Article Text

Longitudinal evaluation of milk type consumed and weight status in preschoolers
  1. Rebecca J Scharf1,
  2. Ryan T Demmer2,
  3. Mark D DeBoer1
  1. 1Division of Developmental Pediatrics, Department of Pediatrics, University of Virginia School of Medicine, Charlottesville, Virginia, USA
  2. 2Department of Epidemiology, Mailman School of Public Health, Columbia University, New York, New York, USA
  1. Correspondence to Dr Mark Daniel DeBoer, Division of Pediatric Endocrinology, University of Virginia School of Medicine, PO Box 800386, Charlottesville, VA 22908, USA; deboer{at}


Objective To evaluate relationships between type of milk consumed and weight status among preschool children.

Design Longitudinal cohort study.

Setting The Early Childhood Longitudinal Study, Birth Cohort, a representative sample of US children.

Participants 10 700 US children examined at age 2 and 4 years.

Main outcome measures Body mass index (BMI) z score and overweight/obese status as a function of milk type intake.

Results The majority of children drank whole or 2% milk (87% at 2 years, 79.3% at 4 years). Across racial/ethnic and socio-economic status subgroups, 1%/skim milk drinkers had higher BMI z scores than 2%/whole milk drinkers. In multivariable analyses, increasing fat content in the type of milk consumed was inversely associated with BMI z score (p<0.0001). Compared to those drinking 2%/whole milk, 2- and 4-year-old children drinking 1%/skim milk had an increased adjusted odds of being overweight (age 2 OR 1.64, p<0.0001; age 4 OR 1.63, p<0.0001) or obese (age 2 OR 1.57, p<0.01; age 4 OR 1.64, p<0.0001). In longitudinal analysis, children drinking 1%/skim milk at both 2 and 4 years were more likely to become overweight/obese between these time points (adjusted OR 1.57, p<0.05).

Conclusions Consumption of 1%/skim milk is more common among overweight/obese preschoolers, potentially reflecting the choice of parents to give overweight/obese children low-fat milk to drink. Nevertheless, 1%/skim milk does not appear to restrain body weight gain between 2 and 4 years of age in this age range, emphasising a need for weight-targeted recommendations with a stronger evidence base.

  • Nutrition
  • Growth
  • Epidemiology

Statistics from

What is already known on this topic

  • The American Academy of Pediatrics recommends that children ≥2 years old consume low-fat milk.

  • Data have been mixed regarding associations between consumption of low-fat milk, weight status and weight gain over time in preschoolers.

What this study adds

  • Consumption of low-fat milk did not restrain weight gain in preschoolers over time.

  • Consumption of low-fat milk was associated with increased risk of overweight/obesity between 2 and 4 years of age.

  • Healthcare practitioners seeing preschool children may wish to focus on weight-control practices with a stronger evidence base than available for consumption of low-fat milk.


The current epidemic of childhood obesity is apparent before preschool,1 contributing to health consequences for the current generation.2 ,3 This increases the need for evidence-based prevention and treatment approaches for effective weight control among preschoolers.

One point of emphasis has been on lowering the fat content of milk the children consume. The American Academy of Pediatrics (AAP) and the American Heart Association have advocated the use of low-fat (1%) or skim milk for all children after 2 years of age to reduce saturated fat intake given the potential effects on weight gain.4 ,5 However, data linking milk type to weight gain in preschoolers are mixed. One cross-sectional study of preschoolers evaluated National Health and Nutrition Survey data for 1999–2002 and reported no relationship between milk type and obesity status.6 Another group reported higher body mass index (BMI) among children drinking 1%/skim milk,7 while a third group reported higher BMI among a cluster of preschoolers drinking 2%/whole milk.8 Still other researchers have suggested a low prevalence of preschoolers drinking low-fat/non-fat milk.9 Thus , the efficacy of the AAP recommendations and overall adherence to them remain unclear.

