Published online December 1, 2005
PEDIATRICS Vol. 116 No. 6 December 2005, pp. 1329-1338 (doi:10.1542/peds.2004-2583)
This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Right arrow P3Rs: Submit a response
Right arrow Alert me when this article is cited
Right arrow Alert me when P3Rs are posted
Right arrow Alert me if a correction is posted
Right arrow Citation Map
Services
Right arrow E-mail this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Add to My File Cabinet
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via CrossRef
Right arrow Citing Articles via ISI Web of Science (23)
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Salsberry, P. J.
Right arrow Articles by Reagan, P. B.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Salsberry, P. J.
Right arrow Articles by Reagan, P. B.
Related Collections
Right arrow Nutrition & Metabolism

Dynamics of Early Childhood Overweight

Pamela J. Salsberry, PhD* and Patricia B. Reagan, PhD{ddagger}

* College of Nursing and School of Public Health
{ddagger} Department of Economics and Center for Human Resource Research, Ohio State University, Columbus, Ohio


   ABSTRACT
TOP
 ABSTRACT
METHODS
 RESULTS
 DISCUSSION
 REFERENCES
 
Objective. To study the dynamic processes that drive development of childhood overweight by examining the effects of prenatal characteristics and early-life feeding (breastfeeding versus bottle feeding) on weight states through age 7 years. We test a model to determine whether prenatal characteristics and early-life feeding influence the development of a persistent early tendency toward overweight and/or whether prenatal characteristics and early-life feeding factors influence the likelihood that children will change weight states as they get older.

Methods. Data from the National Longitudinal Survey of Youth’s Child-Mother file were used to implement these analyses. A total of 3022 children were included in this sample. For inclusion in this sample, valid information on height and weight during 3 consecutive interviews when the child was aged 24 to 95 months as well as valid data on prenatal and birth characteristics were needed. The primary outcome measure was childhood overweight (BMI >95th percentile). Multivariate logistic models and first-order Markov models were estimated.

Results. Early development of childhood overweight was associated with race, ethnicity, maternal prepregnancy obesity, maternal smoking during pregnancy, and later birth years. In later years, the factor that contributed the most to being overweight was having been overweight in the previous observation period. However, with conditioning on the child’s having been overweight in the previous observation period, the prenatal factors that contributed to early childhood overweight, except for birth cohort, were also associated with development of overweight among children who had previously been normal weight and perpetuated the persistence of overweight over time.

Conclusions. This research suggests that prenatal characteristics, particularly race, ethnicity, maternal smoking during pregnancy, and maternal prepregnancy obesity, exert influence on the child’s weight states through an early tendency toward overweight, which then is perpetuated as the child ages. These findings are intriguing as they provide additional clues to the genesis of childhood overweight and suggest that overweight prevention may need to begin before pregnancy and in early childhood.

Key Words: child • obesity • mothers • prepregnancy • dynamics

Abbreviations: CDC, Centers for Disease Control and Prevention • OR, odds ratio • CI, confidence interval • SES, socioeconomic status

During the past decade, life-course studies have provided significant evidence for a link between prenatal and early-life exposures and later development of a range of chronic diseases.1–3 The life-course approach conceptualizes health development as a lifelong dynamic process in which genetic, biological, social, and environmental factors interact to produce health states.4–7 The life-course approach is well suited for the study of childhood overweight because it is widely acknowledged that the development of childhood overweight is a multifactor disorder involving genetic, biological, social, and environmental pathways.8 Prenatal characteristics and early-life feeding factors, including maternal weight before and during pregnancy, smoking during pregnancy, infant feeding, and birth weight, have been associated with overweight during childhood.9–12

Maternal nutrition is widely recognized as critical to the developing fetus.13–16 The availability of macronutrients and micronutrients to the fetus is associated with the mother’s prepregnancy nutritional status, as well as her food intake during pregnancy.10,17–19 Evidence suggests that the development of obesity in adolescents and adults may be related to nutritional programming that occurs very early in life.19,20 For example, maternal undernutrition during the Dutch Famine of 1944 and 1945 has been associated with adult obesity, although the timing and the mechanisms are not clear.2,21 Male children of mothers who were exposed to severe malnutrition during the first trimester were at greater risk for obesity at 18 years but not at 50 years. Female children who were exposed to severe malnutrition during the first trimester were at greater risk for obesity at age 50.22 In addition, children of mothers with diabetes have been the subject of numerous studies linking the intrauterine environment with fetal growth and development.14,19,23–27 These studies suggest that the altered glucose-insulin metabolism of the mother may affect a lifelong change in insulin production and sensitivity in her offspring, increasing the likelihood that these offspring will become overweight and develop type 2 diabetes later in life.14,24,28,29 Furthermore, recent studies using animal models have suggested a hyperinsulinemia effect with maternal diets high in carbohydrates and overnutrition.30,31 Contemporaneous maternal overweight or obesity is consistently related to childhood overweight,32–38 but to our knowledge, only 1 study linked maternal weight during early pregnancy to child overweight among preschoolers.11

Other exposures in utero may also have a lasting effect on overweight development. Several studies have documented an association between maternal smoking during pregnancy and childhood overweight.11,39–45 Fetal physiologic changes in response to maternal smoking during pregnancy have been hypothesized. These include maternal poor nutrition influenced by a suppression of appetite by nicotine, a nicotine-induced vasoconstriction that compromises the uteroplacental vasculature, and an increased exposure to carbon monoxide that decreases the fetal uploading of oxygen, affecting fetal growth and development.46 Impaired fetal growth has been strongly associated with smoking during pregnancy, and a dose-dependent relationship has been found between serum cotinine and restricted growth.47

Birth weight is thought to be a crude measure of these exposures in utero. The attained birth weight is a result of the interactive effects of length of gestation, rate of fetal growth, adequacy of the intrauterine environment, uteroplacental function, and genetic potential.29,46,48 There has been a positive relationship between birth weight and later weight outcomes in most studies,11,35,36,38,49–51 with a reported U- or J-shaped relationship noted.52

After birth, influences within the early-life period are also linked to later overweight. In particular, the method of infant feeding, breast- or bottle-feeding, has been studied with mixed results.53–56 Some studies have found a dose-response effect, with an increased protective effect associated with increased duration of breastfeeding.35,53,57,58 One of the mechanisms posited for a protective effect of breastfeeding is that breastfeeding may permit the infant to establish self-regulation in feeding that is not available to the bottle-fed infant.55

Although these studies have tested for an effect between prenatal characteristics and early-life feeding on overweight development at various ages, they have not tracked the dynamics of the process of overweight development. The purpose of this article is to develop and test a model of the extent, timing, and persistence of prenatal characteristics and early-life feeding on the development of childhood overweight across time.

