Introduction
Noncommunicable diseases (NCDs), such as cardiovascular diseases, type 2 diabetes and some cancers, are collectively responsible for almost 70% of all deaths worldwide. The current prevalence of NCDs poses devastating health outcomes and constitutes a serious threat to global health systems. To reduce the number of deaths caused by NCDs, a better understanding of the potential risk factors is needed.
Unhealthy diets are recognized as a major determinant of the occurrence of NCDs. With the increasing trend of NCDs, a steady rise in the share of processing foods has been seen. In the last half century food processing has evolved greatly as a consequence of the industrialization and globalization of food systems [
1]. Negative effects on nutritional dietary quality emerged subsequently, such as higher content in free sugars, saturated fats, energy density and sodium, and less content in protein, fiber and micronutrients. It is believed that most NCDs can be prevented by changes in diet patterns.
Ultra-processed foods (UPFs) are defined as formulations of ingredients derived from foods and additives, coupled with substances including colorings, flavorings, sweeteners, and emulsifiers [
2]. They contain little if any intact food. Included in this definition are sugar-sweetened beverages, sweets, ice cream, chocolates, savoury snacks, burgers, processed meat and frozen dishes. Compared with other food groups, UPFs are typically durable, ready to consume, low-cost and hyper-palatable. They tend to be packaged delicately and marketed concentratedly. They are characteristically fatty, sugary or salty, energy-dense and lack of protein, dietary fibre, micronutrients and several bioactive compounds [
3‐
5]. Furthermore, they may contain neo-formed contaminants derived from industrial processing, as well as substances from additives and packaging [
6,
7]. Considering the association between UPFs and poorer dietary quality, the share of UPFs has been proposed as an effective predictor of population diet quality [
8‐
10].
The whole world has witnessed a dramatic transition in food consumption patterns. Unprocessed or minimally processed foods and freshly prepared meals are gradually displaced by UPFs. The shift appeared initially in high and middle income countries, and then worldwide [
11,
12]. Transnational corporations are major factors that drive the production and sales of UPFs, along with their convenience, branding and aggressive marketing [
13]. These characteristics create massive market advantages for UPFs over other food groups [
14]. In high income countries, more than half of the foods consumed are UPFs for most of the age groups, and consumption decreases with age [
15,
16]. Purchase surveys and dietary trends on UPFs consumption have been performed in Asia and many western countries [
17‐
21]. It has been evaluated that the energy contribution of UPFs ranged from 25 to 60% [
22].
The existing evidence indicates that displacement by UPFs is driving a rising prevalence of obesity and other diet-related NCDs [
23]. A growing body of evidence suggests that increases of UPFs in dietary proportion were associated with a higher incidence of adverse health outcomes [
16,
24‐
42]. Decreasing the dietary share of UPFs may notably contribute to the prevention of diet-related NCDs [
43‐
45].
As UPFs are increasing dominantly during the past decades, understanding their potential impacts on health outcomes has become a major imperative. To date, however, this literature has not been comprehensively evaluated. No reviews have been conducted on this topic previously. To address these concerns, this systematic review was conducted to summarize the evidence for the association between UPFs consumption and health outcomes.
Discussion
To the best of our knowledge, this is the first systematic review of available epidemiological evidence on the association between UPFs consumption and health outcomes. We identified 12 cohort and 8 cross-sectional studies, and found that there was a positive association between UPFs consumption and risk of all-cause mortality, overall cardiovascular diseases, coronary heart diseases, cerebrovascular diseases, hypertension, metabolic syndrome, overweight and obesity, depression, irritable bowel syndrome, overall cancer, postmenopausal breast cancer, gestational obesity, adolescent asthma and wheezing, and frailty. It showed no obvious association with cardiovascular disease mortality, prostate and colorectal cancer, gestational diabetes mellitus and gestational overweight.
Among the included studies, different methods were applied to estimate intake of UPFs. In some studies portion sizes were estimated using validated photographs. They calculated energy and weight relative to total food intake according to specific food composition databases. Some studies calculated daily intake by multiplying the portion size by the frequency of consumption, which had been proved validated. A majority of included studies evaluated intake as the percentage to total energy while the others selected weight proportion. The application of energy proportion contributed to reduction of variation due to body size, metabolic efficiency, and physical activity [
33]. Weight proportion was taken into account for UPFs that did not provide any energy intake such as artificially sweetened drinks, and factors related to food processing [
38]. Although methods vary, UPFs intake was all modeled as quantiles (e.g., tertiles, quartiles and quintiles). We selected effect estimates adjusted by the most factors for the highest versus the lowest consumption levels, which made them comparable. In light of the increasing concern of UPFs, further research is needed to set standard methods for UPFs intake estimation.
