Dietary intakes and their association with weight status in male and female school children in South India: A cross-sectional study

Background With The present analysis was done to cross-sectionally examine the diets of school-aged children by sex and weight status. Urban school-going children 8-15 years of age, representing all strata of socio-economic status were recruited through stratified quota sampling. Dietary intake was assessed by three interviewer-administered 24-hour dietary recalls. The proportion with inadequate intakes for macro- and micronutrients and percentage energy intake of macronutrients below and above the Acceptable Macronutrient Distribution Range (AMDR) were compared between sexes. Association of dietary intake with weight status were examined by multinomial logistic regression in boys and girls separately with the reference category being the highest tertile of each food group intake.


Conclusions
Children's diets need to be improved to meet nutrient requirements. Promotion of healthy diets in children which reduce prevalence of underweight and arrest increase in overweight is required.

Background
Globalization and urbanization have influenced diets of adults and children in terms of food options and choices in both developing and developed countries (1)(2)(3). Over the past few decades, the diets of children have changed rapidly and this has been attributed to the ongoing nutrition transition (4)(5)(6). Present day diets have led to an increased consumption of higher fat and sodium and less of fibre and essential micronutrient intakes mainly through consumption of processed foods, resulting in the triple burden of malnutrition (undernutrition, overnutrition and micronutrient deficiencies) (3). Diets of children in developing countries including India, continue to be of poor dietary quality with poor dietary diversity (7,8). A study conducted across 9 to 11-year-old children across 12 countries in the world including India, showed an association of lower income or levels of parental education with higher consumption of unhealthy foods and lower consumption of healthy foods patterns irrespective of whether children were from countries with high or low development; this was attributed to easy availability because of low costs of unhealthy foods (9). Childhood is a critical period when dietary habits formed have a bearing on later adult health outcomes particularly of chronic diseases (8).
Considering these factors, it is essential to evaluate dietary intakes of children to understand what they currently eat so that diets can be improved to meet nutrient requirements. Evaluating and ensuring adequate intakes of both macro-and micro-nutrients is important as it is possible that the quality of intake is not adequate to meet both the macro-and micro-nutrient requirements.
By assessing nutrient intakes the prevalence of inadequate intakes of nutrients for groups or populations can be determined (10). Such monitoring is essential in a country like India where deficiencies of several macro-and micronutrient deficiencies exist and are attributed largely to dietary insufficiencies. The type of diet eaten is also influenced by decreased bio-availability of some micronutrients due to inherent inhibitors in the foods eaten as for example, phytates (11). Evidence from repeated national intake surveys conducted by the National Nutrition Monitoring Bureau largely in the rural areas, show highest deficits in intake for folic acid followed by iron and riboflavin and then thiamine and niacin with deficits increasing with increase in age in children (12). Such data is however, lacking in urban children in India.
The period of childhood and adolescence is critical in growth and development. With India facing a triple burden of malnutrition with under-nutrition, over-nutrition and micronutrient deficiencies observed, it becomes relevant to assess dietary intakes of this population. Children 6 to 18 years of age constitute about 27.5% of the total population of India and 23.9% of population of the state of Karnataka as per the latest census of India (13). The effects of nutrition transition are likely to be observed in diets of urban children where availability, accessibility and affordability to a variety of foods is presumably higher. This has so far not been studied in detail in India. The objective of this analysis was to examine dietary intakes of urban male and female school children including the level of inadequacy and to compare dietary intakes between sexes based on their weight status.

Study population and sample size
In a cross-sectional study, children 8 to 15 years were recruited from schools in urban Bangalore in the state of Karnataka, India from January to October 2015. A stratified quota sampling design based on socio-economic level of children attending schools and their annual fee (low, medium and high fees) was employed. Thus, a total of 7 schools were identified. The annual fees of each school were obtained to classify the socio-economic status (SES) as catering to students from low, middle and high SES groups.
With hitherto unpublished data in school children on dietary intake, micronutrient inadequacy for iron, folate, thiamine, riboflavin, niacin and vitamin B 6 and B 12 ranged between 30 and 56%. Hence assuming an inadequacy level of intake of micronutrients of 45%, the sample size required to estimate at a precision of 15% and a confidence interval of 95% was 210 per strata (total of 630 children). In total to achieve the proposed sample size 3 low SES, 2 middle SES and 2 high SES schools were contacted and all schools permitted their children to participate in the study.
Ethical approval was obtained from the Institutional Ethics Committee of the St John's Medical College, Bangalore. The management of the school or the principals were contacted for permissions to conduct the study. Written informed consent was from parents and oral assent from children were obtained. A target of 90 children at each age from 8 to 15 years was proposed to be achieved to attain the required sample size of 630 children by recruiting based on date of consent received. A total of 2065 consent forms were distributed in the 7 schools, out of which 837 were willing to participate in the study and 634 were finally recruited. In total, 203 children were not recruited, 164 of them due to the reason that the target of 90 children were achieved for the particular age (as for example 8 to 9 years), 21 due to incomplete consent forms, 2 were above the age of 15 years, 4 below the age of 8 years and 12 were absent during the period of study in the school.

Demography and dietary intake
Demographic data on age and sex of the child, education and occupation of both parents were collected. Based on age, children were categorized as below 11 years of age and ≥11 years of age, as age 11 marks the onset of adolescence. Education of parents was categorized as up to primary which included those uneducated and those who had availed up to grade 5 of education, secondary which included a further 7 years of education and university level.
For each child, dietary intake was assessed by three 24-hour dietary recalls as usual intakes can be calculated. Further, Burrows et al (14) indicated that 24-hour recalls collected over 3 days gave the best estimate of energy intake in children 4 to 11 years of age. Two non-consecutive weekday recalls, and one weekend/holiday recall were taken by trained personnel from the child using an interviewer administered questionnaire. The 24-hour recall was administered individually to each child by trained interviewers with the subject first listing the foods and beverages consumed during the preceding day, including vitamin and mineral supplements, followed by the reporting of portion size using food portion size aids and then probing for forgotten foods after which the whole recall was reviewed with the subject.
The data collected was entered into a software created using the food composition database developed specifically for this purpose (15) and macronutrients (energy, protein, fat, carbohydrates) and micronutrients (iron, thiamine, riboflavin, niacin, vitamin B 6 , vitamin B 12 , folate and vitamin A) values for each of the 3 recalls were obtained. To estimate usual intake distributions and obtain the proportion at risk of inadequate intake of protein, iron, thiamine, riboflavin, niacin, vitamin B 6 , vitamin B 12 , folate and vitamin A, the within-and between-person variability was determined. The day-to-day variability in intakes was partially removed by fitting a measurement error model to daily log transformed intake data to minimize bias in estimates of prevalence of inadequacy using the Estimated Average Requirement (EAR) (Supplementary table S1). All statistical adjustments were done for the distribution of intake of each individual in the group as per the National Research Council's approach proposed by Institute of Medicine (IOM) (16).
The proportion of children with inadequate intake of the nutrients was determined based on age and sex specific Estimated average requirement (EAR) as per recommendations by IOM (16,17) for protein, thiamine, riboflavin, niacin, vitamin B 6 , vitamin B 12 , folate and vitamin A using EAR cut point method (16,18). To identify children at risk of inadequate intake for iron, the estimated average requirement recently provided in a publication on iron requirements in Indian children was used (19).
The risk of inadequate intake of iron was computed by the probability method which applies a continuous risk-probability function to each individual's estimated intake and then averages the individual probabilities across the population or group.
Further, macronutrient intake were categorised based on acceptable macronutrient distribution range (AMDR) as defined by the Institute of Medicine for children 4 to 18 years of age (16). Carbohydrate intakes 45 to 65% of energy, protein intake between 10 to 30% of energy and fat between 25 and 35% of energy were considered as adequate intake; values below 45%, 10% and 25% of energy for carbohydrate, protein and fat respectively, were considered as "below AMDR", values above 65%, 30% and 35% of energy for carbohydrate, protein and fat were termed as "above AMDR". Cereals, pulses, root vegetables, other vegetables, fruits, milk (milk and milk products) nuts, meat (red and white meat, fish and egg), added fats and added sugar were the food groups considered for analysis.
The nutrient intake during breakfast and the use of fortified foods which included commercially fortified foods and beverages and processed foods were examined and used for analysis. The nutrient intake consumed during breakfast was compared to the whole day's intake.

Anthropometric measurements
Height was measured to the nearest 0.1 cm using a stadiometer (SECA 213) and weight to the nearest 0.1 kg using a digital weighing scale (SECA 803). Body mass index (BMI) was computed and both the height for-age z scores and the BMI-for-age Z score values were obtained using the World Health Organization Anthroplus software version 1.0.2 (WHO, Geneva, Switzerland) using the reference data for school aged children (20). Children were then categorized by actual weight status as overweight (>+1 SD), normal (-2 to +1 SD) and underweight (<-2 SD). The categorized weight status was used for analysis.

