Calories, Caffeine and the Onset of Obesity in Young Children

Abstract

Objective:

High calorie foods and beverages, which often contain caffeine, contribute to child overweight/obesity. We evaluated the results of an educational intervention to promote healthy growth in very young children. Secondarily, we used detailed diet data to explore the association of nutrient intake with the early development of overweight and obesity.

Methods:

Mothers were obese Latina women, enrolled prenatally, and their infants. Specially trained community health workers provided breastfeeding support and nutrition education during 10 home visits, birth to 24 months. At follow-up, age 18-36 months, we measured growth and completed detailed diet recalls (1-7 recall days/child).

Results:

Of 174 infants randomized, 106 children were followed for 24-36 months. The educational intervention did not prevent overweight/obesity. Forty-two percent of children became overweight or obese. Fifty-eight percent of children consumed caffeine on at least one recall day. Mean intake was 0.48 mg/kg/day. Caffeine correlated with higher consumption of calories, and added sugar and decreased intake of protein, fiber and dairy. Compared with days without caffeine, on days when caffeine was consumed, children ingested 121 more calories and 3.8 gm less protein. Children frequently consumed less than the recommended daily intake of key nutrients such as fiber, vegetables, whole fruit, and vitamins.

Conclusions:

Caffeine was a marker for increased intake of calories and decreased intake of key nutrients. When discussing dietary intake in early childhood, practitioners should screen for nutrient deficiency in young children and recommend limiting the intake of caffeinated foods and beverages.

Background/Objective

Overweight and obesity occurs during infancy and predisposes to childhood obesity. Some experts believe that the prevention of childhood obesity should begin during the first 12 months, when important health habits are established. We trained community health workers to provide in-home nutrition education to prevent the onset of overweight during infancy (1,2). At age 24 - 36 months we recorded growth and collected detailed diet recall data. Because many children in the study had consumed caffeine, we had an opportunity to explore the possible relationships between calorie intake, caffeine, and the consumption of key nutrients. The child obesity literature has focused primarily on the sugar/calorie content of caffeinated foods and drinks (3,4). Yet caffeine, because of its neuropsychological effects, might play a role in habituating very young children to the consumption of high calorie foods.

Studies on caffeine consumption in children focus mostly on caffeine intake, exclusive of other nutrients. One study found that 98% of children ages 2-3 years consumed caffeine during the 24-hour period examined (5). Pharmacologic doses of caffeine have multiple effects on children such as sleep disturbance, headache and anxiety (6–14). Sleep deficiency correlates with overeating and overweight/obesity in children (10), partly because children awake at night have more time to snack. Dependence on caffeine in humans is reinforced with repeat consumption; when caffeine is discontinued, regular users experience withdrawal symptoms such as depression, irritability and headache (15).

Caffeine is present, but in varying concentrations, in high-calorie sweetened food and drinks such as carbonated beverages, sweet tea, chocolate breakfast cereals, and chocolate milk. Although the beverage industry has claimed that caffeine is added for flavoring (16), it is more likely that the industry adds caffeine for the stimulant effect. Caffeine in sweet beverages has been shown to reduce sweetness by 10% (17).

Whether and how caffeine itself specifically affects calorie intake and weight gain, separate from its association with high calorie food and drink, is not known and has not previously been explored in adults or children. Excluding its association with excess calories, caffeine could even have beneficial effects on overweight. Caffeine increases energy expenditure and stimulates fat oxidation partially through stimulation of the sympathetic nervous system (18). In young children, because it is a stimulant, caffeine could increase energy expenditure due to increased motor activity.

The aims of this paper were first to report the negative results of the educational intervention to prevent the development of overweight and obesity in very young children. Second, we evaluated possible links between key nutrients, caffeinated food and beverages, and the development of overweight.

Methods

Intervention and data collection:

We have previously published outcomes of the clinical trial (Preventing Childhood Obesity through Early Feeding and Parenting Guidance) at subject age 12 months (1,2). Mothers and their infants were enrolled in the study from March 2013 until October 2014. Mothers were low income, obese, pregnant, Latina participants in the Special Supplemental Nutrition Program for Women, Infants and Children (WIC). Institutional review boards at Arizona State University and the state and city health departments approved the informed consent and research protocol. Written informed consent was obtained from the mothers. The study was registered with ClinicalTrials.gov (NCT01905072). A bilingual research assistant obtained socioeconomic and demographic data from the women antepartum. Infants were enrolled at birth and randomized to intervention versus control. The principal investigators were blinded to randomization status until after data collection was complete.