Our goal was to evaluate relationships between milk fat consumption and BMI among a large cohort of preschool children studied as part of the Early Childhood Longitudinal Survey, Birth Cohort (ECLS-B), a prospective, representative survey of children born in the USA in 2001 and assessed at both 2 and 4 years of age.1 We assessed whether parental choices of milk type for their preschoolers comply with current AAP recommendations and whether milk consumption patterns among US 2-year-olds predicted the development of overweight/obesity during 2 years of longitudinal follow-up. Our original hypothesis was that low-fat milk would be associated with lower BMI z score and less weight gain over time.


Data set

The ECLS-B is a large multi-source, multi-method study sponsored by the National Center for Education Statistics (NCES; part of the US Department of Education) to examine a wide range of influences on early childhood experiences.1 The NCES ethics review board approved the study. This nationally representative sample of children born in 2001 was selected by randomly sampling >14 000 birth certificates, with a final sample of approximately 10 700 completed parent interviews, giving a 77% response rate. Parents gave informed consent. Longitudinal examinations were performed at ages 9 months and 2, 4 and 5 years. We utilised data from the 2-year-old and 4-year-old evaluations, enabling prospective analysis among preschoolers.


During the 2-year-old and 4-year-old waves, parents were interviewed in their home by trained assessors. The primary caregiver (most often the mother) completed a computer-assisted interview. Beverage intake was calculated from parental responses to several questions. At the 2-year visit, parents were asked if their child usually drinks 2%/whole milk (combined together as an option), 1%, skim, soy or other. At the 4-year visit, parents were further able to choose between whole milk and 2% as separate options in addition to 1%, skim, soy or other. During the 4-year visit, parents were asked a more detailed set of questions regarding the type and frequency of beverage intake, including: ‘During the past 7 days, how many times did your child drink milk?’. Parents were instructed to include all types of milk and milk from a glass, cup or carton, or with cereal. They were instructed that the ½-pint of milk served at school equals one glass. Regarding juice and sugar-sweetened beverages (SSB), parents were asked how many times their child drank 100% fruit juices not including punch, Sunny Delight, Kool-Aid, sports drinks or other fruit-flavoured drinks, and how many times their child drank sugar soda pop, sports drinks or fruit drinks that are not 100% fruit juice. For each of these drinks—milk, juice and SSB—categories for frequency included no intake during the past 7 days, 1–3 times during the past 7 days, 4–6 times during the past 7 days, once daily, twice daily, three times daily and ≥4 times daily. For purposes of reporting prevalence data, these quantities were converted to the recommendations of the AAP: juice and SSB ≤1 or >1 serving daily4 and for milk <2, 2 and >2 servings daily.4 The daily intake of milk fat in grams for the 4-year wave was calculated by multiplying grams of fat per serving for each milk type (skim: 0 g; 1%: 2.4 g; 2%: 4.8 g; whole, 7.9 g10) by the number of servings consumed daily, with children reported to drink >3 servings considered as having consumed four servings daily.

Direct measurements of height and weight were obtained by trained researchers using standardised protocols and equipment including a portable stadiometer and digital scale. Children were dressed in light clothing without shoes. Measurements were taken twice; if these were within 5% of each other, their average was used, otherwise a third measurement was taken and the three measurements averaged. BMI was calculated as weight (kilograms)/(height (meters))2 and converted to age- and gender-specific percentiles and z scores using the 2000 Centers for Disease Control and Prevention US growth charts.11 Weight categories were normal weight (<85th%), overweight (>85th–95th%) and obese (>95th%). For the 2-year time point, children <24 months old were excluded, as BMI is not a validated measure below this age.

Parents identified their child's gender and race/ethnicity. Race/ethnicity was grouped into five categories: white, black, Asian, Hispanic and other. NCES calculated socio-economic status (SES) based on five items: family income, maternal education, maternal occupation, paternal education and paternal occupation.12 ,13 Participants were categorised into SES quintiles (lowest SES=1; highest SES=5). Caregivers identified if their child was predominantly at home during the day or away from the home at childcare or preschool.