To study the dynamic effects, we propose a model linking prenatal characteristics and early feeding with overweight development in children from birth to 7 years of age. We chose this period because we wanted to examine prenatal characteristics and early feeding practices on overweight among young children but wanted to extend the observation period beyond the period of adiposity rebound. Although childhood overweight is a probabilistic process, a child’s weight status displays persistence over time. In other words, the probability that a child is observed to be overweight across any 2 successive observation periods is a function of whether the child was overweight in the previous observation period. It therefore is natural to conceptualize the dynamic process of childhood overweight as a first-order Markov process.59 Figure 1 presents our model for a child who is observed at 4 points during their life cycle. The initial conditions, observed in period T0, capture the maternal factors of prepregnancy weight, maternal smoking during pregnancy, race, ethnicity, and the early-life influence of feeding method. These factors are primarily either biological, such as intrauterine exposure to nicotine, or social, such as race and access to material resources. Some ways in which social factors influence biological processes are through the quality of diet and lower immunity brought on by exposure to stress.60


Figure 1
View larger version (12K):
[in this window]
[in a new window]
  		Fig 1. Dynamic model of the development of childhood overweight.

 
We hypothesize that the initial conditions can have 2 effects on development of childhood overweight. On the one hand, these initial conditions can "program" structural or functional changes that lead to differences in the child’s lifetime tendency toward overweight. The initial conditions affect the probability of overweight at T1, which through persistence of weight states will result in a lifetime tendency to overweight. The thicker arrows in Fig 1 illustrate the "programmed" lifetime tendency toward overweight hypothesized because of these initial conditions. We also hypothesize that initial conditions may have an independent effect on the dynamics of the process that leads to the development of overweight at T2 and T3 by changing the probability that the child moves between weight states at successive periods. We call this phenomenon the dynamic effect of the initial conditions. The dotted arrows in Fig 1 illustrate dynamic effects. A dynamic effect exerts a direct and independent influence on the probability of transition between weight states across successive time periods conditional on weight in the previous state. Contemporaneous risk factors, represented by the upward arrows in Fig 1, can influence the overweight development process.

Previous studies that analyzed the effects of initial conditions on weight of cross-sections of children at different ages cannot isolate the persistent effects from the dynamic effects of the initial conditions hypothesized by our model. For example, maternal prepregnancy type 2 diabetes, known to occur with increasing frequency in obese women, may affect a lifelong change in insulin production and sensitivity in her offspring, increasing the likelihood that these offspring will become overweight. This represents a persistent effect. Maternal prepregnancy obesity may also be an indicator for a shared postnatal nutritional environment that places children at risk for becoming overweight if he or she was normal weight in the previous observation period or at risk for remaining overweight if he or she was overweight in the previous period. This represents the dynamic effects of the initial conditions. These 2 effects can be identified separately only with longitudinal data over the child’s life course and a dynamic modeling framework in which the child’s weight state is a function of observed weight state in the preceding observation period. The purpose of this article is to use a life-course approach to examine the dynamic processes that drive childhood overweight, examining the effects of prenatal characteristics and early-life feeding practices on childhood overweight through age 7 years.


   METHODS
TOP
 ABSTRACT
 METHODS
RESULTS
 DISCUSSION
 REFERENCES
 
Study Population
We used data from the National Longitudinal Survey of Youth’s Child-Mother file. This data set was based on the National Longitudinal Survey of Youth 1979, which began in 1979 with 6283 young women aged 14 to 21. These women were followed annually until 1994 and biennially thereafter. The military sample (N = 456) was dropped in 1985. The poor white oversample (N = 195) was dropped in 1991. We focus on the 5729 women in the cross-section sample and the oversamples of black and Hispanic individuals. In 1986, the National Longitudinal Survey of Youth’s Child-Mother began biennially surveying the children who were born to these women. As of 2002, the latest round of survey data available, 9595 children were born to the cross-section sample, which included a nationally representative cross-section of women who lived in the United States in 1978 and were born between 1957 and 1964 and the oversamples of black and Hispanic individuals of the comparable birth cohorts.

Our sample is composed of children who had valid information on height and weight and were observed during 3 consecutive biennial interviews when they were aged 24 to 95 months. We required 3 observations on each child. The first observation period, T1, occurred when the child was between 24 and 47 months (age 2–3 years). The second observation period, T2, occurred when the child was between 48 and 71 months (age 4–5 years). The third observation period, T3, occurred when the child was between 72 and 95 months (age 6–7 years). Because the data are collected biennially only beginning in 1986, children who were born before 1982 and after 1996 could not be included in the study. Likewise, children of the 1982 to 1996 birth cohorts who missed an interview between the ages of 2 and 7 years were excluded. Because we are concerned about the effects of prenatal and birth characteristics on subsequent child overweight, we also required valid information on these characteristics.

Of the 6615 children who were born between 1982 and 1996, 4197 had 3 consecutive interviews during the appropriate age ranges. Height and weight were observed in each period on 3331 children. The final sample consists of 3022 children for whom we also observed all of the prenatal characteristics and early feeding practices in the analysis. Because of the loss in subjects as a result of the inclusion criteria, we tested (using t tests) for differences in demographics and outcomes between the eligible sample (6615 children) and the study sample (3022). The study sample overrepresented white individuals by 6 percentage points and underrepresented black and Hispanic individuals by 3 percentage points each. Within age and race/ethnicity groups, the prevalence of overweight for white children was no different between the eligible sample and the study sample. However, the prevalence of overweight for black children aged 72 to 95 months was lower in the study sample than in the eligible sample. The prevalence of overweight for Hispanic children in both of the groups aged 48 to 71 months and aged 72 to 95 months was lower than in the eligible sample.

