There are reports of a study doing the round in the UK press this week that warn us that eating a protein-rich diet is ‘as bad for us a smoking’. See here and here for examples of stories appearing in the ‘respected’ broadsheets the Telegraph and Guardian respectively. Reading the headlines of these pieces gives the distinct impression, I think, that low-carb and ‘paleo’ eating puts people on a fast-track to an early demise, particularly from cancer and diabetes. Given my nutritional stance (generally pro-low-carb and pro-paleo), perhaps not surprisingly, I’ve had quite a few emails asking for my opinion on this research.
The study concerned involved asking people to log what they ate over the course of a day, and then following them over a period of about 18 years. This type of research is ‘epidemiological’ in nature, which means it looks at associations between things. This study found an association between higher animal protein intakes and increased risk of death from diabetes, cancer and death overall in individuals aged 50-65. However, just because such a link was found, does not mean it’s protein that is causing the apparently harmful effects.
It might be, for example, that individuals who eat a lot of animal protein might also smoke more, and this is the real explanation for the link between animal protein and increased risk of death. Researchers can ‘control’ for these other factors (called ‘confounding factors’), and this was the case in this study. However, this process is far from perfect, and still leaves us with data that can only ever show associations (and not causality).
It is perhaps worth bearing in mind that this was a relatively small study. But would a larger study have been better? Or multiple studies showing the same thing? The reality is that epidemiological evidence is limited in its power by the very nature of this research, and having bigger or more studies does not magically transform a sow’s ear into a silk purse.
Imagine a study that finds a correlation between ice cream sales and shark attacks. One can imagine that more ice cream is sold when the weather is warm, and more people are likely to end up in the sea when the weather is warm too. However, the idea that ice cream actually causes shark attacks is probably incorrect, right? And it doesn’t matter if you have a mammoth study or 20 similar studies showing such an association, the fact is ice cream still doesn’t cause shark attacks.
This inherent weakness of epidemiological studies is the major reason why, these days, I hardly ever write about or comment on this sort of research. If it wasn’t for the fact that this recent study has had so much attention and me so many emails about it, I wouldn’t be writing about it at all.
The way the study has been reported, though, is not nearly as circumspect as it should have been, I think. We can lay the blame at the feet of the ‘science’ journalists who have reported this study in what might be termed a ‘sensationalist’ way. But, likening protein-rich diets to smoking appears to have come from one of the study authors: Dr Valter Longo is quoted in the Telegraph as saying:
We provide convincing evidence that a high-protein diet – particularly if the proteins are derived from animals – is nearly as bad a smoking for your health.
I have news for Dr Longo: No he hasn’t.
But there’s another part of this ‘research’ that appears to have been lost in translation. In individuals aged 66 or older, high protein diets were found to be associated with a lower risk of death overall, including a reduced risk of death from cancer. According to the authors of this study, it seems that somewhere around our mid-60s our protein needs flip from low to high. If this sounds faintly ridiculous to you, then that’s because it is (I think).
Crucially, when all individuals over the age of 50 were considered, no relationship was found between protein intakes and risk of death or risk of death from cancer. This, I would say, is the bottom line of this study, but it’s got completely buried in the ‘Atkins gives you cancer’ rhetoric.
Interestingly, this overall result was the first result to be reported by the authors in the study, and my suspicion is that their original intention was to perhaps find something meaningful here and leave it at that. However (and this is idle speculation on my part), it seems that after finding nothing, they decided to see what ‘meaningful’ results they could dredge from the data. In scientific parlance, this is referred to as ‘post-hoc analysis’. A colloquial way to describe it is ‘shifting the goalposts’.
The end result is that a weak study showing nothing much gets interpreted in a way that I believe is sensationalist, over-hyped and misleading. The problem here is that it might lead many to doubt the appropriateness of how they are eating and lead them to a diet that is altogether less healthy (such as one rammed full of carbohydrate).
Researchers and scientists have a responsibility to report their findings accurately and honestly. I’m not sure Dr Longo has done that here, at all. Likening protein-rich diets to smoking on the basis of epidemiological evidence is not only unscientific but potentially damaging. I wonder whether Dr Longo will reflect on his soundbite and wonder whether he might have (even unwittingly) put headlines in front of public health.
1. Levine ME, et al. Low Protein Intake Is Associated with a Major Reduction in IGF-1, Cancer, and Overall Mortality in the 65 and Younger but Not Older Population. Cell Metabolism. 2014;19(3):407-417