuracy is nigh on impossible.
In this reporting of this case the two variables compared are red meat consumption and ‘all causes mortality’. Looking at the opposite ends of the spectrum, they found that those who eat the most red meat and more likely to die then those that eat the least. As phrased in the questionnaire used it essentially compared the group that answered that they ate red meat “never or less than once per month” to those that answered “6 or more times per day” (to be precise, the two groups being compared ate 0.22 and 2.36 and 0.53 and 3.1 portions of red meat per day in the two studies used).
As we have been told for decades now to lower red meat consumption, I should imagine that those that chose to eat the most red meat generally care less about their health and well-being then those that eat the least. I think it is fair to say they are more likely to smoke, drink, take drugs, eat processed food, do less exercise, drive too fast and cross the road without looking (amongst other things). I have seen these people referred to as the ‘don’t give a shit’ group. Is it fair to say that it is the red meat alone that is causing higher death rates in this group? I don’t think so.
Whilst, to their credit, they have made an attempt to mathematically adjust for some of the other variables, the accuracy of these sort of calculations are to be taken with a pinch of salt. Please see Zoe Harcombe’s analysis of the study at https://www.zoeharcombe.com/2012/03/red-meat-mortality-the-usual-bad-science/ for a full dissection of the maths used.
You can use observational studies to prove pretty much anything if you want to. Ancel Key’s 1953 ‘Seven Countries’ experiment supposedly showed that the more ‘Mediterranean’ diet (higher in pasta, bread, etc) reduced all causes mortality. The fact that Keys picked the seven populations he knew would confirm his hypothesis, only measured a few variables and didn’t account for others, and that the experiment, when repeated with seven random populations, always shows no correlation or the opposite, raises questions about it’s validity. These questions should also be asked of this study.
Point 2: It shows a very weak correlation
The increased risk of death for high red meat consumption quoted is 13%. I will ignore the 20% number reported in the press as that was for the higher levels of processed meat and we have already said thousands of times that the more processed foods (of any sort) are worse for you and we constantly try and encourage our clients to eat as naturally as possible. This is, we feel, a no-brainer!
So, a 13% difference. Of dying from ANY cause. You don’t need to be a statistician to see that’s not really a massive number. In an observational study I would say it is so low that it can be ignored. Most epidemiologists would say a correlation of 300-400% would be the minimum they would take seriously and feel worthy of further investigation. Ie: Outcome B is 3 to 4 times more likely in populations where variable A is at it’s highest versus populations where it’s at it’s lowest.
When the first observational studies were done into the effect of smoking on lung cancer it was found that the groups who smoked the most were 20 times more likely to get lung cancer then non-smokers. A 153 times bigger correlation then the red meat one. This number was, obviously, seen as high enough to, at least, warrant further investigation.
Bear in mind that we all have a 100% chance of dying anyway. It is very difficult to ascertain the affect of one variable on the time frame for this.
Point 3: The study also showed that eating some red meat is better for you then eating none
See https://archinte.ama-assn.org/cgi/content/full/archinternmed.2011.2287 for the raw data from the study. It can be seen that death rates fall with increased red meat consumption up to the third and fourth quintile. Ie: those that ate moderate to moderately high amounts of red meat had lower death rates then those that ate none or very little.
It also showed (in one of the studies) across all centiles, that higher red meat consumption is associated with LOWER cholesterol levels. Interesting! In fact, in that experiment, those that ate the most meat were half as likely to have high cholesterol then those that ate the least. Why wasn’t that reported?
So, in summary, please take this study with caution and, if you really want to follow it to the letter, keep your red meat to the 1.5 to 2 portions a day which were shown to be better for you then less then that.