Data Construction and Measures
The primary outcome of interest was childhood overweight, defined as BMI >95th percentile, as determined using the 2000 Centers for Disease Control and Prevention (CDC) Growth Charts for the United States. The CDC has published gender-specific growth charts of BMI for age for children aged 2 to 20 years.61 The 95th percentile of the 1977 National Center for Health Statistics Growth Charts has been used for many years to screen children who are at increased overall health risk. The upper percentiles of the revised 2000 curves are higher than the corresponding percentiles of the 1977 curves and result in fewer children’s being classified as overweight. The BMI percentile scores were determined using the Nutrition program within the CDC’s Epi Info software program. The program calculates a BMI percentile using the child’s height, weight, age in months, and gender. Height and weight were obtained on each child at the interview. The interviewer measured height and used a scale to obtain the weight of the child. Interviewers were trained to conduct these measurements by the National Opinion Research Center at the University of Chicago. Mothers were given the option of reporting the height and weight. The method that was used to obtain the height and weight measures (measured or reported) was recorded as part of the interview in all years except 1986. In 1986, the interviewers measured the children, but the method to obtain height and weight was not recorded. In preliminary analyses, we found that the reported mean values for height were always less than the measured values in each age category. The reported mean value of weight was greater than the measured value at T1 and T2. However, the reported mean value of weight was less than the measured value at T3. Because of this, we controlled for the method of data collection in the analyses.

Explanatory variables included both time-invariant and time-varying characteristics and were chosen because they have been shown in previous studies to be associated with childhood overweight. The time-invariant characteristics included the following:

Time varying characteristics included

Data Analysis
The analyses were conducted separately on children at T1, T2, and T3. The sample was restricted to children with observations in each of the age ranges, because in some of the analysis, we condition on child overweight during the previous observation period, eg, the probability of a child’s being overweight at age 4 years (6 years) is conditioned on the child’s overweight status at age 2 years (4 years). There were 2070 mothers in the sample, 1307 of whom have a single child in the sample and 763 of whom have multiple children, with an average of 2.25 children per multiparous mother.

The analysis proceeded first with {chi}2 tests of the bivariate association between each of the independent factors and child overweight at the 3 ages. To study the relationship between child overweight and several independent variables, we estimated multivariate logistic regressions at the 3 ages. These models were run unconditioned on overweight during the previous observation period. Next, we estimated first-order Markov models of the transition probabilities between 2 observation periods. To do so, we conditioned the logistic models at T2 and T3 on the child’s being overweight during the previous period. By including the child’s previous weight, we controlled for the effect of factors up to the previous period capitalized into the child’s previous weight state and tested for independent dynamic effects of the initial conditions during the current period. These analyses provided information about the persistence of independent risks across time, the timing of the influence, and the magnitude of the risk. Estimates were obtained using Stata (Stata Corp., College Station, TX). A robust Huber/White/sandwich estimator of the variance of the logistic odds ratios was used to account for the nonindependence in the errors between siblings who were born to the same mother.


   RESULTS
TOP
 ABSTRACT
 METHODS
 RESULTS
DISCUSSION
 REFERENCES
 
Table 1 presents the variables used in this analysis. Of the 3022 children, 53% were white, 27% were black, and 20% were Hispanic. The mean age of the child at T1 was 35 months, at T2 was 60 months, and at T3 was 84 months. Thirty-eight percent of the children were first-born. Approximately half of the children were ever breastfed. The mean age of the mother at birth was 26.9 years, and 29% smoked during pregnancy. Sixty-three percent of the mothers were normal weight before pregnancy, 8% were underweight, 19% were overweight, and 10% were obese before pregnancy. The percentage of mothers who were married decreased slightly over time, from 70% at T1 to 67% at T3. Mean number of years of education was ~13 years at each of the time points. The mean percentage of weight for gestational age was 51%. Measurement of height was recorded in 55% of children at T1 and ~80% at T2 and T3. Measurement of weight was recorded in 47% of children at T1 and ~76% at T2 and T3. The lower prevalence of recorded measurement in T1 was driven by the fact that in 1986 it was not recorded whether height and weight were measured or mother reported.


View this table:
[in this window]
[in a new window]
  		TABLE 1. Sample Characteristics

 
Table 2 details the percentage of overweight children by subgroup. At each of the 3 times, a greater percentage of black and Hispanic children were overweight, compared with white children. The percentage of children who were overweight decreased in all race/ethnic groups as the children aged. This is consistent with the fact that we study children over their life cycle up to a point just past their adiposity rebound. There was little difference in overweight between boys and girls or by parity. Method of feeding (breast or bottle) and maternal smoking during pregnancy showed differences as expected. As the mother’s prepregnancy weight increased, so did the percentage of children who were overweight. Marital status was significant in the 2 earliest periods, with a higher percentage of overweight children in households without a married couple. There were significant differences by child’s birth year at T1 but not at T2 or T3.


View this table:
[in this window]
[in a new window]
  		TABLE 2. Bivariate Comparisons of the Percentage of Children Who Were Overweight at Each Time Period According to Major Risk Factors

 
Table 3 presents the results of the 3 cross-sectional logistic models run at 3 different time points. Several of the bivariate relationships were no longer significant in the multivariate models. At the earliest time, T1, black (odds ratio [OR]: 1.61; 95% confidence interval [CI]: 1.24–2.07) and Hispanic (OR: 1.65; 95% CI: 1.26–2.15) individuals were more likely to be overweight than white individuals. Children whose mothers smoked during pregnancy were more likely to be overweight (OR: 1.37; 95% CI: 1.08–1.73). They were also more likely to be overweight if their mother’s prepregnancy BMI was ≥30 (OR: 1.37; 95% CI: 1.02–1.84). Age of the mother at birth, breastfeeding, percentile weight for gestational age, gender of the child, age of child in months, parity, and socioeconomic status (SES; marital status and education) were not significant. When the height (OR: 0.63; 95% CI: 0.47–0.84) and weight (OR: 0.53; 95% CI: 0.39–0.71) were recorded to have been measured, the child was less likely to be overweight. Children who were born in a later birth cohort were at greater risk for overweight (OR: 1.17; 95% CI: 1.11–1.23).