Statistical Methods
All categorical variables are presented as number and percentage. Continuous data were examined for normal distribution using Q-Q plots. The normally distributed continuous variables were reported as mean and standard deviation (SD) and the non-normal variables using median and quartiles.
The various sociodemographic variables such as age group (<11 and ≥11 years), SES (low, middle and high), maternal and paternal education, were compared between sexes using the Pearson's chisquare test or Fischer's exact test as appropriate. Anthropometric measures were compared using independent samples t test or Mann Whitney U test, as appropriate. The macronutrient intakes of energy, protein, fat and carbohydrate and the micronutrient intakes of iron, thiamine, riboflavin, niacin, vitamins B 12 , B 6 , folate and vitamin A in all children and stratified by sex were compared using the Mann Whitney U test. Percent protein, fat and carbohydrate of energy was computed and the comparison between sexes were performed using the independent samples t test. The proportion at risk of inadequate intake of nutrients was compared between the 2 sexes using Chi square test or Fisher's Exact test as appropriate. Association between the proportion of boys and girls with intakes which were adequate, lower, and higher than the recommended AMDR with weight categories was analyzed using chi-square test and the odds ratio with 95% confidence interval has been reported.
Multinomial logistic regression was performed to examine association of food group intake with BMI 8 categories after adjustment for age and total energy intake. Adjusted odds ratios with 95% confidence intervals are reported. The proportion of children consuming fortified foods was reported.
The percentage contribution of macronutrients consumed at breakfast to the whole day's intake was also calculated. A two-sided level of significance less than 5 % was considered statistically significant.
All analyses were performed with the SPSS program (version 25.0, SPSS, Chicago, IL, USA).

Results
A total of 634 school children were recruited. The profile of the children recruited into the study is represented in Table 1. Underweight and overweight prevalence were comparable between boys and girls, but height-for-age z score was higher in boys (p = 0.001). Among the nutrients examined, the intake of energy and all micronutrients other than vitamin B 12 and vitamin A were higher in boys (Table 2). Carbohydrate, fat and protein as percent of energy were not different between the sexes. Food group intake of cereals, nuts, added fats and oils, added sugar and milk intakes were higher in boys. Although the risk of inadequate intakes was over 5% for all nutrients, except for protein in boys, the risk was higher in girls for protein, iron, thiamine, riboflavin, niacin, B 6 and B 12 , but not for dietary folate and vitamin A intakes (Table 3). The association between tertiles of food group intake with weight categories was also performed by sex ( Figure 2). In boys, the proportion of underweight children were higher in the lowest tertiles of  our study population had an intake of about 50 kcal/kg for boys and 45 kcal/kg for girls. Although this may seem lower than the requirement, it should be noted that studies have reported low physical activity levels in children in India (23,40,41) and the requirement could be lower as these are set based on the assumption that children have a moderate level of activity.
In our study, the percent of energy contributed by macronutrient intakes was not significantly different between boys and girls. That macronutrient intakes are higher among boys is not surprising as the difference in rates of growth and the change in fat mass and fat free mass warrant a higher requirement among males (21). Sex (biological or physiological characteristics) or gender-based (socio-cultural aspects, behaviours and attitudes) differences in diet intakes and behaviours have been noted in a few studies in adult men, women and children (22)(23)(24)(25)(26)(27). In a study conducted in Canada, it was reported that men reported higher intakes of energy, increased energy density and a higher percentage of energy consumed from fats and a decreased score for emotional susceptibility to disinhibition compared to women (28). A food choice favouring high energy foods is likely to lead to weight gain and subsequently to overweight and obesity (29). The physiological controls for eating and energy homeostasis are purportedly different between the sexes although delineating the differences between biological and non-biological causes is difficult (30). In a study across 23 countries, it was reported that women reportedly chose lesser high fat foods, more high fibre foods and fruits with 50% of the food choice based on health beliefs and 20% on dieting status (30,31). The choice of food and composition of the diet are generally culture specific (20); but the type and quality of diet affect the weight status (22). Differences in consumption among male and female children, however, has not been explored in other countries.
In general, the risk of inadequate intakes was high for most micronutrients examined, with the highest reported for vitamin A, folate, B 12 and iron. This is certainly a matter of concern implying that efforts to improve or diversify diet intakes is essential. Inadequate intakes of protein, iron and the B vitamins, except for folate and B 12 were significantly higher in girls compared to boys. Thus, the micronutrient density of the diets of school-going children is inadequate to meet requirements and education to both the child and the family on methods to improve nutrient density of diets is required.
The inadequacy of intake of micronutrients has not been conclusively demonstrated earlier among school going children in India as diets were analysed using household consumption units and not using an individual's dietary data using 24-hour dietary recalls, estimates of which are closer to the true value of consumption.
That over 50% of children had low intakes of fat and 23% for protein compared to the recommendations for the acceptable macronutrient range is a matter of concern as it indicates a low diet quality. The association of low fat and protein intakes and high carbohydrate intakes with underweight, accentuates this finding further and implies that carbohydrate intakes largely contributed by cereals is higher in those underweight, as diverse diets are probably costlier (3).
In our study, overweight boys consumed a higher percentage of energy from proteins and lesser percentage of energy from carbohydrates rather than increasing their fat percentage from energy.
This was largely through increase in intake of milk and milk products, and nuts and an upward trend towards consumption of eggs, fish and meat rather than an increased consumption of pulses. But this finding needs to be validated further as even with lower intakes compared to the developed countries an association has been shown. A similar observation was observed among school aged children in Germany where children with medium or high protein intakes had higher BMI z scores than those with low protein intakes (32). This could translate to a higher lean body mass in boys. In girls, the proportion of children being in the highest tertile of carbohydrate intake and lowest tertile of fat intake was significantly higher.
Among these children 13.7% were underweight and 18.  (34). The present study conducted a decade later has shown a substantial increase in overweight and obesity in school children. With childhood overweight and obesity rates increasing across socio-economic groups in urban areas, there is a need to plan public health programmes accordingly as it is likely to track to adulthood and to later chronic disease (5,35).
Further, prevalence of overweight and obesity among women is much higher compared to men as per the latest National Family Health Survey (NFHS 4) (36), a phenomenon observed across all the continents. In a review of overweight and obesity in children across several countries (37) a similar trend was seen where prevalence was higher in female children. However, in our study on school going children, this difference between boys and girls was not observed The fact that there are substantial increases in overweight and obesity levels in school going children as compared to children below the age of 5 years as reported in NFHS 4 (~ 2.0%), indicates that the problem is likely to exacerbate further with obesity tracking into adulthood and probably further escalating the problem of non-communicable diseases in India.
There is a possibility that overweight children consistently underreport their intakes and hence associations are not evident. Although it has been noted that children's accuracy for reporting items and amounts improved with multiple recalls, discrepancies in reporting varied across individual children with underreporting increasing with increase in BMI (38)(39)(40)(41). The accuracy of reporting was also found to be better among females than males (39).
The strength of the study is that it is representative of children from different socio-economic strata of society and included diet intakes of children with weight status ranging from underweight to overweight. Most studies on diet intakes are from the developed countries which have marginal rates of undernutrition, and not from a developing country. Capturing individual diets of children across India is required to understand how children eat, the nutrient density and inadequacy of their diets.
A primary limitation, however, is that data capturing diet intakes cannot be without errors. It has also been reported (42) that children do have the ability to participate in recalls without any assistance, but the recall needs to be in the last 24 hours. However, the method of collection and the reporting by subjects influence the nature and quantum of error (18). The study used three 24-hour recalls adjusted for within-and between-person estimates using statistical techniques for nutrients of interest (16,18), that can provide an intake value closer to the true intake of the population group and thus is a valid measure of dietary intake. The data was also collected over a period of one school year thus reducing the degree of error.
Another limitation of the study was that physical activity was not captured and thus activity levels could not be adjusted for in this analysis. Although subjective measures of dietary data collection using a standardized protocol were used, the likelihood of underreporting and/or misreporting is possible as reported in other studies (43,44). With the study being cross-sectional, conclusions on cause-and-effect relationships and the direction of the associations observed cannot be made.
The fact that a large proportion of overweight children used fortified foods needs to be explored further. Unlike other countries, all children consumed breakfast and there was no association with the percentage of macronutrient intake consumed through breakfast with weight categories. Similar studies also need to be conducted across different states in India as the extent of nutrition transition can be different across states.
It should be noted that it is difficult to decipher whether the differences between girls and boys can be attributed to biological factors related to sex or social factors related to gender. It is most likely that both factors contribute to these differences (22,37).

Conclusion
There is a dearth of data on nutrient and food intakes of children in India, A significant difference in nutrient and food group intake was observed between male and female children in India. Overall, energy intakes were lower in this population of children from the urban area. The risk of inadequate intake of protein and micronutrients were high especially for vitamin A, folate, vitamin B 12 and iron.
Except for vitamins B 12 and A, all other nutrients were different between males and females. About 50% of children had fat intakes below the AMDR. In general, the quality of diets in children need to be improved to reduce underweight and arrest the increase in overweight and obesity in children which is likely to track into adulthood. Public health programmes combining messages on healthy eating and adequate physical activity need to be designed to address the double burden of malnutrition through Adjusted OR for the association between food group intake tertiles and being underweight and overweight in boys and girls. Data for 10 food groups adjusted for age and energy intake is shown. The dotted line at 1 in Y axis indicates the reference line which is the highest tertile of intake. The adjusted odds ratio (95% CI) is represented as error bars for the lower and mid tertiles for each weight category. * p <0.05.

Supplementary Files
This is a list of supplementary files associated with this preprint. Click to download. SupplementaryDietaryintakesassocwithweightstatus.docx