Re: Consumption of ultra-processed foods and cancer risk: results from NutriNet-Santé prospective cohort
This paper and its media coverage epitomizes what is wrong with much of current nutrition research. It uses weak associations from dubious and flawed data to reinforce existing prejudices (prejudices which I partly share). Even if one accepts the findings at face value they offer no prospect of better understanding of the relations between dietary factors and health and no prospect of making any realistic recommendations for dietary change.
It is no surprise that people who eat a lot of processed food have a higher cancer risk because they are likely to be very different in many ways from people who eat little of them. The authors have identified many potential confounding variables which they have tried to quantify, often quite crudely. They then use sophisticated mathematical modelling to correct for many of the potential confounders. There is no statistical magic wand that accurately corrects for all potential confounders. The dietary assessment is also necessarily relatively crude. They then suggest that the residual small excess risk after the correction processes is caused by the high intake of “ultra-processed food”. Given the size of the effect and the inherent weaknesses in the methodology, I am unconvinced by that conclusion.
The range of foods of many different types included under the blanket heading “ultra-processed” is enormous and makes up a high proportion of most French and British people’s diets. We may yearn for a bygone era where people ate largely home-cooked food produced from fresh unprocessed ingredients or bought freshly processed foods like bread from artisan local producers. However, public health advisers must give realistic practical recommendations based upon robust evidence that it will be effective and have a reasonable prospect of being implemented. This paper offers no such prospects.
The authors call for bigger studies in different situations but how would these clarify matters? A weak association remains just that whatever the size of the cohort. They call for more studies about mechanisms but studies of this type have been ongoing for decades. What extra mechanistic studies are justified by their findings?
This study is just likely to encourage a stream of related studies that will soak up research resources and increase the mountain of published papers that add little to our scientific understanding or our ability to give clear and consistent advice about dietary improvement.
Rapid Response:
Re: Consumption of ultra-processed foods and cancer risk: results from NutriNet-Santé prospective cohort
This paper and its media coverage epitomizes what is wrong with much of current nutrition research. It uses weak associations from dubious and flawed data to reinforce existing prejudices (prejudices which I partly share). Even if one accepts the findings at face value they offer no prospect of better understanding of the relations between dietary factors and health and no prospect of making any realistic recommendations for dietary change.
It is no surprise that people who eat a lot of processed food have a higher cancer risk because they are likely to be very different in many ways from people who eat little of them. The authors have identified many potential confounding variables which they have tried to quantify, often quite crudely. They then use sophisticated mathematical modelling to correct for many of the potential confounders. There is no statistical magic wand that accurately corrects for all potential confounders. The dietary assessment is also necessarily relatively crude. They then suggest that the residual small excess risk after the correction processes is caused by the high intake of “ultra-processed food”. Given the size of the effect and the inherent weaknesses in the methodology, I am unconvinced by that conclusion.
The range of foods of many different types included under the blanket heading “ultra-processed” is enormous and makes up a high proportion of most French and British people’s diets. We may yearn for a bygone era where people ate largely home-cooked food produced from fresh unprocessed ingredients or bought freshly processed foods like bread from artisan local producers. However, public health advisers must give realistic practical recommendations based upon robust evidence that it will be effective and have a reasonable prospect of being implemented. This paper offers no such prospects.
The authors call for bigger studies in different situations but how would these clarify matters? A weak association remains just that whatever the size of the cohort. They call for more studies about mechanisms but studies of this type have been ongoing for decades. What extra mechanistic studies are justified by their findings?
This study is just likely to encourage a stream of related studies that will soak up research resources and increase the mountain of published papers that add little to our scientific understanding or our ability to give clear and consistent advice about dietary improvement.
“Substantial reliance on observational data for which causal inference is notoriously difficult also limits the clarifying ability of nutrition science” Ioannidis and Trepanowski (2018) https://jamanetwork.com/journals/jama/article-abstract/2666008?redirect=...
Competing interests: No competing interests