Intervention mothers received home visits by trained Spanish-fluent community health workers who visited the homes once before delivery and at ages 1 and 2 weeks, and at 2, 4, 6, 9, 12 18 and 24 months. Workers documented and plotted growth for the parent and provided counseling on breastfeeding, nutrition, growth, physical activity, child development, sleep, and safety. Workers counseled on feeding at each visit. Mothers were encouraged to breastfeed and/or avoid giving excess formula or adding sugar or solids to the bottle. Workers did not visit the mothers randomized to the control group.

A trained bilingual research assistant visited the homes of all children in the intervention and control groups to obtain growth measurements and diet recall data at ages 18, 24, 30, and 36 months. Multi-day diet recalls were recorded using the National Cancer Institute’s Automated Self-administered 24-Hour Dietary Assessment Tool (ASA24®) (19–22). At each visit, in conversation with the parent, the assistant attempted to obtain nutrition data on at least three recall days: one weekend day and two weekdays. This was not always possible because on a given day someone other than the parent may have been caring for the child, or the parent could not provide accurate recall. Although the original intent was to evaluate all children at age 36 months, diet recalls were begun at 18 months due to concern that some families were planning to move from the area or would be lost to follow up. Regardless of the number and timing of diet recalls, no recall data were excluded from the analysis. The assistant recorded the diet data immediately on data collection sheets; data were entered later on the ASA24 website. Dietary supplements were not recorded. Based on the amount of each food, the ASA24 program calculated the nutrients present and input to the ASA24 TNMYPHEI file (Total Nutrients My Personal Healthy Eating Index) (21,22) the 24-hour intake, in the appropriate units, for each nutrient. In addition, the ASA24 MS (My Selection) listed all foods ingested during recall days, from which we identified the caffeine-containing foods the children had consumed.

Data management:

Demographic and growth data were entered on REDCap (Research Electronic Data Capture) (23). Data were downloaded from REDCap and ASA24 to Microsoft Excel® spreadsheets and converted to SAS® (version 9.4, Cary, NC) databases on a personal computer. Body mass index values were calculated using a macro from the US Centers for Disease Control and Prevention that draws on reference values from the World Health Organization growth data for healthy children (24). Growth classifications were: underweight < 10^th percentile (BMI-z < −1.28), normal weight ,10^th to <85^th percentile (BMI-z −1.28 to <1.04) , overweight 85^th to < 95^th percentile (BMI-z 1.04 to < 1.64) , and obese ≥95 percentile (BMI-z ≥1.64). We classified malnutrition by World Health Organization definitions: severe BMI z-score < − 3.0, and moderate between −2.0 and −3.0 (25).

Data were evaluated for outliers, and implausible values were checked against the original files and corrected or deleted as previously described (2). Extreme values were checked for plausibility by scrutinizing the daily food records in the ASA24 MS (My Selections) file and comparing these with usual energy and nutrient intake from foods and beverages for children aged 24-35.9 months (26,27). To account for incomplete parent recall, caffeine and nutrient intakes were standardized per 1000 Kcal/day.

Statistics, results of the clinical trial:

We used the chi-squared, Fisher exact test, to evaluate the primary aim of the clinical trial, underweight, normal, overweight and obese at the final measurement. Comparisons were between the intervention group versus control. Because healthy growth at age 12 months was influenced by breastfeeding (2), the analysis also assessed growth by breastfeeding status at age two months, any versus none,

Co-correlations and rationale for selection of key nutrients:

ASA 24 provides nutrition data on 100 variables. In order to decrease the likelihood of finding a statistically significant result by chance, the analysis favored a parsimonious approach to variable selection. We identified the smallest number of key variables that represented distinct non-overlapping domains, as determined by evaluating correlation coefficients with P set at <0.05. We calculated correlations using SAS® PROC CORR, Spearman. The final set of variables selected for detailed analysis were caffeine, total calories, total protein, fiber, total fat, added sugar, total dairy and folate. Total protein was selected because it correlated only with zinc (P<0.0001) but not with fiber, whole fruit, total fat, dairy, or any vitamin or mineral. Total fiber served as the representative for carbohydrate, total vegetables, whole fruit, and vitamin C (all P<0.0001). Total fat correlated with zinc (P=0.0082), and folate (P=0.0012), but not with protein (P=0.0568). Added sugar correlated positively only with caffeine. Total dairy correlated only with vitamin D and calcium (both P<0.0001). Folate represented the vitamin group because of high correlation with vitamins A, B1, B2, B6, B12, niacin (all P<0.0001) and vitamin E (P=0.0146). For interest, the following additional variables were selected for comparison against recommended daily requirement: milk, calcium, vitamin D, vitamin E, vegetables, whole fruit, and sodium.

