点击显示 收起
1 From the Division of Gastroenterology and Nutrition, The Children’s Hospital of Philadelphia (NS, BSZ, and VAS), the Center for Clinical Epidemiology and Biostatistics (NS and SKK), and the WM Krogman Center for Research in Child Growth and Development (SHK), University of Pennsylvania School of Medicine, Philadelphia.
2 The National Collaborative Perinatal Project was supported by the National Institute of Neurologic Disease and Stroke. The Philadelphia Blood Pressure Project was supported by the National Institutes of Health. The present study was supported by the Nutrition Center of The Children’s Hospital of Philadelphia. NS is supported in part by NIH grant K23 RR16073.
3 Address reprint requests to N Stettler, Division of Gastroenterology and Nutrition, North 1559, The Children’s Hospital of Philadelphia, 34th Street and Civic Center Boulevard, Philadelphia, PA 19104-4399. E-mail: nstettle{at}cceb.med.upenn.edu.
See corresponding editorial on page 1350.
ABSTRACT
Background: Obesity is increasing in the United States. Evidence-based prevention is a public health priority and should target well-defined risk factors and critical periods.
Objective: We tested the hypothesis that rapid weight gain during early infancy is associated with obesity in African American young adults, a group at increased risk of obesity.
Design: A cohort of 300 African Americans born at full term was followed from birth to 20 y of age. A pattern of rapid weight gain was defined as an increase in weight-for-age ≥ 1 SD between birth and 4 mo.
Results: About 29% of subjects had a pattern of rapid weight gain during infancy; 8% were obese [body mass index (in kg/m2) ≥ 30] at age 20 y. One-third of the obesity at age 20 y could be attributed to rapid weight gain in the first 4 mo of life. After adjustment for confounding factors, subjects with rapid weight gain during early infancy were more likely to become obese at age 20 y (odds ratio = 5.22; 95% CI: 1.55, 17.6; P = 0.008). The results were confirmed by using a combination of body mass index and skinfold thickness (odds ratio = 6.72; 95% CI: 1.93, 23.4; P = 0.003).
Conclusions: The results of the present study provide evidence that a pattern of rapid weight gain during early infancy is associated with obesity not only in childhood but also in young adulthood. We propose that early infancy constitutes a critical period for the development of obesity. Mechanisms of action and prevention strategies require further investigation.
Key Words: Birth order • birth weight • African Americans • children • cohort studies • growth • infants • obesity • skinfold thickness • weight gain
INTRODUCTION
As in many other countries, in the United States, the prevalence of obesity is increasing among both adults (1) and children, particularly African American children (2, 3). The complications of obesity, such as heart disease and type 2 diabetes, are common causes of morbidity and account for a large portion of preventable deaths and health care costs in the United States, particularly in the African American minority (4–7). Therefore, in addition to treatment, prevention of obesity has become a global public health priority and may be one way to reduce health disparities in the United States (8).
To prevent obesity, both the risk factors and the critical periods for the development of obesity must be clearly identified. Childhood obesity often tracks into adulthood, especially in children of obese parents (9–12). Fetal life, the period of adiposity rebound (age 4–7 y), and adolescence have been identified as potentially critical periods for the development of obesity (13). In addition, a rapid rate of weight gain during infancy has been associated with obesity in childhood (14–16). It is unknown whether rapid weight gain during infancy is also associated with adult obesity in humans, although it has been shown in animal models (17, 18). The aim of this study was to determine whether a pattern of rapid weight gain during early infancy is a risk factor for the development of obesity in African American young adults, a group known to be at high risk of obesity.
SUBJECTS AND METHODS
Study design and subjects
The association of a rapid weight gain pattern during the first 4 mo of life with obesity in young adulthood was assessed through the retrospective analysis of a prospective cohort study. The National Collaborative Perinatal Project was a multicenter cohort study initiated in 1959 to investigate risk factors for cerebral palsy at 12 US sites. In Philadelphia, between 1959 and 1966, 9020 African American women with singleton pregnancies obtaining prenatal care at Pennsylvania Hospital were enrolled in the study after giving informed consent. Data were collected on the children on several occasions, including at ages 4 mo, 12 mo, and 7 y. Among the 74% of subjects who completed the 7-y assessment in Philadelphia, a sample of 446 African American subjects representative of the subjects who completed the National Collaborative Perinatal Project and born at full-term gestation between 1962 and 1966, when gestational age was routinely assessed, was enrolled in a 1977 follow-up study: the Philadelphia Blood Pressure Project. The 300 subjects measured between the ages of 18.0 and 22.9 y constitute the present study sample; the other 146 were lost to follow-up. The sample was restricted to subjects with birth documented to be at full term (gestational age 37–42 wk) to eliminate potential confounding by differences in patterns of weight gain in infants born prematurely or postterm (19). The study was approved by the Institutional Review Board of The Children’s Hospital of Philadelphia.
