|Home | About | Journals | Submit | Contact Us | Français|
Childhood obesity has become a public health crisis in the US . Meanwhile, eating disorders have become a particular concern among adolescent girls, who are vulnerable to social and peer pressures [2,3]. It is challenging to measure these pressures. One way of doing so is to examine the role of schooling on their weight status. Ideally, girls would be randomized into schooled and non-schooled groups, which is unethical. Alternatively, age of enrollment works as a natural proxy for the number of years of schooling [4,5]. State-specific age-enrollment policies require children born after a given cut-off date to delay admission until a year later than those born prior to that date . Therefore, children born a few days apart may be in different grades. Stratifying by age-at-enrollment creates two comparable groups with similar developmental stages but with a one year difference in their level of schooling. We utilized this approach to examine whether earlier school enrollment was negatively associated with weight status among adolescent girls.
This study used the National Longitudinal Survey of Youth 1997 cohort (NLSY97), an annual survey of a nationally representative cohort of adolescents born between 1980 and 1984. We utilized data collected between 1997 and 2006, restricting our sample to girls not yet graduated from high school and excluding those with missing body measures, those who were pregnant, and those who had repeated or skipped grades. Our final sample included 14,413 person-year records.
The body mass index (BMI) is defined as 1. We calculated participants’ BMI z-score following the CDC’s formula for the age-specific percentile . Overweight and obesity were defined as BMI ≥ 85th percentile and ≥ 95th percentile, respectively .
School entry status was defined as (a) “early birth date”, i.e., respondents’ birth dates were before the admission cut-off dates, (b)“late-starting” means a one year delay of normal admission into the education system2. The interactive term indicates the marginal effect of early birth date on late starters. For example, if the cut-off-date was September 1, 2000, girls born on August 31, 1994 were late starters if they started their first grade on September1, 2001, while girls born on September 2, 1994 were late starters if they started on September 1, 2002. Therefore, late starters born in August 1994 still had one more year of schooling than late starters born in September 1994.
Linear and logistic models were used with controlling year- and state-fixed effects. Age, race/ethnicity, maternal education and maternal body weight status were controlled. Model A used the school year as an explanatory variable, which was approximately the age effect. For Model B, we created three subgroups whose birth dates were within 1, 3, and 6 months of the cutoff-dates. Hypothetically, the effect of early school entry on obesity should be the most significant between girls born no earlier or later than (“within”) one month of the cutoffs, while the schooling effect should be gradually dominated by the age effect within the 3 and 6 months’ gap. Huber-White standard errors were estimated.
The mean respondent age was 16.6 years. About 51.5%, 26.4%, and 22.1% of the sample were white, black, and Hispanic, respectively. Forty-percent of the girls had mothers who were overweight or obese; 46.9% had mothers who attended college, 32.8% of their mothers graduated from high-school and 20.2% of their mothers were high-school dropouts. The mean BMI of the girls was 23.4, with 18% being overweight and 12.2% being obese. About half of the girls were born before the admission cut-off-dates. Overall, 10% of girls with early birthdates and 3.4% with late birthdates entered school later than expected. Late starters had higher BMIs and a higher prevalence of overweight and obesity; the results were consistent across age groups (Table 1).
Model A in Table 2 shows that one additional year of schooling was correlated with a reduced BMI, meaning adolescent girls who had more schooling were less likely to be obese (OR = 0.862, p < 0.001).
Model B in Table 2 demonstrates the effect of school entry on body weight status. For BMI and BMI z-score, all except one coefficient of an early birth date were significantly negative, indicating that one additional school year was associated with reduced BMIs. Among girls with birthdays within one-month of the cutoff-dates, the coefficient of late starting was significantly positive (β = 0.311; p = 0.02), indicating that late starting may be correlated with weight gain in adolescence. The interactive term was significantly negative in the sample within one month of the cut-off dates, indicating that late starters with early birthdays were slimmer than those with later birthdays, possibly due to the additional one year of schooling. As expected, these terms became insignificant within 3 and 6 months of the cutoff-dates.
For overweight and obesity, all ORs of early birth date were below 1, indicating that one additional school year for girls decreased the risk of overweight and obesity in adolescence.
This study explores the impact of the age of starting education on obesity among adolescent girls. It found that early school entry may reduce girls’ likelihood of becoming obese in their teen years. Girls born a few days earlier than the cut-off dates were thus likely to interact with more mature adolescents in their educational cohort who might be more motivated to be slim . An additional motivational factor may be the U.S. school health curriculum, which is grade-specific .3 Adolescent girls who were admitted to school earlier may benefit from the higher standard of the more advanced-grade curriculum to maintain their body weight.
Our research suggests that earlier admission to school may have a long-term protective effect in terms of adolescent girls’ obesity risks. It indicates that utilizing the age at enrollment, a natural experiment to assign students randomly into different amounts of schooling, better estimates the effect on obesity of time spent in school. This study improves the measurement of schooling on obesity among adolescent girls.
We are very grateful to John Cawley, Jordan Matsudaira, Kosali Simon, Donald Kenkel, George Jakubson, Byung-Kwang Yoo, and Obesity Luncheon at Cornell for many valuable suggestions. Thanks also go to Pinky Chandra at the Cornell Restricted Access Data Center (CRADC) for her assistance to work on NLSY97-Geocode files. Dr. Q. Zhang was partially supported by National Institute of Child Health and Human Development (R03 HD056073).
1To correct for the potential self-reporting bias, we regressed measured height and weight on self-reported measures using National Health and Nutrition Survey data, using regression coefficients to predict the measured height and weight in NLSY97 .
2Specifically, a student was a “late starter if the calculated number of years of schooling based on her birth-date and the state’s admission cut-off-date was larger than her reported highest grade attended or completed.
3For example, Maryland’s Voluntary State Curriculum for Physical Education provides a “grade specific delineation of what students know and be able to do” .
Conflict of interest: There is no conflict of interest.
Financial disclosure: Dr. Q. Zhang was partially supported by National Institute of Child Health and Human Development (R03 HD056073).
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.