View this table:
[in this window]
[in a new window]
  		TABLE 3. Multivariate Logistic Regressions of Childhood Overweight

 
At time T2, when the child was between 48 and 71 months, race/ethnicity (black: OR: 1.57; 95% CI: 1.17–2.10; Hispanic: OR: 1.43; 95% CI: 1.05–1.96), smoking during pregnancy (OR: 1.43; 95% CI: 1.11–1.84), and the birth cohort (OR: 1.07; 95% CI: 1.01–1.13) remained statistically significant. When the mother was overweight (OR: 1.40; 95% CI: 1.07–1.83) or obese (OR: 1.69; 95% CI: 1.22–2.34) before pregnancy, the child was more likely to be overweight. Percentile birth weight for gestational age was significant, with increasing weights associated with an increased risk for overweight (OR: 1.50; 95% CI: 1.02–2.19). Older mothers were less likely to have an overweight child (OR: 0.95; 95% CI: 0.90–0.99). Breastfeeding, gender of the child, child age in months, parity, and SES (marital status and education) remained insignificant. When the height (OR: 0.63; 95% CI: 0.44–0.91) or weight (OR: 0.71; 95% CI: 0.52–0.97) was measured, the child was less likely to be overweight.

At time T3, when the child was between 72 and 95 months, race was significantly associated with overweight for black (OR: 1.73; 95% CI: 1.28–2.35) and Hispanic (OR: 1.64; 95% CI: 1.18–2.28) individuals. Smoking during pregnancy (OR: 1.74; 95% CI: 1.32–2.29), maternal prepregnancy weight (overweight OR: 2.11; 95% CI: 1.60–2.78; obese OR: 2.91; 95% CI: 2.09–4.03), percentage weight for gestational age (OR: 1.61; 95% CI: 1.07–1.74), and birth cohort (OR: 1.07; 95% CI: 1.01–1.14) remained significant. In addition, parity (OR: 1.36; 95% CI: 1.07–1.74) and child’s age in months (OR: 1.02; 95% CI: 1.01–1.04) now exerted significant influences. Age of the mother at birth, gender of the child, and SES (marital status and education) continued to be statistically insignificant. When height (OR: 0.55; 95% CI: 0.37–0.81) was recorded to have been measured, the child was less likely to be overweight, but there were no statistical differences between children whose weight was recorded as measured.

Table 4 presents results of the first-order Markov models at times T2 and T3. The greatest risk for overweight is the weight state in the previous period. At T2, the child was approximately twice as likely to be overweight when he or she was overweight at T1 (OR: 2.21; 95% CI: 1.68–2.92 when not measured and OR: 2.70; 95% CI: 1.83–3.99 when measured). At T3, the child was 6 to 16 times more likely to be overweight (OR: 6.22; 95% CI: 4.33–8.92 when not measured and OR: 16.47; 95% CI: 12.04–22.54 when measured) when he or she was overweight at T2. When conditioned on the previous observation state, percentile birth weight for gestational age and birth cohort were no longer significant at either period. Parity was no longer significant at T3. At T2, Hispanic ethnicity was no longer significantly associated with overweight in the conditioned models but remained significant in the conditioned model at T3 (OR: 1.60; 95% CI: 1.13–2.25). Black children were at higher risk for overweight than white children in the conditioned models at T2 (OR: 1.46; 95% CI: 1.09–1.96) and T3 (OR: 1.59; 95% CI: 1.14–2.23). Maternal smoking during pregnancy at T2 (OR: 1.37; 95% CI: 1.05–1.78) and T3 (OR: 1.71; 95% CI: 1.28–2.31) remained significant and of similar magnitude to that found in the cross-section models. Maternal prepregnancy overweight at T2 (OR: 1.40; 95% CI: 1.07–1.85) and T3 (OR: 2.13; 95% CI: 1.57–2.88) remained significant and of similar magnitude to that found in the cross-section models. Maternal prepregnancy obesity at T2 (OR: 1.64; 95% CI: 1.18–2.28) and T3 (OR: 2.89; 95% CI: 2.02–4.15) remained significant and of similar magnitude to that found in the cross-section models. Gender of the child, maternal SES, and maternal age at birth remained insignificant.


View this table:
[in this window]
[in a new window]
  		TABLE 4. Comparison Between Multivariate Logistic Models of Childhood Overweight Unconditioned and Conditioned at T2 and T3 on Overweight in Previous Time Period

 

   DISCUSSION
TOP
 ABSTRACT
 METHODS
 RESULTS
 DISCUSSION
REFERENCES
 
The findings from this study suggest that a dynamic model can yield insights into the effects of the initial conditions on the development of childhood overweight. Such a model allows separation of the persistent effects from the dynamic effects of the initial conditions. Referring to Fig 1, we found that the variables that had a persistent effect on childhood overweight, ie, those represented by the heavy arrows that affected the probability that the child was overweight between 24 and 47 months, were black race, Hispanic ethnicity, maternal smoking during pregnancy, and maternal obesity before pregnancy.

Estimates from the Markov models of overweight in subsequent periods suggest that the greatest risk for overweight is the child’s having been overweight in the previous observation period. Thus, risk factors for overweight between the ages of 24 and 47 months have a persistent effect consistent with the heavy arrows in the model described by Fig 1. In a separate analysis, we examined the previous weight states for children who were overweight and of normal weight at T3. Two thirds of the children who were overweight at T3 had been overweight during at least 1 previous period, whereas three fourths of the normal-weight children had always been of normal weight. This illustrates a strong tendency for weight states to track across time. Although there is movement across weight states, there is also a pattern established at a very young age that can be used to identify children who are at risk and target specify preventive strategies.