Despite different ways to estimate intake of UPFs, all the studies conducted food classification using the NOVA system, except one [
33]. Louzada et al. divided foods into three groups according to the degree of processing, which was also consistent with the NOVA system. NOVA, a food classification system which classifies foods into four groups according to the nature, extent and purpose of industrial processing, has now been applied globally [
49]. Groups are as follows: (i) Unprocessed or minimally processed foods; (ii) Processed culinary ingredients; (iii) Processed foods; (iv) Ultra-processed foods. Instead of focusing on nutrient composition of the diet, it takes into consideration all physical, chemical and biological methods used during the food productive process [
50]. The NOVA system has been used to describe population dietary patterns, assess changes in the dietary share of UPFs, and analyze the association of the dietary share with nutrient profile and with health outcomes [
50]. It will contribute to the prevention of NCDs and the improvement of public health worldwide [
51]. Nevertheless, when using the NOVA system to make dietary recommendations, it should be considered that some foods are difficult to classify [
2]. Further research is needed to promote the application of this food classification system.
Considering the synergistic health-related effects of foods, it is of great importance to study dietary patterns instead of single foods or nutrients. UPFs consumption is increasing dominantly across the globe, especially in Western countries. It is consistent with the increased burden of NCDs attributable to unhealthy diets. The role of some specific UPFs has been assessed, such as processed meats and sweetened beverages, showing positive associations with NCDs [
52,
53]. In line with our findings, previous studies reported an inverse association between higher diet quality and risk of all-cause, cardiovascular disease, and cancer mortality [
54‐
56]. Adherence to healthier diet patterns, which are characterized by a high consumption of unprocessed or minimally processed foods, was promoted to prevent NCDs. Decrease of cardiovascular disease burden with a healthier food system was observed in two modelling studies [
57,
58]. A review of systematic reviews found that grain products and tea were protective, while processed meats and soft drinks tended to increase the risk [
59]. Existing meta-analyses demonstrated that an optimal intake of several food groups could decrease the risk of coronary heart disease, stroke and heart failure [
60]. Overall, UPFs consumption should be limited in prevention of NCDs.
The mechanism is multi-faceted. First, UPFs consumption is usually accompanied with high intake of fats, calories, sugars and salt, and low intake of micronutrients and fibre. The poor quality of dietary nutrients leads to the development of NCDs [
61]. Processing, especially heat treatments, food additives and food packaging, can generate carcinogenicity and genotoxicity [
25]. UPFs consumption increases added sugar intake, which is associated with obesity and several other health outcomes [
19,
21,
62,
63]. Moreover, higher intake of UPFs induces changes in gut microbiota, serum C-reactive protein levels and lipoprotein profiles [
64‐
67]. Displacement of unprocessed or minimally processed foods might play a potential role in decreased diet quality [
37]. However, it still remains unclear what plays a leading role in the association. A better understanding of what really matters and how various aspects contribute to the effects is highly needed.
Our study has strengths. To our knowledge, this is the most comprehensive systematic review of the topic to date. We carried out extensive literature research. The occurrence of selection bias was reduced greatly due to the prospective design of all the cohort studies. Large number of participants might compensate for the inadequate number of studies of each health outcome. All the eligible studies made adequate adjustments of the potential confounding factors. In general, we provide strong implications for dietary policies and guidelines.
Several limitations should also be acknowledged. First, most cohort studies recruited university graduates or volunteers as study objects, who tended to be more health-conscious and had a lower UPFs consumption than the general population. This probably resulted in an underestimation of the association of interest. Second, as occurrence of some health outcomes took a long time, such as carcinogenic processes, the median follow-up was relatively inadequate. Besides, complete detection of outcomes could not be guaranteed. Third, epidemiological studies could not exclude reverse causality and residual confoundings. For cross-sectional studies, probability existed that participants changed their dietary habits after the occurrence of diseases. It tended to cause an underestimation of the results. Fourth, some misclassification in the NOVA system could not be ruled out. Ways applied for dietary assessment were not specifically designed for NOVA classification and UPFs consumption yet. However, substantial differences between the highest and the lowest group might reduce the bias to a great extent. It is also worth mentioning that no quantitative meta-analysis was conducted, which we hoped to be overcome with further research. Given ethical issues of conducting randomized controlled trials of risk factors, more well-designed epidemiological studies are needed to confirm these findings.
There are significant public health implications in our study. To date, prevention and control of NCDs are becoming a growing concern. UPFs are beginning to be recognized as an emerging health risk. The positive association between UPFs consumption and adverse health outcomes provides insights into dietary policies and guidelines. Encouraging a decrease in UPFs consumption and an increase in the proportion of unprocessed or minimally processed foods are a direct way to resolve the issue. Food taxation and surveillance on food marketing still play a vital role. Dietary guidelines, in accordance with the shift of the global food system and health, are a necessity to slow the prevalence of NCDs.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.