Measurements
Maternal variables were those with previously reported or theoretical influence on the offspring’s weight status. Maternal prepregnancy weight, age, smoking status, and education were obtained by interview at enrollment (20). Maternal height measured at time of delivery was used to calculate prepregnancy body mass index (BMI) as prepregnancy weight in kg divided by height in meters squared.
The subjects’ variables of interest included sex, gestational age at birth, birth order, year of birth, weight and length (or height) at each assessment, exact age, and triceps and subscapular skinfold thicknesses in young adulthood, measured as previously described (20–22). According to the study protocol, gestational age was determined by "the senior physician assuming responsibility for the delivery using all clinical and historical information." The subjects were considered as firstborn when the mother reported "no children living in the household" at enrollment. The method of infant feeding was also of interest, but only the initial (first few days) feeding method was documented in a subset of 50 subjects. Because only 3 mothers initiated breastfeeding, this variable could not be used in this analysis.
Analytic approach
Weight-for-age z scores were calculated as recommended by the Centers for Disease Control and Prevention by using the LMS method and a representative sample of the US population (23). The main exposure, a pattern of rapid rate of weight gain during the first 4 mo of life, was defined as an increase in weight-for-age z score ≥ 1 SD between birth and 4 mo. By definition, an average weight gain would correspond to a change of 0 SD. The main outcome, obesity in young adulthood, was defined by using the standard definition of BMI ≥ 30, as was maternal obesity (24). The analysis was then repeated with the use of an alternate definition requiring a subject to have both a BMI ≥ 25 (overweight) and a sum of the triceps and subscapular skinfold thicknesses ≥ 85th percentile (overfat) (25). This combined definition allowed the identification of subjects who were both overweight and overfat, some of whom were also obese on the basis of the definition of BMI ≥ 30. Overweight status in childhood (age 7 y) was defined as a BMI ≥ 95th percentile of a US reference population (23). A continuous outcome, such as the BMI value, was not used because most of the variability in the outcome would be within the healthy range of weight.
The analysis was conducted by using STATA 6.0 (26). Most variables were skewed and therefore are described by using medians, percentiles, or proportions. Chi-square or Wilcoxon’s rank-sum tests were used to compare subjects with complete and incomplete follow-up. The unadjusted associations were estimated by simple logistic regression or chi-square and are expressed as odds ratios (ORs) and 95% CIs. Because adulthood obesity was relatively infrequent, the OR is an appropriate estimate of relative risk. The population attributable risk percent was calculated as the difference in prevalence between the exposed and the unexposed groups divided by the prevalence in the exposed group, multiplied by 100 (27). Possible interactions were explored for variables with previously reported or theoretical influence by using stratified analyses (Mantel-Haenszel test) or logistic regression with an interaction term. Finally, multiple logistic regressions were used to adjust for potentially confounding variables. In a post hoc analysis, the multiple logistic regression models were repeated with additional adjustment for overweight status at age 7 y. All statistical tests were two-tailed and P values < 0.05 were considered statistically significant.
RESULTS
The 300 study subjects were not significantly different from the other 4008 subjects eligible at birth in sex distribution, birth weight, gestational age, firstborn status, maternal prepregnancy BMI, or maternal prepregnancy obesity. However, they were born later in the study (P < 0.001), had older mothers (
View this table:
TABLE 1 . Characteristics of the subjects at infancy and in young adulthood and of their mothers at enrollment
The unadjusted associations of young adulthood obese or overweight-overfat status with rapid early infancy weight gain and potential confounding variables are presented in Table 2. No significant interaction in the association between a rapid early infancy weight gain and adult obesity or overweight-overfat status was detected by sex, birth weight, gestational age, firstborn status, birth year, maternal BMI, maternal age, maternal smoking status, or maternal education. The unadjusted association between a rapid weight gain during early infancy and obesity at age 20 y was confirmed after adjustment for potentially confounding variables.
