Ultra-Processed Food: Assessing Uncertainty in an Imperfect World

Abstract

Food processing has existed for millennia to ensure food is edible, safe and palatable. This has largely been considered to date based on the physical extent of processing. However, recent approaches have expanded this to consider the purpose (i.e. the reason) for the processing. Most commonly, this has been defined by Nova classification.

In this regard, mounting evidence links higher intake of ultra-processed food (UPF) with a range of poor health outcomes. However, the value of classifying food and drink by the extent and particularly the purpose of processing has been debated, with evidence limitations highlighted by scientific advisory committees for both UK and US governments to formulate dietary guidance.

Unfortunately, evidence is not perfect and rarely clear cut, particularly so in the world of diet and nutrition. Therefore, when assessing links between nutrients, food and diets with health outcomes, we must the consider the sources and extent of uncertainty in the evidence, and then evaluate whether the uncertainty may meaningfully impact on the conclusions. The evidence regarding UPF and the Nova classification is no different.

In this presentation, I outline in detail the Nova classification and what it captures in contrast to other processing classifications, before discussing sources of uncertainty and error when measuring UPF intake, including the latest evidence on validation assessments of diet assessment tools. I then review the evidence of associations in the observational literature between UPF and health outcomes, dissecting methodological aspects and sensitivity analyses to consider the extent of uncertainty and its impact on effect estimates. I consider alignment with experimental evidence, and conclude by summarising strengths and limitations of the evidence base, and key research recommendations to address current limitations.