Data analysis

We performed all analyses using SAS software, V.9.2 (SAS Institute Inc., Cary, North Carolina, USA), utilising survey procedures with sampling weights provided by the NCES to account for the complex sampling design. All statistical significance tests were two-sided with significance of α=0.05. Unweighted sample sizes were rounded to the nearest 50 in compliance with NCES rules. Using multivariable linear regression models, we performed both cross-sectional and longitudinal analyses as follows. First we regressed: (i) age 2- and 4-year BMI z score on milk-type categories (skim, 1%, 2% or whole milk) cross-sectionally; and (ii) longitudinal change in BMI z score (4-year BMI z score minus 2-year BMI z score) on baseline milk-type categories. Similarly, multivariable logistic regression models were used to examine the odds of overweight/obese across the milk-type categories in both cross-sectional and longitudinal analyses. Regression coefficients, ORs and CIs are reported in the tables. We adjusted 4-year multivariable models for sex, race, SES, juice and SSB intake,14 number of glasses of milk daily and maternal BMI.15 For juice, SSB and milk intake, we used the amount of each of these reported by the parents as the number of daily servings with one to three times weekly=0.29 servings/day and four to six times weekly=0.71 servings/day. We adjusted 2-year models for sex, race, SES and maternal BMI as the other measures were not available.

To assess longitudinal associations of milk type with weight gain over time, we selected children reported to drink 1%/skim milk (low-fat) at both 2 and 4 years and those reported to drink 2%/whole milk (high-fat) at both time points. This approach obtained the purest contrast between milk type and BMI change. For both groups of consistent low-fat or high-fat milk drinkers, we assessed BMI z score at both time points, as well as the intra-individual change in BMI z score between time points. Given known difficulties in the use of BMI z scores over time at BMI extremes16 and given matching ages at time of assessment, we also assessed longitudinal change in raw BMI. In assessing the odds of becoming overweight between time points, we restricted the analysis to children who were normal weight at age 2 years and adjusted for baseline BMI z score, in addition to adjusting for the potential confounders listed above.



We analysed data from 10 700 ECLS-B participants. Of these 7450 at the 2-year wave were aged >24 months and had complete data on milk type (see online supplementary table S1) and BMI, while 8300 at the 4-year wave had complete information (table 1). An additional 200 non-milk drinkers at age 4 years were excluded. Compared to the original data set, the children remaining at the 4-year wave had a slightly higher prevalence in the upper two SES quintiles (41% vs 43%, p<0.05). At both the 2- and 4-year time points, most children drank whole or 2% milk (86% at 2 years and 81% at 4 years). Among those who consistently drank either high-fat or low-fat milk at both time points (as opposed to changing between milk types), 95% drank 2%/whole while 5% drank 1%/skim.

Table 1

Descriptive statistics overall and by type of milk consumed among 8300 children aged 4 years in the Early Childhood Longitudinal Study, Birth Cohort

Milk type and weight status

Overweight/obesity was highly prevalent at both waves, being 30.1% at 2 years and 32.2% at 4 years. The prevalence of 1%/skim milk consumption was higher among overweight/obese children (14% at 2 years, 16% at 4 years) than among normal weight children (9% at 2 years, 13% at 4 years, p<0.01 at both years; table 1 and see online supplementary table S1).

Mean BMI z scores varied significantly across milk type with lower mean BMI z score among 2%/whole milk drinkers compared to 1%/skim milk drinkers (figure 1). These patterns were consistent at both 2 and 4 years and among race/ethnic subgroups (figure 1A,C) as well as across SES quintiles (figure 1B,D). Similarly, linear regression revealed that consumption of higher fat content in milk was associated with lower BMI z score, including after multivariable adjustment for sex, race/ethnicity, SES, intake of juice and SSB, and maternal BMI (all p<0.0001). These findings did not change when evaluated only among the subset of children in childcare or preschool (data not shown). A weaker inverse association was found when relationships between total daily grams of milk fat consumed and BMI z score at 4 years were assessed (p<0.001) (see online supplementary table S2). Finally, we assessed the odds of overweight or obesity according to milk type (table 2). Preschoolers drinking 1%/skim milk (vs 2%/whole milk) had higher odds of being overweight or obese, findings that were strengthened after multivariable adjustment.

Table 2

Logistic regression of milk type on obesity status at 2 and 4 years of age

Figure 1

Mean body mass index (BMI) z scores by milk type, race/ethnicity and socio-economic status (SES) at age 2 and 4 years. Data shown reflect mean BMI z scores for 7450 2-year-old (A, B) and 8300 4-year-old (C, D) participants in the Early Childhood Longitudinal Study, Birth Cohort, broken down by racial/ethnic (A, C) and SES (B, D) groups. p Values for comparisons of mean BMI z scores for drinkers of 1%/skim milk and 2%/whole milk: *p<0.05; **p<0.01; ***p<0.001; NS, not significant (p≥0.05). Hisp., Hispanic.

Longitudinal change in BMI by milk-type group

We next assessed whether 1%/skim milk consumption was associated with increased weight gain over time. As seen in figure 2, children consistently drinking 1%/skim milk at both the 2- and 4-year time points had higher BMI z scores at both evaluations than those drinking 2%/whole milk. When we used linear regression and adjusted for sex, race/ethnicity and SES, there was no significant difference between the low-fat group and the high-fat group in the change in BMI z score over time (p=0.6). These results persisted when change in raw BMI was assessed between the time points (data not shown). However, consistent drinkers of 1%/skim milk who were not overweight/obese at baseline were more likely in a regression model adjusted for baseline BMI to become overweight/obese between 2 and 4 years (OR 1.57, 95% CI 1.03 to 2.42; table 3).

Table 3

Odds of children who were normal weight at 2 years becoming overweight/obese by 4 years among consistent drinkers of 1%/skim milk (n=250) compared to consistent drinkers of 2%/whole milk (n=4900).

Figure 2

Longitudinal analysis of body mass index (BMI) z scores among consistent drinkers of 1%/skim milk and 2%/whole milk at age 2 and 4 years. Mean BMI z scores among children reported to drink 1%/skim milk (n=250) and 2%/whole milk (n=4900) at both time points are shown. Significance is following adjustment for sex, race/ethnicity and socio-economic status. BMI z scores were higher at 4 years compared to 2 years for both groups (p<0.001), but change in BMI z score over time was not different between groups. BMI z score for consistent drinkers of 1%/skim milk versus 2%/whole milk at each time point: **p<0.01, ***p<0.001.


The AAP first started recommending low-fat milk for all children >2 years old in 2005,4 ,5 after the onset of the current epidemic of obesity.17 While prior reports have since noted low adherence to these recommendations,9 at least one report noted lower BMI between preschoolers drinking 2%/whole milk compared to 1%/skim milk.6 Using a large, nationally representative database, we found multiple associations between intake of 1%/skim milk and higher BMI z scores in preschoolers. Across racial/ethnic and SES categories, children drinking 1%/skim milk had higher BMI z scores than those drinking 2%/whole milk. Similarly, preschoolers drinking 1%/skim milk had higher adjusted odds of overweight or obesity than those drinking 2%/whole milk. These data may reflect the fact that parents of children with higher BMIs are more likely to adhere to recommendations of healthcare providers in selecting low-fat milk.

The logic behind these AAP recommendations is that if children drink reduced-fat milk, this results in overall fewer calories consumed.4 It has been well established that as compared to traditional plant-based diets, Western diets high in saturated fat are associated with increased weight gain.18 In both children and adults, key contributors to the current obesity epidemic are the high-fat diets increasingly consumed worldwide.19 Encouraging consumption of low-fat/skim milk instead of high-fat milk promotes a reduction of 5.5–22.5 g of fat (50–202 kcal) daily among children drinking one to three cups of milk per day.10

While the logic of low-fat milk consumption is sound, we are not aware of studies that have randomised preschoolers to low-fat versus high-fat milk to test effects on weight status. Prospective observational studies in children20 and adults21 have associated whole milk intake with lower BMI than low-fat milk intake. At least theoretically high-fat milk intake may result in less weight gain if its consumption leads to an overall decrease in calories consumed. The presence of fat can induce satiety through the release of cholecystokinin and other factors.22 This could potentially lower appetite for other calorically dense foods, as noted in preschoolers who drink excessive volumes of milk and concurrently eat less iron-containing food, contributing to iron deficiency anaemia.23 ,24 In addition, high-fat, low-carbohydrate diets have been associated with improved short-term weight loss, although much of this weight loss was either not sustained25 or was not better than following low-fat diets.26

After noting consistent trends of higher BMI among preschoolers drinking 1%/skim milk, we proceeded to test our original hypothesis that consumption of 1%/skim milk is associated with decreased weight gain over time. We thus evaluated children drinking 1%/skim milk at both 2 and 4 years and compared them to children drinking 2%/whole milk at both time points. Contrary to our original hypothesis, consistent drinkers of 1%/skim milk had a higher OR for becoming overweight between 2 and 4 years. This may have been related to residual confounding factors that we did not account for in our analysis. Overall, there were no significant differences in the absolute increase in BMI between groups, suggesting against low-fat milk consumption as a cause of additional weight gain beyond that seen for 2%/whole milk.

Certainly there remained among consistent 1%/skim milk drinkers an overall increase in BMI z score over time, potentially emphasising that obesity is a multi-factorial disorder, with contributions from genetic and environmental factors of which dietary patterns are just one component.18 Healthcare practitioners seeing children are faced with limited clinical time to make numerous health recommendations and need to select advice most likely to be efficacious—in this case in protecting against excessive gain in BMI. Our data reveal that intake of 1%/skim milk did not achieve the control of weight gain (compared to 2%/whole milk) that logic would have predicted, although it may be that drinkers of 1%/skim milk would have gained even more weight had they not been drinking low-fat milk. Nevertheless, national scientific societies—and practitioners following their recommendations—may need to reconsider current recommendations regarding low-fat milk intake without further dietary guidance as a means of weight control, choosing to instead emphasise other noted interventions such as decreased television viewing,27 increased physical activity28 ,29 and decreased juice and SSB intake,14 ,30 as well as a focus on non-Western diets with higher vegetable content.18 This emphasis on more efficacious recommendations is particularly relevant when one considers data that indicate that the simpler a set of recommendations parents are given, the more likely they are to retain and follow these recommendations.31

This study had several weaknesses. We employed secondary analysis of data from ECLS-B using measures that were not themselves primary outcome measures. The type of milk consumed by children in the study was reported by parents and not directly observed. Also, we lacked data on other forms of food intake, which could have enabled assessment of the association of milk type with total calorie consumption, and we lacked data on physical and sedentary activities that may have represented important confounders. Further research will be needed to assess whether the associations persist when these other measures are accounted for. However, this study also had significant strengths, particularly its use of a large, nationally representative database to address concerns related to AAP guidelines and use of prospectively gathered observational information to assess the effects of lifestyle factors on measured BMI over time.

In conclusion, we found that among preschoolers, consumption of 1%/skim milk was associated with overweight and obesity. While uncertain, these findings may reflect an increase in adherence to the recommendations of physicians and the AAP among families of children who are overweight/obese. Nevertheless, the prevalence of consumption of 1%/skim milk in this age range remains low, as less than 20% of overweight or obese children drink 1% or skim milk. Our data do not support 1%/skim milk consumption as the sole way to restrain gains in BMI among preschoolers. This may mean that efforts towards weight control among overweight/obese preschoolers would be better directed at other interventions with established efficacy.


Supplementary materials

  • Supplementary Data

    This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.

    Files in this Data Supplement:


  • Contributors RJS , RTD and MDD participated in the design and analysis of the research. MDD and RJS were responsible for writing the manuscript and MDD had primary responsibility for the final content. All authors read and approved of the final manuscript.

  • Funding This work was supported by NIH grant 5K08HD060739-03 (MDD).

  • Competing interests None.

  • Ethics approval The Ethics Review Board of the National Center for Education Statistics (US Department of Education) approved this study.

  • Provenance and peer review Not commissioned; externally peer reviewed.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.