View this table:
TABLE 2 . Unadjusted and adjusted associations between the subjects’ characteristics and their obese or overweight-overfat status in young adulthood1
In a post hoc analysis, we investigated whether the association between early infancy weight gain and adult obesity was already present in childhood and was followed by persistence of overweight status from childhood to adulthood. Among the 12 subjects with a rapid weight gain during early infancy who became obese adults, 6 were overweight at age 7 y and 6 were not. Similarly, among the 13 subjects with a rapid weight gain during early infancy who were overweight-overfat in adulthood, 5 were overweight at age 7 y and 8 were not. In the multivariate analyses, with adjustment for overweight status at age 7 y by multiple logistic regression, there was a trend toward rapid weight gain during early infancy being associated with adult obesity (OR = 3.39; 95% CI: 0.94, 12.2; P = 0.063; n = 213). With use of the combined definition, overweight-overfat status in young adulthood remained significantly associated with early infancy weight gain, independently of childhood overweight status (OR = 4.92; 95% CI: 1.32, 18.3; P = 0.017; n = 213). As expected, overweight status in childhood (age 7 y) was associated with adult (age 18.0–22.9 y) obesity (OR = 11.4; 95% CI: 2.23, 57.6; P = 0.003) and overweightoverfat status (OR = 8.44; 95% CI: 1.48, 48.0; P = 0.016), reflecting the persistence of overweight from childhood to adulthood.
DISCUSSION
The present study establishes, for the first time to our knowledge, that a rapid weight gain pattern from birth to the age of 4 mo is associated with obesity in young adulthood. This suggests that, in addition to fetal life, the period of adiposity rebound, and adolescence (13), early infancy also constitutes a critical period for the development of adulthood obesity. In this sample of African American subjects born at full term, about one-half of the obese young adults experienced a rapid weight gain during early infancy. With the use of alternate definitions based on BMI or BMI combined with skinfold thickness, a more direct measure of adiposity, and after adjustment for confounding factors, strong associations were confirmed between early infancy weight gain and obesity 20 y later. In our post hoc analyses, only part of this association was explained by the previously described association of rapid weight gain during infancy with overweight status at age 7 y (16), followed by persistence of overweight status from childhood to adulthood. Because of the large CIs, however, the magnitude of these independent associations is difficult to fully evaluate by using the present data.
Previous studies have linked rapid weight gain during infancy to childhood overweight status (14–16). Other studies have described the tracking of overweight from childhood to adulthood or, to a lesser extent, from infancy to adulthood (9, 11, 12), but these earlier studies were based on weight status at one point in time and did not assess weight change in individual subjects during critical periods. For example, in the earlier analysis of a sample of African American subjects overlapping with the present study sample, Mack and Johnston (10) described a significant association between relative weight at the age of 12 mo and obesity at the age of 9–14 y. The observed association between early infancy weight gain and young adulthood obesity does not exclude the possibility that weight gain during other periods also may be critical for the development of obesity (13). For example, the present study confirms the previously described association between birth weight and young adulthood obesity (9, 13, 14, 16), even after adjustment for maternal BMI. It should be noted, however, that because birth weight and early weight gain are expressed with different units, the ORs and effect sizes are not directly comparable between the 2 variables. Interestingly, birth weight was not significantly associated with overweight-overfat status when the definition combining BMI and skinfold thickness was used.
If our results showing an association between early infancy growth and young adulthood obesity are confirmed in other populations, they will support new etiologic hypotheses about critical periods or pathways for obesity development and, therefore, new hypotheses about how to prevent this prevalent health problem. Some hypotheses about possible mechanisms have already been suggested on the basis of previous studies and experimental models (28, 29). Studies of primate models (30) did not confirm the initial hypothesis proposed in the 1970s of adipocyte hyperplasia suggested by studies in rodent models (17). More recently, studies have shown life-long changes in the appetite regulating centers and in the insulin secretion of rats overfed for short periods after birth (18, 31). Formula feeding has been associated with more rapid weight gain than breastfeeding (32) and with an increased risk of childhood and adolescent obesity (33), but no data are available on the persistence of this association into adulthood. A pattern of rapid weight gain during infancy may also reflect the early expression of a genetic predisposition to excessive weight, rather than a causal link.
