Published by the Blogs Team, June 2023
Dr Guess is a clinical and academic dietician, with research and clinical interests focusing on type two diabetes and cardiometabolic disease. She has been the principle or coinvestigator of more than 10 trials in diet and type two diabetes. She currently leads a large type two diabetes remission program at the University of Oxford. Dr Guess kindly agreed to be our first webinar guest of the ENLP 2023 webinar series.
Personalised nutrition has been one of the hotly debated topics in the recent years – from a research and business point of view, and Dr Guess navigated brilliantly the evidence up to date to compare its benefits with more generic healthy diets.
Personalisation of diets can be carried out in different ways, said Dr Guess. She grouped these into 3 main categories:
- Personalised diets based on preferences and taste
- Personalised diets based on measurable phenotypes (e.g. high blood cholesterol)
- Personalised diets based on biological differences (genome, microbiome, metabolome…)
The first benefit of personalisation would be a better adherence to that diet, that we expect to lead to better outcomes, which will be the desired end result.
Dr Guess has focused her critique of the evidence in this space on the 3rd type of personalisation, i.e., based on biological differences. This is gaining the most traction for businesses and within popular media, but she briefly reviewed the others within the webinar as well:.
- To illustrate a personalised diet based on preferences, she mentioned the Food4Me randomised European trial. This trial showed how a personalised diet based on what people are currently eating, there is a better improvement of behaviour thanks to a better adherence to dietary changes.
- Personalised diets based on phenotype (e.g. tailored to cholesterol, glucose, blood pressure) has been illustrated by the DASH study. The trial has shown very strongly that tailoring a diet (to lower blood pressure) results in better outcomes than generic healthy eating advice.
- Dr Guess then focused on the “trendy personalisation” -personalised diets based on “-omics”! Why that is trendy? Because personalising based on the genome is exciting and so is the machine learning to optimize nutrition. We see this area raising with a lot of funding.
Even though people have a unique biology, do we see any evidence that their biological state is leading to a different diet recommendation? This uniqueness is the selling point of companies working in the space of personalised nutrition: you are unique – therefore, you need a unique diet!
To answer this question, she shared the very first study which has presented data for that. This was an ambitious study design with 800 people, with an algorithm designing a diet based on the data collected in order to get lower glycaemic responses. The result? The personalised diet was recommending roughly a low carb high fat (or high protein) diet. There was no evidence of variations of dietary recommendation based on measures that were inputted into the algorithm.
Similarly, PREVENTOMICS study’s algorithm didn’t result in differential dietary changes compared to their control group, except on higher fibre recommendations. Her take-home was:
- There doesn’t seem to be evidence that a unique biology could lead to differential diet recommendations.
The second crucial question when it comes to personalised diets, she underlined, is whether the analysis of “-omics” leads to better outcomes and ultimately if these diets are more effective.
Dr Guess showed us how a study having better outcomes (on blood glucose) with a personalised diet compared to a Mediterranean diet, has in fact very similar outcomes to a study looking at a high protein low carb diet, putting into question the “personalisation”.
The PREVENTOMICS study didn’t show either any evidence that different biological states could lead to differential & personalised diets, where outcomes were very similar to their control diet. She summarised:
- As far as the evidence suggests, there is no greater benefits related to personalised diets.
Dr Guess concluded that, to date, there no evidence that personalised diets are better or more efficient than generic healthy advice. Of course, with a lot of data collected, and a lot of publications she emphasized that the temptation to rely on the endless correlations we find is huge. The question is, do these correlations matter clinically in the middle of the amount of data we have?
As of today, we still need evidence that algorithms make unique dietary recommendations, and well-designed trial data with pre-specified outcomes. And to recognize that we already have good evidence that some types of biologically driven diets that have great outcomes (e.g. phenylketonuria).
You can find the full replay of the webinar online via our You Tube channel:
