Energy and protein intake in the Colombian population: results of the 2015 ENSIN population survey

The present study was aimed at (1) the differences between current weight v. ideal weight, (2) total energy intake and comparing it with required energy (Rkeer), (3) absolute protein intake in g/kg per d and g/1000 calories, (4) how energy and protein intake relate to the nutritional status of the subjects in terms of overall overweight (OEW) [overweight + obesity] and conservative overweight (CEW) [obesity] and (5) the contribution (%) of protein to total energy intake based on the acceptable macronutrient distribution range (AMDR). A dietary study was carried out in Colombia with 29 259 subjects between 1 and 64 years of age, based on cross-sectional data collected in 2015 by a 24-h dietary recall (24HR) administered as part of the National Nutrition Survey. Energy and protein intake did not differ by nutritional status. In the general population, energy intake was 2117 kcal/d (95 % CI 1969, 2264). The total protein intake was 64⋅3 g/d (95 % CI 61⋅4, 67⋅3). Adequate energy intake ranged from 90 to 100 %, except for the 1–4-year-old group, which ranged from 144 to 155 %. Protein intake was 1⋅64 g/kg per d (95 % CI 1⋅53, 1⋅75). The mean AMDR for protein to total energy intake was 13⋅3 % (95 % CI 12⋅9, 13⋅7). Excess weight began during the first 4 years of age. In conclusion, it is worth reviewing and updating energy and protein intake recommendations and dietary guidelines for the Colombian population and designing and modifying public policy.


Introduction
In Colombia, South America, demographic, economic and epidemiological transitions have been accompanied by two other simultaneous transitions (1) : (1) nutritional, where overweight is more prevalent than obesity (2) , and (2) alimentary (3) , where the traditional dietary pattern may be a protective factor against developing overweight and obesity, which are mediators of cardiovascular disease and cancer (3)(4)(5) .
Given the limited or non-existent scientific evidence at the local level in Colombia, public policies on food insecurity, nutritional deficiencies and nutritional and dietary status have followed approaches used in other cultural and socioeconomic contexts. In the general population, individual dietary practices stem from a combination of various aspects that involve the context in which people live, such as personal or individual factors, as well as external factors in the food environment, including physical, economic, sociocultural and political (6) . The explanations of these practices have lacked epidemiological thought, for example, multicausality.
Food and nutrition policies impact the environment, the economy and other public policies; one example is a recommendation to update the Dietary Guidelines for Americans (9) . Both the amount and quality of protein consumption have been of interest due to their association with serum iron levels, cholesterol, saturated fat, the development of chronic illness, and more recently, with the carbon footprint. The impact that diet has on different environments (economic, social, cultural, industrial and technological, among others) is a complex issue and a new field of formal research. The United States Department of Agriculture proposes six categories that have an environmental impact and political importance: (1) global warming potential, (2) land use, (3) water depletion, (4) marine eutrophication, (5) availability of potable water and (6) respiratory particulate or organic material (9) .
In Colombia, the 2005 ENSIN (13) and the Energy and Nutrients Intake Recommendations (RIEN in Spanish) (14) indicate a deficiency of 36-50⋅1 % in usual protein intake. Colombia updated the RIEN in 2016 based on the epidemiological and nutritional profile of the Colombia population. This guideline sets an acceptable macronutrient distribution range (AMDR) of 10-20 % for protein for the different population groups and suggests a minimum of 15 % in order to meet the average estimated micronutrient requirement (14) .
The present study was aimed at determining (1) the differences between current weight v. ideal weight, (2) total energy intake and comparing it with required energy (Rkeer), (3) absolute protein intake in g/kg per d and g/1000 calories, (4) how energy and protein intake relate to the nutritional status of the subjects in terms of overall overweight (OEW) [overweight + obesity] and conservative overweight (CEW) [obesity] and (5) the contribution (%) of protein to total energy intake based on the AMDR.

Materials and methods
Colombia (South America) is a middle-income country with large socio-economic inequalities. Three national nutrition surveys (ENSIN in Spanish [Encuestas Nacionales de la Situación Nutricional]) were conducted between 2005 and 2015 using a cross-sectional design. The analysis was based on data on energy (kcal/d) and protein (g/d) intake obtained from a 24HR conducted as part of the 2015 ENSIN, which was a nationally representative survey with a complex and multistage sample design. The methodological details have previously been published (15) .

Population and sample
The 2015 ENSIN surveyed 44 202 urban and rural households and interviewed 151 343 subjects. The sample included 4739 segments located in 295 municipalities in the country's 32 departments and in Bogota, Colombia's capital. Various probabilistic subsamples were calculated to study specific topics, including nutrient intake, nutritional practices of interest and breast-feeding, among others. The 2015 ENSIN sample was based on the Master Household Sample for Health Studies, of the National System of Health Population Studies and Surveys, developed and implemented in 2013 by the Ministry of Health and Social Protection (7). Subjects were selected with various sampling methods, including probability, cluster, stratified and multistage. The primary sampling unit was the department, the secondary was the municipalities and the tertiary was the housing of the subjects. The subsample that constituted the 24HR included 34 099 subjects between 0 and 64 years old, 4589 of whom filled out a second 24HR. Excluded from the 34 099 were children between 0 and 1 year of age (n 1770), pregnant women (n 2589) and those without data on weight, height or both (n 481). The 24HR response rate was 84 %. The final sample that was analysed included 29 259 subjects.

Source of data
Trained personnel administered the questionnaire to heads of household to obtain information on sociodemographics, food security and household wealth. Moreover, nutritionists administered the 24HR using the Automated Multiple-Pass Method (AMPM) developed in 1999 by the United States Department of Agriculture (USDA) (15,16) . For children under 12 years old, consumption information was supplied by the person who was responsible for having prepared and served their food the previous day and/or who accompanied the child while eating. For adolescents between 12 and 14 years old and adults 60 years or older, a rapid memory test was conducted before administering the 24HR, which included 4 of the 10 items proposed by Hodkinson in 1972 to identify dementia in elderly patients (17) . When one of the four questions was not adequately answered by the subject, the information was provided by the caregiver or the person responsible for feeding the child. Anthropometric measurements were taken by trained interviewers using standardised techniques and calibrated equipment. Size was determined with stadiometers (Shorr Productions LCC, Olney, MD, USA) to the nearest millimetre. SECA scales (model 874) were used to determine weight to the nearest 100 g.
The main outcome variables were OEW, CEW and energy (kcal/d) and protein (g/d) intake. Eleven covariables were also studied: sex, age, weight circumference, adherence to weekly physical activity goals, number of household members, household food security, wealth index, educational level of the head of household, ethnicity, degree of urbanism and geographic region.

Nutritional status
Values for weight, size and sex were converted to Z scores based on growth references by the World Health Organization (WHO) (17,18) . For children up to 17 years old, OEW was based on body mass index for age (BMI/A) and Z-score >1, and CEW was based on Z-score >2. For adults, overweight was defined as ≥25 BMI <30 (kg/mt 2 ), obesity as BMI ≥30, OEW = overweight + obesity and CEW = obesity. In addition, for each individual with low or excess weight, the 2015 ENSIN estimated 'adequate weight' based on its own data using two predictive equations (linear regressions): one for subjects between 1 and 17 years old and another for those between 18 and 64 years old. The equations included data on the sex, weight, size and age of the individuals with normal nutritional status for the age group (19) .

Energy and protein intake
After rigorous quality control, the 24HR were converted to nutrients based on a food composition database specifically designed for the 2015 ENSIN. This database contains 2703 items and 9 nutritional variables. In addition, following recommendations by the 2001 FAO/WHO/UNU Expert Consultation (20) , the Estimated Energy Requirement (EER) was calculated for each individual based on RIEN (14) . Energy (kcal/d) and protein (g/d) intake were reported in absolute terms based on the first 24HR. Protein intake was reported as nutrient density in grams per 1000 calories (g/ 1000) and as the ratio of absolute protein intake/day (g/d) to adequate weight (Rprotein). Lastly, the ratio of absolute kcal intake/day to EER was calculated and expressed as requirement units (Rkeer) or adequacy (%EER).

Adherence to weekly physical activity goals
The 2015 ENSIN calculated this potential confounder based on minimum adherence per week, using different methods for different age groups (15) . For children under 6 years old, adherence was defined as 180 min or more of play per day over the prior 7 d, according to C-MAFYCS (21,22) . For adolescents between 6 and 17 years old, adherence to physical activity recommendations was defined as 60 or more min/d of moderate or vigorous physical activity, according to the Youth Risk Behavior Surveillance System (23) . Adherence for adults was defined as at least 150 min of moderate or 75 min of vigorous aerobic physical activity/week, according to the International Physical Activity Questionnaire (IPAQ) developed by the WHO (24) .
Large waist circumference was defined as ≥90 cm for men and ≥80 cm for women. The degree of urbanism was categorised as urban and rural based on the population density reported by the 2015 ENSIN (15) . Household food security was determined with the Latin American and Caribbean Food Security Scale (ELCSA in Spanish) (25) . Wealth level was designated based on the index designed for the international demographics and health survey (26) , with the highest values representing the wealthiest subjects.

Statistical analysis
All analyses were performed using the routines for complex sampling designs in Stata software version 14.1 (27) . The statistical analysis was aimed at (1) describing socio-demographic characteristics, (2) for each category of covariates, describing the energy and protein intake of the subjects with prevalences (%) or averages, and presenting each one with its standard error (SE) or 95 % confidence interval (95 % CI), (3) determining the prevalence ratio of protein intake to adequate weight (g/kg per d) in subjects with and without excess weight and (4) determining the average prevalence ratio of energy intake to energy required in subjects with and without excess weight.
Multiple linear regressions were performed to obtain adjusted differences in protein and energy intake between subjects with and without excess weight, and with and without conservative excess weight, with their different expressions for each category of covariates. To this end, a new term was created with the result of the cross product of intake and each category of covariates (interaction). The adjusted differences and their respective 95 % CIs incorporated the complex sample design, and the multiple regression model included the covariates sex, age, adherence to physical activity, household size, household food insecurity, wealth index, ethnicity, education level of head of household, geographic area and region.

Ethics approval and consent to participate
All analyses were performed in accordance with the principles of the Helsinki Declaration (28) . The databases used are available to the public. This research is classified as 'without risk' according to Resolution 8430 of the Colombian Ministry of Health (1993) (29) . Since this is a secondary analysis of population studies with anonymised data, no authorisation was required from the Health Research Ethics Committee of the Industrial University of Santander.

Results
A total of 29 259 subjects were studied, 48⋅8 % of which were men. The OEW of the population was 28⋅7 %: 27⋅6 % for men and 30⋅0 % for women (P = 0⋅123). Nine percent (9⋅0 %) of the population had CEW: 8⋅7 % of men and 9⋅2 % of women (P = 0⋅506). A total of 25⋅8 % of subjects met the physical activity recommendations, 37⋅5 % of households were food secure and 77⋅3 % of subjects lived in urban regions. Table 1 presents the remaining socio-demographic characteristics of the subjects studied, according to excess weight categories.

Discussion
This work found that excess weight in the Colombian population is detectable by the age of 4 years, which may suggest that the excessive energy and protein consumption described herein may be the causal pathway to excess weight and obesity at an early age. Once excess weight begins, it increases steadily until age 40, and although it decreases after age 40, there continues to be a gap between adequate and current weight (Fig. 1). With regard to current mean energy and protein intake, no significant differences were found between subjects with and without OEW and CEW, with the exception of ethnicity. In general, the Rkeer was found to fall within the suggested range of 90-110 %, except for children between 1 and 4 years old (144 and 155 %). Given the low level of physical activity on the part of the Colombian population (7) , if a range of 0⋅8-1⋅2 g/kg per d is considered adequate protein consumption regardless of the nutritional condition of the subjects (30) , then the population consumes more protein than the limit. Nevertheless, long-term protein consumption of up to 2⋅0 g/kg per d has been considered to be safe (30) , and protein intake has been considered inadequate when the relative protein contribution is under the RIEN (14) recommendation of 15 %. The prevalence of overweight and obesity among Colombian adults was greater for women than for men: 8 and 6⋅8 %, respectively (7) . Since overweight, obesity, OEW and CEW cannot be attributed to the prior day's energy or macronutrient intake, as expected, absolute and relative energy and protein intake did not differ according to the nutritional condition of these subjects. Excess weight is a long-term condition that is related only to usual intake and not to current intake based on a single 24HR (31) .
Proteins are one of the most commonly studied macronutrients. They are required for growth and development, and for the synthesis of essential amino acids (30) . The U.S. National Academy of Medicine (formerly the Institute of Medicine [IOM]) established the estimated average requirement (EAR) and recommended dietary requirement (RDA) in grams per kilograms per day (g/kg per d) (EAR of 0⋅66 g/kg per d and RDA of 0⋅80 g/kg per d) (32) , while RIEN established the relative contribution of protein to total energy intake, with a relative contribution between 10 and 20 % (equivalent to 45 g/d for children between 2 and 5 years of age and up to 103 g/d for subjects between 14 and 17 years old) (14) . In all cases studied herein, the estimates exceeded both IOM and RIEN recommendations and were similar to those reported by the 2003-2004 National Health and Nutrition Examination Survey (NHANES): average of 56 g/d ± 14 SD for children and adolescents, 91 ± 22 g/d for subjects between 19 and 30 years of age, 86 g/d ± 20 SD for adults for adults between 30 and 50 years and 66 ± 17 g/d for adults over 50 years old (33) . NHANES reported that between 1999 and 2016, the AMDR for total proteins (both animal and vegetable) increased from 15⋅5 to 16⋅4 % in the U.S. population over 20 years of age (12) . Furthermore, protein intake in 2003-6 did not differ significantly from protein intake in 2015-16 (34) .
Excess protein intake has been associated with increased mortality from cardiovascular disease, with a hazard ratio of 1⋅08 per 10 % increase in energy consumption (95 % CI 1⋅01, 1⋅16; P = 0⋅04 for the trend). This was found among those who had at least one risk factor associated with smoking, heavy alcohol consumption, overweight, obesity or sedentarism (35,36) . In a cohort study of Swedish women between 30 and 49 years old with a 16-year follow-up, regular consumption of high-protein, low-carbohydrate diets was associated with a greater risk of cardiovascular disease, with an incidence rate of 1⋅05 (95 % CI 1⋅02, 1⋅08) (37) . A case-control study of the Korean population between 30 and 76 years old reported an inverse association, though not statistically significant, between vegetable protein intake and the prevalence of colorectal adenoma, after adjusting for age, total energy intake, waist circumference, BMI, HDL-cholesterol, fasting glucose, alcohol intake and smoking. Nevertheless, when adjusted by potential confounders, that association was not significant (OR 0⋅54 [95 % CI 0⋅27, 1⋅11]; P = 0⋅13 for the trend) (38) . With regard to the risk of type 2 diabetes mellitus, a cohort study of the population in China between 20 and 74 years old found that, over the long term, diets low in carbohydrates, high in fats and high in proteins were associated with a higher risk of type 2 diabetes for subjects who consumed excess calories (RR 1⋅64 [95 % CI 1⋅03, 2⋅61]; P = 0⋅040 for the trend) (39) . A cohort study with an Iranian population over 20 years old found that the risk of incidence of chronic kidney disease increased with an increase in high-protein, lowcarbohydrate diets (OR 1⋅48 [95 % CI 1⋅03, 2⋅15]; P = 0⋅027 for the trend) (40) . Excess protein consumption of over 0⋅8 g/kg per d increased intraglomerular pressure and complications from chronic kidney disease (41) . Lastly, the EAT-Lancet Commission has presented robust evidence of the health and environmental consequences from consuming over 0⋅8 g/kg per d of protein, or ≥10 % of the AMDR, predominantly from animal sources. The commission reported health consequences such as an association with overall mortality from cancer, cardiovascular disease and type 2 diabetes, and an environmental impact on the production of greenhouse gases (methane, nitrous oxide and carbon dioxide) from food production (42) .

Public policy implications
The development of the science of nutrition cannot be understood without the institutions and policies that are associated with them (43) . In this case, energy and protein intake is not Table 2. Adjusted difference in the ratio between protein intake and adequate weight (g/kg per d), between overall excess weight a and non-overall excess weight b for subjects in the Colombian population, 2015 an isolated event that occurs outside the socio-economic and cultural context of the subjects. Furthermore, public policy can provide proper guidance on multiple levels (43) . The evidence presented herein, based on data on protein and energy intake from the 2015 ENSIN, will help institutions that are responsible for improving diets and nutrition (14) to not only review and modify nutritional goals but also sustain them, as suggested by the Dietary Guidelines for Americans (9) . Incorporating sustainability in a possible update of the RIEN and the Food-based Dietary Guidelines (GABAS in In children and adolescents based on Z ≤ +2, in adults <30 (kg/m 2 ). * Conservative overweight n may be less than 2631 for missing values. Non-conservative overweight n may be less than 26 628 for missing values. † Adjusted difference and 95 % confidence intervals calculated with a linear regression model based on consumption/day of kilocalories as a dependent variable and predictors that include indicator variables for each socio-demographic correlate, Non-conservative overweight (Conservative overweight) and cross-product (interaction) terms between obese and indicator variables of the correlate. In addition, the linear regression model was adjusted by the following covariables: sex, age, physical activity, household size, food security, wealth index, ethnicity, education of the head, area and region. The complex sampling survey design was used in all multivariate regression models. ‡ In men ≥ 90 cm, in women ≥ 80 cm. § The rural category included suburban population centres close to small cities, towns in rural areas distant from small towns and places dispersed or very distant from rural towns.
Spanish) (44) may also improve food security in the country (9,45) . Future modifications could include conscious eating as promoted by Canadian dietary guidelines (46) and standardising recommendations in g/kg per d (33) . Protein intake limits and the sources of protein can also be adapted, given that red meat and eggs can play a protective role in iron deficiency anemia, thereby reducing its prevalence in Colombia (47) . By having identified the age at which the current and adequate weight of the subjects began to deviate from each other, dietary and nutritional education as well as other multi-sector dietary interventions can be better targeted before excess weight develops. Various actions need to be taken to prevent pseudoscience from replacing food and nutrition education, such as compelling messages and actions regarding how long a diet needs to be followed before harm or benefits result, the complexity of interactions among different levels of risk factors and differences in individual and population interventions carried out in particular cultural contexts. Lastly, the data on the prevalence of protein intake deficiency reported in the 2005 ENSIN should be reviewed to take into account habitual or long-term consumption (13) .

Strengths and limitations of the study
The primary limitation of the present study was the inability to determine causal associations due to the cross-sectional design of the ENSIN. Another limitation was that the present study reported current intake rather than usual or long-term intake, given the use of the 24HR method. Sources of protein could also not be determined (animal or vegetable). Nevertheless, this analysis also has its strengths. The data were taken from a nationally representative survey that estimated energy and protein intake based on a 24HR that was administered by welltrained nutritionists. Furthermore, the 24HR was translated into nutrients based on the best food composition database available in Colombia.

Supplementary material
The supplementary material for this article can be found at https://doi.org/10.1017/jns.2021.2.