Although the population attributable risk for obesity related to rapid weight gain during early infancy was 30%, no inference can be made from this observational study about causality. Therefore, before any recommendations are given, interventions aimed at altering weight gain in early infancy should be carefully studied for safety and their potential ability to alter the long-term development of obesity. The American Academy of Pediatrics (34) recommends exclusive breastfeeding during the first 6 mo of life, a practice generally associated with a slower rate of weight gain and possibly a decreased risk of overweight in childhood and adolescence. At present, however, there are no known safe and effective interventions in early infancy to prevent adulthood obesity.
Early infancy may be a particularly important period for the development of obesity in African Americans. Because birth-weight-for-gestational-age is lower in African Americans than in European Americans (35), a pattern of rapid weight gain during infancy, as catch-up growth, may be more frequent in African American infants. Additionally, African American infants are less frequently breastfed than are European American infants (8), possibly further increasing their risk for rapid weight gain during infancy and obesity development (32, 33). In 1998, 45% of African American mothers were breastfeeding in early postpartum, compared with 68% of white mothers (8). Note also that the prevalence of obesity in the young African American adults of the present study (8.0%) and their mothers (8.2%) was considerably lower than the prevalence observed today in African American adults (39.9%) and women of childbearing age (46.2%) (1). Therefore, the public health impact of our finding may be more important today than was observed in the present study.
Our study has some limitations. The relatively small sample size resulted in large CIs and relatively unstable OR estimates. However, the main association of interest was consistent in direction and, we believe, provides valuable information for further testing of this hypothesis in larger and more diverse cohorts. Although the study subjects were similar to the potentially eligible subjects for most variables, they were less likely to have a rapid weight gain during early infancy. The reason for this difference is unclear, but may be related to a lower participation of obese subjects. Such a difference would, however, introduce a bias only if the reported associations were different between subjects with or without complete follow-up. We believe that this is unlikely to be the case. Because the infant feeding mode was not adequately recorded in the present study, its potential role as a confounding factor could not be assessed. Furthermore, the present data do not allow us to test whether the association of a rapid rate of weight gain during early infancy with obesity in young adulthood is limited to formula-fed subjects, as suggested by a similar study in European American children (36). On the basis of the usual feeding practice at the time of the study, it can be assumed that the study subjects were mostly formula-fed, but, because the formulas used at the time of the study differ considerably in composition from formulas used now, our results may not be generalizable to infants who are formula-fed today. Environmental and behavioral factors between infancy and young adulthood (patterns of television viewing, physical activity, and dietary intake) were not assessed in the present study and may be potential confounding factors. Finally, the association described in this group of low-income African Americans born in Philadelphia in the 1960s may not be generalizable to other populations, and further testing of this hypothesis is needed to confirm our findings.
Our study also has unique strengths. Unlike similar studies, these data were collected prospectively as part of a research protocol in which growth was an important variable, thus markedly decreasing the risks for recall bias or unreliable measurements. Furthermore, skinfold thicknesses were measured with a high degree of reliability by experienced research anthropometrists (21), allowing for the assessment of adiposity in addition to overweight status as indicated by BMI.
In conclusion, the results of the present study support the hypothesis that early infancy constitutes a critical period for the development of obesity in African American young adults born at full term. This study suggests new hypotheses about the long-term consequences of early growth patterns on adult obesity and related health complications in the context of the lifecourse approach to chronic disease (37). These results may also provide new approaches to identifying persons early in life who are at increased risk of obesity. These are necessary steps for the development of safe and effective strategies to prevent obesity in childhood and adulthood.
ACKNOWLEDGMENTS
We thank James R Coleman for his support in accessing and managing the Collaborative Perinatal Project data in Philadelphia.
NS participated in the design of the present retrospective study, in the analysis and interpretation of the data, and in the writing of the manuscript. SKK participated in the analysis and interpretation of the data and in the writing of the manuscript. SHK participated in the design of the study, in the data collection, in the interpretation of the data, and in the writing of the manuscript. BSZ participated in the design of the present retrospective study, in the data collection, in the interpretation of the data, and in the writing of the manuscript. VAS participated in the design of the present retrospective study, in the interpretation of the data, and in the writing of the manuscript. None of the authors had a conflict of interest to declare.
REFERENCES
Related articles in AJCN: