While debates about the relative healthfulness of carbohydrates/type of fat/salt/sweeteners/etc. never cease, there is one nutritional message that can unite (almost) anyone: eat less processed foods. On the surface, this seems obvious; after all, the top sources of calories for adults in the US include grain-based desserts, burgers, sugar-sweetened beverages and chips (1,2), which are certainly not what dietary recommendations are going for. But when we pose “What are the effects of food processing on health?” as a scientific question, things suddenly become very complex. Is there a consensus on what food processing is? Is all food processing unhealthy? What research has been done on processed food?
What is Processed Food?
First, we should start on the tricky subject of definitions.
Jones and Clemens published a nice summary of how various organizations differ in their definitions of “processed food” (3). The United States Department of Agriculture and other government organizations, for example, tend to use a broad definition that includes any changes from a natural state – such as washing, heating, and cooking, to adding preservatives, flavors, additives, etc. The American Institute for Cancer Research uses the term “minimally processed” to denote vegetables, grains, and beans prepared without much added to them, or minimal refining. Then there is the “NOVA” system, an attempt by Dr. Carlos Monteiro and colleagues to categorize food processing into: “unprocessed or minimally processed foods”, which are essentially unmodified from nature or include light processing such as drying, boiling, freezing, etc., “processed culinary ingredients”, “processed foods”, and “ultra-processed foods” as the extreme, which includes foods made with ingredients with limited “unprocessed foods”, such as concentrated energy sources like sugars, oils, fat, salt, extracts like casein, lactose, and whey, or additives like dyes, flavors, sweeteners, etc. (4). There doesn’t appear to be good published data on what the public considers processed food, although the International Food Information Council (IFIC) has developed definitions with input from consumer focus groups, discussed below.
What the Critics Say
Some critiques of definitions push against a broad classification of food processing because it could put washing an apple and baking a cookie in the same basket. To this end, while the NOVA system is an attempt to standardize food processing classification for study, not everyone agrees with it. For instance, Dr. Mike Gibney and colleagues published a commentary last month in AJCN arguing that the system is not superior to using associations of nutrient intakes with disease (5). In essence, they argue that the NOVA definition of “ultra-processed” foods and beverages is too subjective and simplistic; that classifying solely by processing would not be of use to study disease links to specific micronutrients (i.e., goiter, allergies and intolerances, anemia, etc.); that it does not improve upon using already established dietary pattern analysis in research; that analysis of the NOVA system indicates that it does not predict nutrients that are suggested to be increased by processed foods such as fat and salt; that certain nutritional requirements such as folic acid could not be met with solely unprocessed foods; and that there is little evidence to date that processing relates to measures of satiety or “hyper-palatability”/”food addiction”. Further, changes to modern eating that relate to consuming more energy including increased portion sizes and energy density, may not necessarily be related to food processing. (As such, processing can be used to formulate smaller portion sizes and reduce energy density.)
Similarly, the 2014 American Society for Nutrition scientific statement on the nutritional role of processed foods by Dr. Connie Weaver and colleagues (6) takes issue with using the NOVA classification based on subjectivity, favoring instead IFIC definitions of processed foods as described below:
|Type of Food||Examples|
|Foods that require little processing or production (also called “minimally processed”).||Washed and packaged fruits and vegetables; bagged salads; roasted and ground nuts and coffee beans|
|Foods processed to help preserve and enhance nutrients and freshness of foods at their peak.||Canned tuna, beans and tomatoes; frozen fruits and vegetables; pureed and jarred baby foods|
|Foods that combine ingredients such as sweeteners, spices, oils, flavors, colors, and preservatives to improve safety and taste and/or add visual appeal. (Does not include “ready-to-eat” foods listed below.)||Some packaged foods, such as instant potato mix, rice, cake mix, jarred tomato sauce, spice mixes, dressings and sauces, and gelatin|
|“Ready-to-eat” foods needing minimal or no preparation.||Breakfast cereal, flavored oatmeal, crackers, jams and jellies, nut butters, ice cream, yogurt, garlic bread, granola bars, cookies, fruit chews, rotisserie chicken, luncheon meats, honey-baked ham, cheese spreads, fruit drinks and carbonated beverages|
|Foods packaged to stay fresh and save time||Prepared deli foods and frozen meals, entrées, pot pies and pizzas|
The statement addressed how processed foods contribute to the health and nutrition of populations, stakeholders in improving diet, and research still needed. Analysis of foods that provide nutrient enrichment and fortification (added by processing) indicates that much of the population would fall below adequate intakes for several nutrients. On the other hand, on average they contribute to nutrients that are recommended to be limited such as added sugar, sodium, saturated fat, and calories. They critique the term “ultra-processed” because the degree of processing does not necessarily reflect nutrient content of a food. The paper also summarizes what the future of processed food should look like to better address nutritional and food security around the world, including more cross-discipline collaboration to improve upon processed products and better communication between consumers and relevant stakeholders (6). It also remains to be seen whether the NOVA classification system would lead consumers to choose better diets compared to healthy diet patterns developed by nutrient epidemiology such as MyPlate, DASH, or Mediterranean-style diets, which can include foods with various levels of processing (7).
Thus, depending on how you define it, food processing can contribute necessary nutrients but also nutrients that should be reduced, and how different forms of processing may affect health is understudied.
Research on Processed Foods
Most research on “processed foods” as a whole-diet approach has come from epidemiological studies or animal experiments. Many challenges exist that may prohibit effective research from being done, including defining and classifying processed foods, accurate recalling of foods in dietary surveys, utilizing suitable comparisons, getting study subjects to stay on an assigned diet for the necessary period of time, etc.
Instead of looking at processing, per se, as predictive of health effects, much research currently looks at individual aspects of food processing that may contribute to increased (or decreased) health risks. For example, my dissertation research focuses on dietary phosphorus, which is widely utilized as a food additive (8). Growing evidence suggests that excess phosphorus in the diet, particularly in the form that is added for processing purposes, may increase risks for cardiovascular and bone disease, particularly in those who have kidney disease. For example, a human trial directly compared foods with elevated levels of phosphorus additives vs. those that do not have additives and measured surrogate outcomes for bone and mineral metabolism, and observed changes that would predict long-term bone loss (9). For more, see recent reviews from our group (10,11). It is, however, difficult to definitively tease out the effect of individual components like phosphorus additives to long-term health from the whole food packages that they accompany.
An upcoming trial by Dr. Kevin Hall and others will test the effect of “ultra-processed foods” (based on the NOVA definition) in a highly controlled setting to see what effect on insulin sensitivity and other outcomes such foods have that should prove very interesting and continue the discussion on the utility of such classification schemes (12).
Like any nutrition guidance, we must rely on imperfect evidence to communicate dietary recommendations. Which is why telling the public to choose alternatives to “processed foods” when possible is likely an appropriate message despite scientific disagreement on classifications and health implications. Indeed, the 2015-2020 Dietary Guidelines for Americans contain 25 instances and the scientific report 67 instances of the word “processed” (13,14), in context implying that we should limit processed food, mostly focusing on processed meats where more research has been done. This is because processed foods tend to contribute nutrients that Americans already consume high amounts of, such as sodium and saturated fat. At the same time, it will be interesting to see the results of natural experiments such as Brazil’s Dietary Guidelines, which put the focus on food processing instead of nutrient levels (15). It is an area ripe for research and cross-disciplinary collaborations.