Nutrient intake in relation to caffeine consumption:

Children were the unit of study; each child served as his/her own comparison. Using the TNMYPHEI file, we identified all children whose diet data included both days when caffeine was and was not consumed. Because some data were skewed, we compared median nutrient intakes using the Wilcoxon rank sum test with the Hodges Lehmann estimator (SAS® NPAR1WAY, Wilcoxon HL, alpha=0.05).

Children’s nutrient intake versus recommended daily intake:

For each nutrient, children were classified as meeting or not meeting the recommended mean daily intake. Recommended intakes were average daily requirement (EAR) or adequate intake (AI) (26) and/or the U.S. Dept. of Agriculture MyPlate recommendations for children age three years (27).

Results

Demographic data, birth weight, outcome weight and BMI-z (Table 1):

Table 1.

Demographic data including gender, intervention group, breastfeeding status at age two months, primary language spoken at home, mother’s academic attainment, income, parent employment status at enrollment, type of delivery, age at last visit, and weight status at birth and last visit. Baseline demographic data were obtained at the prenatal visit. N = 106 children.

Variable	Category or mean	N (%) or range
Gender	Female	42 (40%)
Groupa	Control	52 (49%)
Breastfed ≥ 2 months	Yes	56 (53%)
Speak Spanish at homeb	Almost always	84 (79%)
	Very often	15 (14%)
	Moderately	4 (4%)
	Very little	1 (1%)
	Not at all	2 (2%)
Highest education level achieved by mother	Some college	13 (12%)
	High school degree	37 (35%)
	Some high school	18 (17%)
	< High school	38 (36%)
Income	$40-50,000	2 (2%)
	$30-40,000	22 (21%)
	$<30,000	82 (77%)
Father employed	Yes	84 (79%)
Mother employed	Yes	16 (15%)
Delivery	Caesarean section	39 (37%)
Birthweight (kg)	3.51	2.50 to 4.80
Age at last weight visit (mo)	33.5	20.2 to 40.6
Weight at last visit (kg)	15.88	10.9 to 40.4
BMI-z at last visit	0.79	−3.38 to 6.35

^a)Control mothers in the study did not receive the educational intervention or home visits by the community health workers; intervention mothers did.

^b)The parent estimated the frequency with which Spanish was spoken at home.

A total of 807 pregnant women were screened at WIC. A convenience sample of 174 mother/infant dyads were randomized; 119 infants were followed to age 12 months (See flow diagram, reference 2). During the following two years thirteen children were subsequently lost to follow up. Six were in the control group and seven in the intervention group, resulting in 13 children with final growth outcomes measured at age 24 months, 26 at age 30 months, and 67 at age 36 months (Total N=106). Fifty-three percent of infants breastfed for a minimum of two months, 79% of mothers spoke Spanish at home “almost always”, and 47% of mothers had completed high school. BMI-z scores ranged from −3.38 to 6.35.

Results of the Intervention:

Growth at follow-up is summarized in Table 2. At the outcome visit 20% of the children were overweight and 23% were obese. The weight status at outcome was influenced by neither the educational intervention (P=0.68) nor breastfeeding (any at age 2 months, P=0.96). Two children in the control group and three children in the intervention group were malnourished at the final visit: BMI-z scores were −3.38 and −3.28 versus −3.07, −2.27, and −2.07, respectively.

Table 2.

Results of the clinical trial at the final visit, classified by weight category underweight, normal, overweight, and obese. (N=106).

Growth status at final visit	Groupa		Breastfed age 2 monthsb
	Intervention	Control	Yes	No
Underweight: < 10th percentile (BMI-z < −1.28)	6	4	6	4
Normal weight: 10th to <85th percentile (BMI-z −1.28 to <1.04)	26	25	26	25
Overweight: 85th to < 95th percentile (BMI-z 1.04 to < 1.64)	12	9	11	10
Obese: ≥95 percentile (BMI-z ≥1.64).	10	14	13	11

^a)P=0.68.

^b)P=0.96.

Correlation between key nutrients (Table 3, N=106):

Table 3.

Correlation of caffeine, calories/day/kg and key nutrients. All nutrient values were standardized on a daily intake of 1000 kcal. Spearman correlation coefficients and probabilities > |r_s| under H0: Rho=0, SAS® PROC CORR. N=301 recall days, 106 children.

	Caffeine	Kcal/day/kg	Protein	Fiber	Total fat	Added sugar	Total Dairy	Folate
Caffeine, mg	1.00	0.27 <0.0001	−0.19 0.0007	−0.12 0.03	0.11 0.06	0.33 <0.0001	−0.15 0.01	−0.09 0.10
Kcal/day/kg	0.27 <0.0001	1.00	−0.03 0.61	−0.01 0.86	0.24 <0.0001	0.09 0.11	−0.06 0.33	0.01 0.91
Protein, gm	−0.19 0.0007	−0.03 0.61	1.00	−0.28 <0.0001	−0.44 <0.0001	−0.22 0.0001	0.24 <0.0001	−0.22 0.0001
Fiber, gm	−0.12 0.03	−0.01 0.86	−0.28 <0.0001	1.00	−0.44 <0.0001	−0.15 0.01	−0.35 <0.0001	0.31 <0.0001
Total fat, gm	0.11 0.06	0.24 <0.0001	0.21 0.0004	−0.44 <0.0001	1.00	−0.27 <0.0001	0.10 0.09	−0.33 <0.0001
Added sugar, tsp	0.33 <0.0001	0.09 0.11	−0.32 <0.0001	−0.15 0.01	−0.27 <0.0001	1.00	−0.09 0.13	0.17 0.004
Total dairy, cups	−0.15 0.01	−0.06 0.33	0.24 <0.0001	−0.35 <0.0001	0.10 0.09	−0.09 0.13	1.00	−0.10 0.07
Folate	−0.09 0.10	0.01 0.91	−0.22 0.0001	0.31 <0.0001	−0.33 <0.0001	0.17 0.004	−0.10 0.07	1.00

There were positive correlations between milligrams of caffeine consumed, calories (r_s = 0.27, P<0.0001) and added sugar (r_s = 0.33, P<0.0001). Negative correlations were observed between caffeine intake and protein (r_s= − 0.19, P<0.0007), fiber (r_s = − 0.12, P=0.03), and total dairy (r_s = −0.15, P=0.01). Added sugar did not correlate with calorie intake but did show inverse correlations with protein (r_s = −0.32, P<0.0001), fiber (r_s = −0.15, P=0.01), and total fat (r_s = −0.27, P<0.0001). Calorie intake correlated with only caffeine (r_s = 0.27, P<0.0001) and total fat (r_s = 0.24, P<0.0001).

Nutrient intake in relation to caffeine consumption:

A total of 20 weekend and 281 weekday diet recall days (N=301 days), consisting of 16,499 individual food entries in the ASA24 MS file, provided the data for each child’s daily nutrient intake in the ASA24 TNMYPHEI file. The mean age at diet recall was 35 months (minimum=18 months, maximum=46 months). The number of recall days per parent varied from 1 day for 11 parents to 7 days for four parents (47 parents provided 2 days, 28 parents 3 days, 5 parents 4 days, 2 parents 5 days, 9 parents 6 days; mean = 2 days per parent). The main sources of caffeine were chocolate milk and caffeinated sodas (Table 4). Sixty-one children (58%) consumed caffeine on at least one day. Twenty children (33%) consumed caffeine on more than one recall day. Overall, children consumed caffeine on 29% of recall days (87/301). Mean caffeine intake, for all days that caffeine was consumed, was 7.26 mg/day (0.57 to 38.80 mg/day) or 0.48 mg/kg/day (0.04 to 2.70 mg/kg/day). Five of 61 children (8%) consumed ≥1.25 mg/kg/day.

Table 4.

Sources of caffeine by number and percent of recall days that each food was documented as having been consumed by children.

Food/drink	Number of days (% of total days with caffeine)*
Chocolate milk	25 (35%)
Caffeinated sodas	18 (25%)
Chocolate cookies	14 (19%)
Caffeinated teas	6 (8%)
Chocolate cereals	6 (8%)
Coffee	2 (3%)
Chocolate syrup	1 (1%)

^*N=61 children consumed caffeine on at least one recall day. Twenty children consumed caffeine on more than one recall day.

Table 5 summarizes the diet data for the 48 children whose data included both days with and without caffeine. Nutrients were normalized per 1000 Kcal/day. Caffeine was associated with increased intake of calories (P=0.03) and added sugar (P=0.001), and decreased intake of protein (P=0.03). On days when caffeine was consumed, children consumed 2.78 more teaspoons of added sugar, and 7.6 more calories per kg. Based on the mean weight of children at the final visit (15.88 kg), on average, children consumed 121 more calories per day and 3.8 gm less protein on days that they consumed caffeine than on days they did not. Based on the children’s average daily intake of protein (45 gm/day/1000 Kcal), compared with days when caffeine was not consumed, children consumed 9% less protein on days they consumed caffeine.

Table 5.

Daily intake of nutrients, normalized to 1000 Kcal/day. Diet recall data are from all 48 children who consumed caffeine on at least one, but not all, recall days. Nutrients are compared, for each child, between days when caffeine was consumed (N=64 days) versus days caffeine was not consumed (N=95 days).

Variable	Caffeine	Mean	Median	Max	Min	Quartile Range	Skewness	Between group comparison: 95% confidence limits (Hodges-Lehmann)	Pr > |z|
Caffeine (mg/day, mg/kg/day)	yes	7.04	2.76	43.77	0.37	7.39	2.22
		0.47	0.20	3.54	0.32	0.69	2.66
Kcal/day/kg	yes	68.07	62.71	208.93	10.42	37.22	1.81	−0.70 to −8.77	0.03
	no	60.46	56.06	168.13	19.17	27.73
Total protein (g/day)	yes	43.29	44.60	69.92	22.60	14.25	0.18	0.63 to 8.84	0.03
	no	47.12	45.99	90.02	9.07	15.59	0.30
Fiber (g/day)	yes	8.87	7.22	29.86	0.89	6.67	1.35	−1.25 to 2.18	0.38
	no	9.29	9.06	23.69	0	6.12	0.56
Total fat (g/day)	yes	36.67	30.79	54.30	17.34	11.59	−0.07	−4.19 to 2.91	0.78
	no	35.60	34.95	69.70	11.26	13.43	0.25
Added sugar (tsp/day)	yes	6.20	4.98	20.09	0.05	5.24	1.14	−3.61 to −0.85	0.001
	no	3.42	2.51	21.09	0	4.27	2.20
Total dairy (cups/day)	yes	1.60	1.35	4.86	0	1.48	0.93	−0.12 to 0.73	0.21
	no	2.18	1.87	8.2	0	2.35	1.27
DFE folate (mcg/day)	yes	278.45	221.68	760.55	82.09	171.22	1.23	−34.58 to 65.08	0.63
	no	307.44	237.18	1086.73	70.31	228.71	1.60

Children’s nutrient intake versus recommended daily intake (Table 6):

Table 6.

Recommended daily intake of nutrients and number (percent) of children whose mean intake did not meet recommended intake. (N=106 children).

Nutrient	Recommended daily intake per 1000 calories	Total not meeting recommended
Protein	44 gm	47/106 (44%)
Total dairy	>1.79 cups	72/106 (68%)
Cups milk	>1 cup	50/106 (47%)
Fiber	13.8 gm	95/106 (90%)
Calcium	>363 mg	11/106 (10%)
Vitamin D	>7.25 mcg	83/106 (78%)
Vitamin E	>3.6 mg	67/106 (63%)
Whole fruit	>1 cup	73/106 (69%)
Cups vegetables	> 1 cup	68/106 (64%)
Upper limit sodium	1088 mg	99/106 (93%)

All the following comparisons, including the recommended requirements, were standardized to a 1000 Kcal/day diet. Taken as a group (N=106), 44% consumed less than the recommended amount of protein, 68% of the children consumed less than the USDA recommended quantity of dairy (recommended = 1.8 cups), 47%, consumed less than one cup of milk; 19% consumed no milk. Ninety percent of children ingested insufficient fiber (EAR fiber = 13.8 gm), 10% consumed insufficient calcium (EAR calcium = 363 mg), 78% and 63% of children did not meet the recommended intake of vitamins D and E, respectively (EAR vitamin D =7.25 mcg, EAR vitamin E= 3.6 mg/day). Sixty-four percent of children consumed less than one cup of vegetables, and 69 percent ingested less than one cup of whole fruit. Ninety-three percent took in more than the upper limit of sodium (upper limit =1088 mg).

Discussion

In these high-risk Mexican-American children, the educational intervention, consisting of multiple home visits by community health workers who promoted breastfeeding and healthy eating, was not successful in reducing the development of overweight and obesity. Several randomized prospective studies have attempted to promote healthy weight gain in children beginning during the first year. Paul et al (28) used research nurses at four home visits during infancy to provide education focused on feeding, sleep, interactive play, and emotional regulation. A control group was only provided information on safety. Compared with control, at age 3-years, the intervention group demonstrated small reductions in BMI-z scores (P=0.04) but not in overweight (P=0.09). Taylor et al (29) evaluated the effect of prenatal parent education about infant sleep. Compared with control, at age 5 years, children whose parents received the sleep intervention were less likely to be obese. Dodd et al. (30) provided dietary, exercise, behavioral strategies and goal setting during pregnancy by a registered dietitian and research assistants. Compared with the control group, at age 18 months children did not show significant differences in overweight. Wen et al. (31) provided eight home visits from specially trained community nurses, one visit in the antenatal period, and seven at 1, 3, 5, 9, 12, 18 and 24 months. Nurses taught mothers healthy infant feeding practices and promoted active play. Compared with control infants, at age 24 months, the mean BMI of the intervention children was slightly lower (P=0.04), but there were no differences in overweight/obesity between the groups.

We scrutinized our nutrition data to evaluate possible explanations for the excess weight gain. In reviewing data from the children as a group (Table 3), calorie intake showed positive correlations with the intake of caffeine and total fat, but not added sugar. Caffeine was closely linked to the intake of added sugar, but added sugar, taken alone, did not correlate with calorie intake. One tsp of sugar (4 gm) contains 16 calories. The same quantity of fat contains 36 calories (9 calories /gm). A small increase in the amount of fat consumed per day, in a young child, would have a greater effect, over time, than the same increase in sugar consumption. Chocolate milk, which contains fat, was the caffeine-containing food most often consumed (Table 4) and could account for the correlation between caffeine and fat.

On average, children ingested 121 more calories on days caffeine was consumed. Overall, children consumed caffeine on 29% percent of recall days. Excess weight gain in children and adults can result from the habitual intake of extra calories. If calories consumed with caffeine were excess, and caffeine intake continued during 29% of days, in 12 months this would have amounted to 10,527 added calories (87 days x 121 calories). This would equate to a gain of 1.4 Kg, assuming 3,500 calories per pound, 7,700 calories per kg. At the final visit, the difference between the lightest study subject meeting the definition of obesity and the mean weight of all study children was 18.18 kg – 15.88 kg = 2.3 kg. Even without considering the contribution of non-caffeinated foods, caffeine-associated calories could have contributed to the onset of overweight.

When caffeine was consumed, children ingested 7.04 mg caffeine/1000 kcal/day, 0.48 mg/kg/day, or 7.46 mg per day (Table 5), based on 1083 kcal/day, mean weight=15.88 kg. The maximum 24-hour caffeine intake was 39 mg. We compared these results with reports from the literature, using the mg/kg/day and/or mg/day intake for a comparable child (age 36 months, 15.88 kg). Caffeine intake, from highest to lowest were reported as ranging from 101 to 4.6 mg/day (32–37). Data from the Bogalusa Heart Study (5) reported a maximum intake at age two years, 7.8 mg/kg/day, tapering off to 2.3 mg/kg/day at age 17 years; white children consumed twice as much caffeine as black. Ahluwalia (37) reported daily intake for Non-Hispanic white, Non-Hispanic black and Mexican-American children: 6.4, 1.3 and 4.2 mg/kg/day, respectively.

