Presentation Details
| A critical review of the epidemiological, randomized controlled trial, and mechanistic data on the health effects of ultra-processed foods Richard Mattes, . Purdue University, West Lafayette, IN, USA |
Abstract
There is widespread concern about potential adverse health effects of consuming ultra-processed foods (UPF). Indeed, many nations are incorporating guidance on the intake of such items in their national policies. However, there are multiple definitions of UPF with low concordance on how to classify foods and, as result, their association with various health outcomes. To set health policy, it is desirable to have a convergence of epidemiological evidence (to established associations and at-risk populations), randomized controlled trial (RCT) data (to establish causality) and mechanistic findings (to guide interventions). To-date, there is ample, consistent epidemiological evidence linking UPF intake with adverse health outcomes. However, effect sizes are small, trends in intake are not consistent with trends in obesity (the focus of this presentation) and some sub-populations with high UPF intake have especially good health and longevity. Thus, the epidemiological evidence is still wanting. With respect to RCTs, no educational intervention has yielded beneficial effects on body weight and clinical trials show effects on food intake are transient and effects on body weight range from increased, to no change with high UPF intake, to decreased. Thus, causality has not been established. Multiple mechanisms for UPF effects on ingestive behavior and body weight have been proposed, but none, including a purported contribution of “hyper-palatability” is empirically supported. Some suggest existing data are sufficient to implement dietary guidance to eliminate UPF from the diet, but concerns about resulting increased food-borne illness, food waste, disproportionate burden on food insecure and single-parent households, and possible decreased diet quality suggest an unfavorable risk benefit assessment.
No part of this publication may be reproduced, distributed, or transmitted in any form or by any means, including photocopying, recording, or other electronic or mechanical methods, without the prior written permission of the author.
No part of this publication may be reproduced, distributed, or transmitted in any form or by any means, including photocopying, recording, or other electronic or mechanical methods, without the prior written permission of the author.