A UK based study into the health benefits of vegetarian diets – the largest of it’s kind to date – has just published its findings in The American Journal of Clinical Nutrition showing that long-term diets reducing or eliminating meat and other animal products did not lead to significant reductions in early death. Not only does this call into question the claimed health benefits of vegetarian diets, but is also challenges recent warnings about the risks of eating meat.
The study by Appleby et al at the Cancer Epidemiology Unit at the University of Oxford wanted to find out if previous evidence indicating protective effects of vegetarian diets against certain chronic diseases actually translated into extended life (i.e. reduced mortality).
They used data from two UK based studies that tracked a total of 60,310 persons recruited between 1980 and 1993 and tracked until 2014, representing a total of over 1 million person-years of follow-up.
Based on a simple questionnaire participants were categorised to one of five dietary groups: Regular meat eaters – who ate meat 5 or more times per week; Reduced meat eaters – who ate meat less than 5 times per week; Fish eaters – who avoided meat but ate fish; Vegetarians – who ate eggs and/or dairy products but no meat or fish; and vegans who ate no food of animal origin at all. These dietary patterns were rechecked every five years.
During the study period over 5000 deaths were recorded, providing data to estimate the relative risk of death among each group. The graph above shows the relative risk of death for each group after adjustments for the usual suspects: smoking, alcohol, age, marital status, supplements etc. As you can see there was very little difference between the groups.
Although the vegans appear to have a raised risk, there is more uncertainty in the association as they were the smallest diet group. This is indicated by the larger 95% confidence interval whiskers, so the apparent increase is actually non-significant. It might be tempting to use these data to criticise vegan diets, but that would not be statistically valid.
This is still interesting, as the vegan group had several dietary factors which conventional wisdom would expect to be in their favour: they consumed the lowest level of saturated fat and had the highest intakes of fruit, vegetables and fibre. All of these would usually be considered part of a healthy diet. In contrast, the meat eaters had the highest levels of saturated fat and the lowest levels of fruit and vegetables, yet their all-cause mortality was similar to that of the vegans.
So what about earlier studies which found evidence for protective effects from vegetarian diets, and those that indicated an increased risk of bowel cancer from red meat – you know – the ones that made the headlines and six o’clock news? How do we square those with the ‘no overall difference’ found in this study? The answer appears to be that each diet increases risks of some diseases whilst reducing risk of others.
When Appleby et al looked at the causes of death between the groups they found some clearly significant differences, which we will look at below. This is somewhat surprising considering that the overall mortality was so similar between groups, but it highlights how focussing on a single cause of death can lead to mistaken conclusions. Let’s take a look at some of those between-group differences.
Following October’s media frenzy surrounding the decision of the WHO to classify red meat as a ‘probable carcinogen’ and processed meat as a ‘definite carcinogen’ statistics appeared in many papers claiming to quantify the increased risk of colon cancer caused by consuming red or processed meat. One such from the Guardian on the 26th October 2015 said:
An analysis of data from 10 studies estimated that every 50 gram portion of processed meat, if eaten daily, increases the risk of colorectal cancer by about 18%… [whilst] data from the same studies suggest that the risk of colorectal cancer could increase by 17% for every 100 gram portion of red meat eaten daily.
Fortunately, we can put these claims to the test as in the new study Appleby’s group gives us the daily intake of processed meat for each diet group:
|g/day||Regular meat eaters||Reduced meat eaters||Fish eaters||Vegetarians and vegans|
If the stats in the Guardian are to be believed, then the regular meat eaters should have about a 20% increased incidence of colorectal cancer than the non-meat eaters. But what did the Oxford study find?
Of the non-meat eaters, only the fish eaters had a significantly reduced risk of dying from colorectal cancer, a whopping 40% fewer fish eaters dying of that cause. Yet contrary to expectations, the vegetarians and vegans gained no significant benefit from avoiding meat. This is weird. And it challenges many of the putative mechansisms by which meat might initiate cancer. The front runner in the cancer-inducing steaks (pun intended) is thought to be the oxidation of fatty acids by heme iron (from fatty red meats), but if that were the chief mechanism then vegetarians should be protected from colon cancer to the same extent as fish eaters.
One possibility would be that fish may contain protective factors, rather than red meat containing harmful ones. But this seems unlikely, as the regular and reduced meat eaters ate similar amounts of fish per week as the pescatarians. There is something fishy going on here…
Looking at deaths from all kinds of malignant cancers it is possible to distinguish between vegetarians and vegans, and the picture again is much of a muchness. The fish diet is the only group showing a significant difference:
The second biggest killer after cancer is cardiovascular disease, so lets take a look at what this study found.
It is commonly assumed that meat eating is associated with heart disease. Indeed meat, animal fats and saturated fats are often conflated, and in many people’s minds all contribute to cardiovascular disease. Yet in this study there was no statistical difference between the diet groups for deaths from cardiovascular disease in general or ischaemic heart disease in particular.
In the case of cerebrovascular disease (which includes stroke), however, an interesting picture emerges:
In this case meat eating appears to be protective, with the vegan’s having a significantly increased risk of dying from cerebrovascular disease. There is possibly a smaller raised risk for fish eaters too, although the broad 95% confidence intervals make it hard to draw any firm conclusions. These findings are strange as there were no obvious dietary differences between the vegans and fish eaters when it came to the usual ‘culprits’ saturated fat, total fat, fruit/vegetable consumption etc. As the authors conclude:
Differences found for specific causes of death merit further investigation.
Strengths and weaknesses of this study
Usually observational studies can be criticised because they can only demonstrate correlation not causation. However, when a sufficiently powered observational study like this one finds no correlation (all 5 diets being essentially equal with respect to mortality) then de-facto there can be no causation.
The take home messages seems to be:
- Vegetarian and vegan diets do not appear to be especially protective, and meat eating is not especially harmful.
- Fears about red and processed meat and colorectal cancers appear unfounded.
- Whilst choices around vegetarian v non-vegetarian diets can indeed change risk factors in relation to specific diseases, they are unlikely to affect overall life expectancy.
Does this mean there is no point in trying to eat a healthier diet? No. For one thing, if you have a heightened risk of a particular disease – because you have a specific medical condition, genetic weakness, or family history for example – then it makes sense to eat a diet that minimises that risk. In such a case it will improve your overall mortality risk.
Secondly, the participants in this study were all eating ‘standard’ versions of these diets with macronutrient ratios close to the national norms, i.e. high-carb with lowish protein and fat. Below I have listed the participants’ macro nutrient consumptions and compared them to the national average taken from NDNS data 2011/12. You can see from the variations (+/- figures) that none of the participants’ diets varied much from this norm.
|Baseline diet; All participants||Energy, %||UK average (NDNS 2012)|
|Total fat||31.5||± 6.0||35|
|Saturated fat||11.1||± 3.5||12.6|
So this study tells us nothing about the long term benefits of a low-carb, gluten-free or paleo diet. What it does show though, is that there is little reason to worry about your meat intake and even less reason to be distracted by vegetarian or vegan health rhetoric.