The variables that had dynamic effects in estimates of the Markov models, ie, variables that had a significant effect of weight at T2 and T3 conditional on overweight in the previous period, included black race, Hispanic ethnicity (in T3 only), maternal smoking during pregnancy, maternal prepregnancy overweight, and maternal prepregnancy obesity. These variables increase the probability that a child will transition into overweight or remains overweight conditional on overweight in the previous observation period. These variables correspond to the dynamic effects signified by the dashed arrows in Fig 1.

At T1, the only maternal prepregnancy weight category that had a statistically significant effect on child overweight was maternal obesity. In each of the subsequent periods, the influence of maternal prepregnancy obesity increased substantially. By T2, children of mothers who were overweight before pregnancy were also at greater risk for being overweight. There was a similar time pattern for the effects of maternal overweight in that the magnitude of this risk factor increased in T3. The magnitude of the risk for childhood overweight increased with the weight category of the mother. Conditioning on previous weight resulted in only small changes in the point estimates of the ORs of maternal prepregnancy weight. That the ORs in the cross-sectional models and in the Markov models of transition probabilities have similar magnitudes suggest that the effect of maternal prepregnancy weight not only is the result of an early, persistent propensity to overweight but also affects the dynamics of the process of the development of childhood overweight. Whether these findings result from an intrauterine programming process that affects the dynamics of childhood overweight or that maternal prepregnancy overweight is a proxy for an obesegenic home environment high in caloric foods and not conducive to exercise remains an open question.

Smoking during pregnancy was related to childhood overweight at T1 and had independent effects after conditioning on the weight state during the previous period at T2 and T3. The ORs were similar in the cross-sectional models and the Markov models. These findings add to the growing body of evidence that smoking during pregnancy may be a significant risk factor for overweight development during childhood. However, we do not have data on postnatal maternal smoking and cannot test for an effect of prenatal smoking while controlling for postnatal maternal smoking. A third alternative is that smoking during pregnancy is not itself a risk factor for childhood overweight but is instead a marker for an obesegenic home environment. Thus, it is preliminary to state that smoking may have an effect in utero.

Race and ethnicity were related to childhood overweight at T1 and had independent effects after conditioning on the weight state during the previous period at T2 and T3 for black children. For Hispanic children, ethnicity was significant in the conditioned model only at T3. The influences of race and ethnicity are laid down early in life and operate for black individuals both through early development that persists and an independent risk for development at T2 or T3. Black children are at greater risk for overweight than white children because overweight is more likely to develop at the youngest age, and that has a persistent effect over time. In addition, at older ages, being black is a risk factor for becoming or remaining overweight, even conditional on past weight status. The findings show that at the youngest ages, Hispanic children are more likely than white children to develop overweight. At T2, Hispanic children are more likely to be overweight because of the persistence of overweight from T1, but there is no increased risk for Hispanics to become or remain overweight. By T3, Hispanic individuals are more likely to be overweight as a result of the persistence of previous overweight, and there is also an independent influence of entering or remaining overweight.

We included the birth cohort in our analysis and found that in cross-sectional models, children who were born in later cohorts were more likely to be overweight than those who were born in earlier birth cohorts. In the dynamic models, birth cohort was no longer significant. This raises questions about why children, especially very young children aged 24 to 47 months, who were born in the mid-1990s were more likely to become overweight at very young ages than children who were born 10 years earlier. The rapidity of this change argues against genetics and for environmental conditions. One clear difference is the weight status of women. The prevalence of overweight has been increasing in adults as well as in children. In our sample, the mean prepregnancy weight of the mothers has increased over time (a mean increase of 14 lb, mean weight of 133 lb, BMI = 23 during the period from 1982 through 1986 to a mean weight of 147 lb, mean BMI = 25 a decade later, 1992–1996). The increase in adult BMI has been related to a nutrition transition that occurred in the last 1 or 2 decades of the 20th century. The nutrition transition is characterized by diets high in saturated fats, sugar, and refined foods but low in fiber and on lifestyles characterized by lower levels of activity.63

Another possible contributing factor to this time trend is household changes that have occurred during this period. One of the most notable trends is the rise in maternal employment over that past 30 years as more women have either elected to enter and remain in the workforce or been encouraged to enter the workforce because of welfare reform. From 1970 to 1999, the fraction of women who had children who were younger than 6 years and were in the workforce has doubled, from 30% to 62%. Anderson et al64 found a small association between childhood overweight and the number of maternal hours per week worked over the life of the child, thus providing some evidence for an employment effect, but it does not explain a large portion of this time trend difference, and more needs to be done to understand why young children are at greater risk for overweight now as compared with a decade ago.

A limitation of the study is that height and weight were not measured for all children. Results from the Markov model at T3 suggested that measurement error was a problem because of the large difference in the ORs on weight status at T2 depending on whether it was recorded as being measured. A second major limitation of the study is that the study sample is not representative of the eligible sample. There are more white individuals and fewer minorities in the study sample than in the eligible population. The use of race and ethnic control variables in the statistical analyses ameliorates but does not entirely solve this problem. A more serious sample selection problem is that the prevalence of overweight in black children aged 72 to 95 months is lower in the study sample than in the eligible population. This leads to an underestimation of the effect of black race in both the cross-section and Markov models for children aged 72 to 95 months. The prevalence of overweight among Hispanic children is lower in the study sample than in the eligible population for both ages 48 to 71 months and 72 to 95 months. This leads to an underestimation of the effect of Hispanic ethnicity in the cross-section models for these 2 age groups. It leads to an underestimation of the effect of Hispanic ethnicity in the Markov model for ages 48 to 71 months and has an indeterminate effect on the estimate of Hispanic ethnicity in the Markov model for ages 72 to 95 months. This study is limited further because we cannot shed light on the specific biological mechanisms whereby these initial conditions affect the dynamics of childhood overweight. Other limitations include lack of information on duration of breastfeeding, which may explain the insignificance of the breastfeeding variable in the childhood overweight equations, and mother reports of birth weight and length of gestation.

These findings are clinically relevant in that they suggest 3 areas of emphasis. First, they highlight the importance of very early monitoring (before the child’s second birthday) of healthy weight maintenance. Because weight states have persistence across time, a child who reaches his or her second birthday at a healthy weight is less likely to become overweight at a later age. For children who are already overweight at these young ages, interventions should begin immediately. Women who are overweight or obese and considering becoming pregnant should be counseled to achieve a healthy weight before becoming pregnant. Women who smoke should also be counseled to quit as these results provide additional evidence that smoking during pregnancy may be harmful to the fetus. The mechanisms by which maternal smoking during pregnancy and prepregnancy obesity may influence child weight states are not fully understood. These risk factors may be operating through biological processes and/or as markers for obesegenic environments. From a clinical perspective, however, these findings suggest factors that can be used to identify children who are at high risk for the development of childhood overweight at very young ages, thus providing an opportunity to target intensive preventive strategies before the establishment of an unhealthy weight pattern.


   ACKNOWLEDGMENTS
 
This research was funded by National Institutes of Health, National Institute of Nursing Research grant RO1 NR008512.


   FOOTNOTES
 
Accepted Feb 7, 2005.

Reprint requests to (P.S.) College of Nursing, Ohio State University, 1585 Neil Ave, Columbus, OH 43210. E-mail: salsberry.1{at}osu.edu

No conflict of interest declared.


   REFERENCES
TOP
 ABSTRACT
 METHODS
 RESULTS
 DISCUSSION
 REFERENCES
 

  1. Kuh D, Davey Smith G. The life course and adult chronic disease: an historical perspective with particular reference to coronary heart disease. In: Kuh D, Ben-Shlomo Y, eds. A Life Course Approach to Chronic Disease Epidemiology. 2nd ed. Oxford, United Kingdom: Oxford University Press; 2004:15–37
  2. Michels KB. Early life predictors of chronic disease. J Womens Health (Larchmt). 2003;12 :157 –161
  3. Marmot M, Shipley M, Brunner E, Hemingway H. Relative contribution of early life and adult socioeconomic factors to adult morbidity in the Whitehall II study. J Epidemiol Community Health. 2001;55 :301 –307[Abstract/Free Full Text]
  4. Ben-Shlomo Y, Kuh D. A life course approach to chronic disease epidemiology: conceptual models, empirical challenges and interdisciplinary perspectives. Int J Epidemiol. 2002;31 :285 –293[Free Full Text]
  5. Halfon N, Hochstein M. Life course health development: an integrated framework for developing health, policy and research. Milbank Q. 2002;80 :1 –33
  6. Harper S, Lynch J, Hsu WL, et al. Life course socioeconomic conditions and adult psychosocial functioning. Int J Epidemiol. 2002;31 :395 –403[Abstract/Free Full Text]
  7. Langenberg C, Hardy R, Kuh D, Brunner E, Wadsworth M. Central and total obesity in middle aged men and women in relation to lifetime socioeconomic status: evidence from a national birth cohort. J Epidemiol Community Health. 2003;57 :816 –822[Abstract/Free Full Text]
  8. Parsons TJ, Power C, Logan S, Summerbell CD. Childhood predictors of adult obesity: a systematic review. Int J Obes Relat Metab Disord. 1999;23(suppl 8) :S1 –S107
  9. Eriksson J, Forsen T, Osmond C, Barker D. Obesity from cradle to grave. Int J Obes Relat Metab Disord. 2003;27 :722 –727[CrossRef][ISI][Medline]
  10. Strauss RS. Effects of the intrauterine environment on childhood growth. Br Med Bull. 1997;53 :81 –95[Abstract/Free Full Text]
  11. Whitaker RC. Predicting preschooler obesity at birth: the role of maternal obesity in early pregnancy. Pediatrics. 2004;114 (1). Available at: www.pediatrics.org/cgi/content/full/114/1/e29
  12. Dietz WH, Gortmaker SL. Preventing obesity in children and adolescents. Annu Rev Public Health. 2001;22 :337 –353[CrossRef][ISI][Medline]
  13. Harding J. Nutrition and fetal growth. Reprod Fertil Dev. 1995;7 :539 –547[CrossRef][Medline]
  14. Dabelea D, Hanson R, Lindsay R, et al. Intrauterine exposure to diabetes conveys risks for type 2 diabetes and obesity: a study of discordant sibships. Diabetes. 2000;49 :2208[Abstract]
  15. Godfrey KM, Barker DJ, Robinson TN, Osmond C. Mother’s birth weight and diet in pregnancy in relation to the baby’s thinness at birth. Br J Obstet Gynaecol. 1997;104 :663 –667[ISI][Medline]
  16. Moore VM, Davies MJ, Willson KJ, Worsley A, Robinson JS. Dietary composition of pregnant women is related to size of the baby at birth. J Nutr. 2004;134 :1820 –1826[Abstract/Free Full Text]
  17. Godfrey K, Robinson S. Maternal nutrition, placental growth and fetal programming. Proc Nutr Soc. 1998;57 :105 –111[CrossRef][Medline]
  18. Godfrey K, Robinson S, Barker DJ, Osmond C, Cox V. Maternal nutrition in early and late pregnancy in relation to placental and fetal growth. BMJ. 1996;312 :410 –414[Abstract/Free Full Text]
  19. Yajnik CS. The lifecycle effects of nutrition and body size on adult adiposity, diabetes and cardiovascular disease. Obes Rev. 2002;3 :217 –224[CrossRef][Medline]
  20. Harding J. The nutritional basis of the fetal origins of adult disease. Int J Epidemiol. 2001;30 :15 –23[Free Full Text]
  21. Ravelli G, Stein Z, Susser M. Obesity in young men after famine exposure in utero and early infancy. N Engl J Med. 1976;7 :349 –354
  22. Ravelli A, van Der Meulen J, Osmond C, Barker D, Bleker O. Obesity at the age of 50 y in men and women exposed to famine prenatally. Am J Clin Nutr. 1999;70 :811 –816[Abstract/Free Full Text]
  23. Nold JL, Georgieff MK. Infants of diabetic mothers. Pediatr Clin North Am. 2004;51 :619 –637, viii[ISI][Medline]
  24. Gillman MW, Rifas-Shiman S, Berkey CS, Field AE, Colditz GA. Maternal gestational diabetes, birth weight, and adolescent obesity. Pediatrics. 2003;111 (3). Available at: www.pediatrics.org/cgi/content/full/111/3/e221
  25. Glueck CJ, Goldenberg N, Streicher P, Wang P. The contentious nature of gestational diabetes: diet, insulin, glyburide and metformin. Expert Opin Pharmacother. 2002;3 :1557 –1568[CrossRef][Medline]
  26. Malee MP, Verma A, Messerlian G, Tucker R, Vohr BR. Association between maternal and child leptin levels 9 years after pregnancy complicated by gestational diabetes. Horm Metab Res. 2002;34 :212 –216[CrossRef][Medline]
  27. Cho NH, Silverman BL, Rizzo TA, Metzger BE. Correlations between the intrauterine metabolic environment and blood pressure in adolescent offspring of diabetic mothers. J Pediatr. 2000;136 :587 –592[CrossRef][ISI][Medline]
  28. Whitaker RC, Pepe MS, Seidel KD, Wright A, Knopp R. Gestational diabetes and the risk of offspring obesity. Pediatrics. 1998;101 :e9 . Available at: www.pediatrics.org/cgi/content/full/101/2/e9[Abstract/Free Full Text]
  29. Whitaker RC, Dietz W. Role of the prenatal environment in the development of obesity. J Pediatr. 1998;132 :768 –776[CrossRef][ISI][Medline]
  30. Petry C, Ozanne E, Hales C. Programming of intermediary metabolism. Mol Cell Endorcrinol. 2001;185 :81 –91
  31. Phillips D. Relation of fetal growth to adult muscle mass. Diabet Med. 1995;12 :686 –690[ISI][Medline]
  32. Whitaker RC, Wright JA, Pepe MS, Seidel KD, Dietz WH. Predicting obesity in young adulthood from childhood and parental obesity. N Engl J Med. 1997;337 :869 –873[Abstract/Free Full Text]
  33. Agras W, Hammer L, McNicholas F, Kraemer HC. Risk factors for childhood overweight: a prospective study from birth to 9.5 years. J Pediatr. 2004;145 :20 –25[CrossRef][ISI][Medline]
  34. Magarey AM, Daniels LA, Boulton TJ, Cockington RA. Predicting obesity in early adulthood from childhood and parental obesity. Int J Obes Relat Metab Disord. 2003;27 :505 –513[CrossRef][ISI][Medline]
  35. Kramer MS, Barr RG, Leduc DG, Boisjoly C, Pless IB. Infant determinants of childhood weight and adiposity. J Pediatr. 1985;107 :104 –107[CrossRef][ISI][Medline]
  36. Maffeis C, Micciolo R, Must A, Zaffanello M, Pinelli L. Parental and perinatal factors associated with childhood obesity in north-east Italy. Int J Obes Relat Metab Disord. 1994;18 :301 –305[ISI][Medline]
  37. Guillaume M, Lapidus L, Beckers F, Lambert A, Bjorntorp P. Familial trends of obesity through three generations: the Belgian-Luxembourg child study. Int J Obes Relat Metab Disord. 1995;19 (suppl 3):S5–S9
  38. Danielzik S, Czerwinski-Mast M, Langnase K, Dilba B, Muller MJ. Parental overweight, socioeconomic status and high birth weight are the major determinants of overweight and obesity in 5–7 y-old children: baseline data of the Kiel Obesity Prevention Study (KOPS). Int J Obes Relat Metab Disord. 2004;28 :1494 –1502[CrossRef][ISI][Medline]
  39. Pollack H, Lantz P, Frohna J. Maternal smoking and adverse birth outcomes among singletons and twins. Am J Public Health. 2000;90 :395 –400[Abstract/Free Full Text]
  40. von Kries R, Toschke A, Koletzko B, Slikker W. Maternal smoking during pregnancy and childhood obesity. Am J Epidemiol. 2002;156 :954 –961[Abstract/Free Full Text]
  41. Vik T, Jacobsen G, Vatten L, Bakketeig L. Pre and post-natal growth in children of women who smoked in pregnancy. Early Hum Dev. 1996;45 :245 –255[CrossRef][ISI][Medline]
  42. Montgomery SM, Ekbom A. Smoking during pregnancy and diabetes mellitus in a British longitudinal birth cohort. BMJ. 2002;324 :26 –27[Free Full Text]
  43. Conter V, Cortinovis I, Rogari P, Riva L. Weight growth in infants born to mothers who smoked during pregnancy. BMJ. 1995;310 :768 –771[Abstract/Free Full Text]
  44. Toschke AM, Koletzko B, Slikker W Jr, Hermann M, von Kries R. Childhood obesity is associated with maternal smoking in pregnancy. Eur J Pediatr. 2002;161 :445 –448[CrossRef][ISI][Medline]
  45. Toschke AM, Grote V, Koletzko B, von Kries R. Identifying children at high risk for overweight at school entry by weight gain during the first 2 years. Arch Pediatr Adolesc Med. 2004;158 :449 –452[Abstract/Free Full Text]
  46. Power C, Jefferis B. Fetal environment and subsequent obesity: a study of maternal smoking. Int J Epidemiol. 2002;31 :413 –419[Abstract/Free Full Text]
  47. Oncken C, Hardardottir H, Smeltzer J. Human studies of nicotine replacement during pregnancy. In: Benowitz N, ed. Nicotine Safety and Toxicity. Oxford, United Kingdom: Oxford University Press; 1998:107–118
  48. Oken E, Gillman MW. Fetal origins of obesity. Obes Res. 2003;11 :496 –506[ISI][Medline]
  49. Barker M, Robinson S, Osmond C, Barker DJ. Birth weight and body fat distribution in adolescent girls. Arch Dis Child. 1997;77 :381 –383[Abstract/Free Full Text]
  50. Singhal A, Wells J, Cole T, Fewtrell M, Lucas A. Programming of lean body mass: a link between birth weight, obesity, and cardiovascular disease? Am J Clin Nutr. 2003;77 :726 –730[Abstract/Free Full Text]
  51. Seidman DS, Laor A, Gale R, Stevenson DK, Danon YL. A longitudinal study of birth weight and being overweight in late adolescence. Am J Dis Child. 1991;145 :782 –785[Abstract]
  52. Parsons TJ, Power C, Manor O. Fetal and early life growth and body mass index from birth to early adulthood in 1958 British cohort: longitudinal study. BMJ. 2001;323 :1331 –1335[Abstract/Free Full Text]
  53. Hediger ML, Overpeck MD, Kuczmarski RJ, Ruan WJ. Association between infant breastfeeding and overweight in young children. JAMA. 2004;285 :2453 –2460
  54. Bergman KE, Bergmann RL, von Kries R, Bohm O, Dudenhausen JW, Wahn U. Early determinants of childhood overweight and adiposity in a birth cohort study: role of breast feeding. Int J Obes Relat Metab Disord. 2003;27 :162 –172[CrossRef][ISI][Medline]
  55. Grummer-Strawn LM, Mei Z. Does breastfeeding protect against pediatric overweight: analysis of longitudinal data from the Centers for Disease Control and Prevention Pediatric Nutrition Surveillance System. Pediatrics. 2004;113 (3). Available at: www.pediatrics.org/cgi/content/full/113/3/e81
  56. Li L, Parsons TJ, Power C. Breast feeding and obesity in childhood: cross sectional study. BMJ. 2003;327 :327 –328
  57. Gillman M, Rifas-Shiman S, Camargo CJ, Berkey C, Frazier A, Rockett H. Risk of overweight among adolescents who had been breast fed as infants. JAMA. 2001;285 :2461 –2467[Abstract/Free Full Text]
  58. von Kries R, Koletzko B, Sauerwald T, von Mutius E, Barnert D, Grunert V. Breast feeding and obesity: cross sectional study. BMJ. 1999;319 :147 –150[Abstract/Free Full Text]
  59. Heckman J. Statistical models for discrete panel data. In: Manski CF, McFadden D, eds. Structural Analysis of Discrete Data with Econometric Applications. Cambridge, MA: MIT Press; 1981:179–195
  60. Culhane J, Raub V, McCollum K, Elo I, Hogan V. Exposure to chronic stress and ethnic differences in rates of bacterial vaginosis among pregnant women. Am J Obstet Gynecol. 2002;187 :1272 –1276[CrossRef][ISI][Medline]
  61. Kuczmarski RJ, Ogden CL, Guo S. CDC Growth Charts for the United States: Methods and Development. Hyattsville, MD: National Center for Health Statistics; 2002
  62. Oken E, Kleinman KP, Rich-Edwards J, Gillman MW. A nearly continuous measure of birth weight for gestational age using a United States national reference. BMC Pediatr. 2003;3 :6[CrossRef][Medline]
  63. Popkin BM, Gordon-Larsen P. The nutrition transition: worldwide obesity dynamics and their determinants. Int J Obes Relat Metab Disord. 2004;28 (suppl 3):S2–S9
  64. Anderson PM, Butcher KR, Levine PB. Maternal employment and overweight children. J Health Econ. 2003;22 :477 –504[CrossRef][ISI][Medline]
PEDIATRICS (ISSN 1098-4275). ©2005 by the American Academy of Pediatrics



