Introduction

The use of the NOVA classification system, developed to study the impact of food processing on diet quality and health outcomes, has shown consistent associations between consumption of ultra-processed food and drink products (UPF) and obesity,¹ metabolic syndrome,²and dyslipidemia.³ More recently, the consumption of UPF was associated with higher nine-year incidence of overweight and obesity and hypertension in a Spanish cohort of middle-aged adult university graduates.⁴ Although the mechanisms by which these foods are linked to weight gain and metabolic impairments still need to be fully determined, some explanations have been put forward.

Several studies show that UPF, as a group, are more fatty, sugary, salty and energy-dense than are all other food groups taken together, in Canada,⁵ the United States,⁶the UK,⁷Brazil,^8,9,10and in Chile.^11,12Additionally, UPF, have a high glycemic load, are often sold in large portion sizes, are formulated to be extremely palatable and habit-forming,¹³ and are aggressively advertised and marketed.¹⁴Furthermore, they stimulate appetite,¹⁵ and are more likely to be eaten mindlessly.¹⁶Current evidence suggests that consumption of sweetened ultra-processed beverages such as soft drinks have less impact on satiety compared to solid food.¹⁷ In contrast, minimally processed food have been found to be more satiating and less hyperglycemic.¹⁸

UPF contribute to more than 50% of the daily energy intake in high-income countries such as the US,6 and Canada,⁵while lower values are found in middle-income countries such as Brazil (21.5%),⁸Mexico (29.8%),¹⁹and Chile (28.6%).¹² Given recent changes in market deregulation in Colombia and the introduction of a Trade Promotion Agreement (TPA) with the United States, greater access to UPF is expected in the near future.²⁰ It is likely that traditional diets and cultural aspects related to eating habits will change, as has been the case in other Latin American countries.^21,22

There have been few prior examinations of the consumption of UPF within a representative Colombian population. In order to address this gap, this study uses the NOVA classification system, to quantify the intake of UPF in the Colombian population. It also assesses the relationship between UPF and the nutrient profile of dietary intake in Colombia in 2005.

Materials and methods

Results

The mean age of this Colombian sample was 26.5 years (±0.2 years). A majority of the sample was female (51.9%), of lower-middle socioeconomic status, with secondary education or higher (46.9%). Over 70% of the participants resided in urban areas (table II).

The average daily energy intake from all food and drinks was 1 835 kcal. Unprocessed or minimally processed food accounted for 63.3% of total energy intake, processed culinary ingredients contributed 15.8%, and processed food and UPF accounted for 4.9 and 15.9% respectively (table I). The contribution of culinary preparations to total energy intake, calculated by combing energy intakes from unprocessed or minimally processed food and processed culinary ingredients was 1 418 kcal or 79.1%. Within the unprocessed or minimally processed food category, the largest contributors to energy intake were plantains, roots and tubers (16.0%), followed by cereal grains and flours (14.2%). Freshly prepared food (preparations of primarily unprocessed food) accounted for 7.1% of the total energy, milk and yogurt contributed 5.5%, and red meat and fruits contributed 5.1 and 3.6% respectively.

Among the processed culinary ingredients, sugars were the largest contributor to total energy (8.9%), followed by vegetable oils (6.1%). In the processed food category, the largest contribution came from cheese (1.9%), followed by fresh bread and bakery products (1.7%). Within UPF, the largest contributor to total energy intake were industrial breads (5.0%), sugar-sweetened beverages (2.5%) and packaged sweet and savory snacks (2.5%). Sausages and reconstituted meat (1.3%), confectionery (1.5%) were some of the other subcategories of importance.

The average Colombian diet met WHO recommendations for protein (12.8% compared to the recommendation of 10-15%), carbohydrates (64.6%, compared to the recommendation of 55-75%), total fats (24.8%, compared to the recommendation of 15-30%), and saturated fats (8.6%, recommended value <10%) (table III). However, the fiber density of the average diet was below recommended levels (11.0 g/1 000 kcal compared to the recommendation of 12.5g/1000 kcal), while average energy density, and free sugars clearly exceeded the recommendations. Average values for energy density were 1.5 kcal/g (recommended range 1.25 g/kcal - 1.45 g/kcal), and 14.1% for free sugars (<10% is recommended).³²

Not surprisingly, the nutrient profile of non-ultra-processed items most closely aligned with recommended intake levels. UPF, however, did not meet recommendations for energy density, fiber density, protein, and free sugars. At 25.3%, the free sugar content of UPF was more than twice the recommended level. Between groups comparisons further highlighted the discrepancies in nutrient levels. The energy density of UPF was twice as much as that of non-UPF (3.1 vs 1.4 kcal/g), fiber density was nearly half (5.4 vs 11.9 g/1 000 kcal) and protein content was also much lower (8.8 vs 13.8%).

Significant differences were also found in the nutrient profile across quintiles of UPF consumption. Compared to the lowest quintile, the quintile with the highest intake UPF had substantially higher daily total energy intake (1 511 kcal vs 2 039 kcal), greater energy density and larger percentage contributions to total energy from free sugars, total fats, and saturated fats (table IV). There was a significant reduction in fiber density and percentage contribution to total energy from protein and carbohydrates from the lowest to the highest quintile. After adjusting for sociodemographic variables, a positive and statistically significant trend was found between quintiles of the dietary share of UPF and the dietary energy density (β: 0.24) and the content of free sugars (β: 0.14), total fats (β: 0.16), and saturated fats (β: 0.22). In contrast, a negative and statistically significant trend was observed for dietary content of carbohydrates (β: -0.19), protein (β: -0.07) and fiber (β: -0.26).

Discussion

In 2005, the Colombian diet was largely based on culinary preparations made of unprocessed or minimally processed food, culinary ingredients, and processed foods. Plantains and tubers and cereal grains were the largest sources of energy in the diet, with some important contributions from red meat and poultry. The overall dietary share of fresh or minimally processed foods in Colombia (63.3%) was much higher than in Chile (33.8%),¹² Canada (39.2%),⁵and the US (29.6%),³⁴and similar to Brazil (58%). However, the nutrient profile of the Colombian diet did not quite align with recommended consumption levels. For instance, while the proportion of energy from protein, carbohydrates, total and saturated fat was adequate, levels of free sugars, and energy density were at higher than recommended levels while fiber was less than recommended.

The Colombian diet composed of only non-UPF had a healthier nutrient profile than the UPF component. It had significantly healthier levels of protein, fiber, and energy density, and while free sugars intake was still higher than recommended levels, it was less than half of the levels seen in the UPF component. These results are similar to those found in a recent study within a representative sample of children (5-12 y) from low- to middle income families in Bogota.³⁵The authors found that processed and UPF had a less healthy nutrition profile compared to minimally processed foods; in particular, the processed and UPF were lower in essential vitamins and higher in sodium and trans-fats.

Industrialized breads, packaged snacks, sugar sweetened beverages and confectionery were the most commonly consumed categories of UPF in Colombia. Sugar sweetened beverages form a large proportion of UPF in other South American countries as well. In Brazil they are the third largest contributor of energy among UPF, preceded only by desserts and fast food,⁸ while in Chile they are the primary contributor of energy.¹² Besides contributing to the overall energy intake, sugar sweetened beverages are also an important source of free sugars in the Colombian diet. Along with confectionary, and the use of table sugar in the preparation of coffee, fruit juices and desserts, they explain the high free sugars level of the Colombian diet. As demonstrated by the results of this study, participants in the first quintile of UPF intake, where calories from these foods represented 0.2% of total intake, still derived 12.4% of their total calories in the form of free sugars. Given the overwhelming evidence on the negative effects of sugar on human metabolic health, recommendations to avoid or limit UPF and to use sugar in moderation when preparing drinks, dishes meals, are essential in Colombia.

Beyond increasing free sugars content, a higher consumption of UPF seems to displace fiber-rich, minimally processed food. In our study, as the dietary share of UPF increased, the fiber density significantly decreased as the energy density, and the proportion of total fats and saturated fats of the diet increased. This macronutrient profile of UPF is also mirrored among diets in Brazil,¹ Canada,⁵ and the US,³⁴ suggesting that a higher dietary share of UPF lowers the overall nutritional diet quality.¹ The observed association between the increasing consumption of UPF and lowered diet quality is highly relevant given the negative effects of high energy density, high intake of free sugars, and saturated fats and low intake of fiber 29 on cardiovascular health, obesity and mortality.

Some limitations of this study should be highlighted. These include measurement errors, and the inability to obtain additional information from the respondents. The authors worked closely with the statisticians who developed the actual surveys and the sampling strategy to minimize any errors in the analysis of derived variables. In addition, misclassification of the food using NOVA was minimized by employing an iterative process with multiple expert consultations. In some cases, it was not possible to disaggregate certain culinary preparations into their constituent ingredients or precisely measure the free sugar content of food items (this was estimated using PAHO recommendations) which may have resulted in some loss of precise information. In addition, some typical culinary preparations which were impossible to disaggregate were mostly classified as unprocessed or minimally processed foods. As a result, the macronutrient content from these foods could have been overestimated since it was not possible for us to disaggregate into groups 1 and 2. The total contribution of this food group (culinary preparations) was 7.1% of the total energy in the diet. Future surveys should be designed to collect information on food processing to allow for the use of NOVA, as suggested by FAO.³⁶Finally, data was used from the only available National Nutrition Survey that allowed for categorization of food intake collected using a 24-hour recall diary. Food intake data from the most recent nutrition surveys in Colombia use food frequency questionnaires, which make using the NOVA classification more challenging.