The average U.S. adult weighing 83 kg. consumes 180 mg of caffeine/day, which is two cups of brewed coffee, 2.17 mg/kg/day (38). Published recommendations do not comment on doses of caffeine for children ages < 3 years, but several studies recommend limiting caffeine intake to a maximum of 2.5 mg/kg/day for children aged ≥ three years (13,14,35,38), which would be physiologically equivalent to 40 mg caffeine/day in our children. This dose would be physiologically equivalent to 2.3 cups of brewed coffee consumed by our typical 15.9 kg child. This is close to the maximum dose consumed in our study (42 mg/day, based on a 1083 Kcal/day intake). Eight percent of children who consumed caffeine in the study (8/61) consumed ≥1.25 mg/kg/day, which would be equivalent to one cup of brewed coffee per day. At this dose, some children could have experienced side effects and, if caffeine were consumed regularly, could have developed habituation.

When considered as a group (Table 6, N=106), children often consumed less than the recommended servings of dairy, fiber, vitamin D, vitamin E, fruit and vegetables, while also consuming more sodium than recommended. The USDA MyPlate guidelines recommend 1 cup each of fruit and vegetable per day for a moderately active child age 3 years. This recommendation was met by only 31 and 36% of children for fruit and vegetables, respectively. Five children in the study were significantly underweight, but study children were not evaluated for chemical evidence of malnutrition using measures such as serum iron, vitamins, or total protein.

Regarding validity of ASA24, multiple pass recalls conducted over at least a 3-day period that included weekdays and weekend days and using parents as proxy reporters was the most accurate method to estimate total energy intake in children aged 4 to 11 years (39,40). Measuring food weights would have provided the best estimate for young children, but this was not feasible. A Kids’ version (ASA24-kids-2012) was shown not to be superior to the original ASA24 for the collection of children’s diet data.

A limitation of the study was that ASA24 MYPHEI reports total daily intake of nutrients. Due to this lack of detail, daily intake of caffeine could be reported as the sum of caffeine in a soda, plus a chocolate milk plus a chocolate candy. Despite this, our data are consistent with nutrient totals, similarly standardized per 1000 Kcal per day for a representative sample of children living in the United States (26). Results are not generalizable to a typical U.S. population because mothers were obese, Mexican-American pregnant women enrolled in WIC. Also, children classified as not having consumed caffeine, likely would have been caffeine consumers had more diet recalls been done.

We worry about the habit-forming combination of sugar, fat and caffeine, especially when introduced at a young age. But it would be difficult to sort out the relative effects of sugar, fat and caffeine on eating habits in young children. Caffeine content varies widely between foods and beverages a child might consume. A cup (8 oz) of regular caffeinated soda contains 21 mg caffeine and 93 calories. In contrast a cup of chocolate milk contains 5 mg caffeine and 209 calories. This compares with a cup of whole milk, which contains no caffeine and 149 calories. A chocolate cookie that contains only 1.3 mg caffeine contains 150 calories. Habit forming behaviors and side effects associated with the stimulant effect of caffeine could be considerable in a child regularly consuming coffee, as has been observed in some Hispanic cultures (11), or caffeinated cola, as was observed in some of our study children. But the same could not be said about children whose caffeine consumption consisted only in foods containing chocolate. Clearly, the effect of caffeine on children’s nutritional behaviors will need further study.

Infants have an inborn preference for sweet tastes (41). Within hours after birth newborns show a preference for sugar; when given the choice, they consume more sugar solution than pure water (42). In a large study of sweet preferences, children most preferred a sugar concentration equivalent to 11 teaspoons of sugar/8 oz of water, twice the sugar concentration of a standard cola (43). Evidence from neuroscience suggests that epigenetic changes in gene expression are impacted by early-life experiences (44). The addition of caffeine to sweetened beverages and foods would enhance the risk of habituation at an age when brain plasticity may enable experiences to shape neural circuits and thus behaviors. Food preferences are well-established by age 3-4 years, and track into childhood and adolescence (45). To succeed, preventive interventions should focus on early life.

We conclude that young children who consume caffeine may be at risk for diets excessive in calories and lacking in key nutrients. Practitioners should screen infants and young children for caffeine intake and counsel parents to reduce their child’s consumption of caffeinated foods and beverages. Public health authorities should gather more information and develop recommendations that address the appropriate use of caffeinated foods and beverages by young children.

What’s new:

An educational intervention failed to prevent obesity in young children, 58% percent of whom consumed caffeine. Caffeine correlated with intake of more calories and less protein, fiber and dairy. Children often failed to meet the daily requirement of important nutrients.