This article has been cited by other articles:


Home page
 Diabetes CareHome page
D. Dabelea, E. J. Mayer-Davis, A. P. Lamichhane, R. B. D'Agostino Jr., A. D. Liese, K. S. Vehik, K.M. V. Narayan, P. Zeitler, and R. F. Hamman
Association of Intrauterine Exposure to Maternal Diabetes and Obesity With Type 2 Diabetes in Youth: The SEARCH Case-Control Study
Diabetes Care, July 1, 2008; 31(7): 1422 - 1426.
[Abstract] [Full Text] [PDF]

Home page
 Am. J. Clin. Nutr.Home page
M. A Mendez, M. Torrent, C. Ferrer, N. Ribas-Fito, and J. Sunyer
Maternal smoking very early in pregnancy is related to child overweight at age 5-7 y
Am. J. Clinical Nutrition, June 1, 2008; 87(6): 1906 - 1913.
[Abstract] [Full Text] [PDF]

Home page
 Int J EpidemiolHome page
R. von Kries, G. Bolte, L. Baghi, A. M. Toschke, and for the GME Study Group
Parental smoking and childhood obesity--is maternal smoking in pregnancy the critical exposure?
Int. J. Epidemiol., February 1, 2008; 37(1): 210 - 216.
[Abstract] [Full Text] [PDF]

Home page
 Am. J. Roentgenol.Home page
A. K. Choudhary, L. F. Donnelly, J. M. Racadio, and J. L. Strife
Diseases Associated with Childhood Obesity
Am. J. Roentgenol., April 1, 2007; 188(4): 1118 - 1130.
[Abstract] [Full Text] [PDF]

Home page
 Exp PhysiolHome page
P. D. Taylor and L. Poston
Developmental programming of obesity in mammals
Exp Physiol, March 1, 2007; 92(2): 287 - 298.
[Abstract] [Full Text] [PDF]

This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Right arrow P3Rs: Submit a response
Right arrow Alert me when this article is cited
Right arrow Alert me when P3Rs are posted
Right arrow Alert me if a correction is posted
Right arrow Citation Map
Services
Right arrow E-mail this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Add to My File Cabinet
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via CrossRef
Right arrow Citing Articles via ISI Web of Science (23)
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Salsberry, P. J.