This overview is intended to provide an integrated and interpretive view of the material presented at the conference on Preventing Childhood Obesity. It also expresses the editors’ viewpoints and their recommendations and caveats. The overview was developed after examination and comparison of all of the speakers’ presentations and materials and consideration of comments made during the question-and-answer periods that followed each session. Material from the literature that addressed some unresolved issues raised during the conference is included. The overview follows a chronological sequence from intrauterine life to adult life and integrates, within each section, material from any of the presentations that touched on that topic. Research needs, public health messages, and caveats that were suggested to the editors by the conference presentations and discussions are also included.
Several speakers presented data showing the rapid increases in the rates of obesity and overweight that have occurred in the past few decades in the United States, as well as other countries. Obesity and overweight are increasingly prevalent among adults, but similar problems are affecting progressively younger children. Overweight among children 6 to 11 years of age in the United States increased from ∼4% in the early 1970s to 15% in 1999–2000. The speakers explored changes that have occurred in the past few decades that are likely to have contributed to the increasing rates of obesity.
The prenatal period, infancy, and early childhood may be stages of particular vulnerability to obesity development, because they are unique periods for cellular differentiation and development. This unique vulnerability might make it possible for actions taken at these stages to determine the future course of adiposity. Two speakers focused on whether in utero factors might affect later obesity development. The evidence presented by Dr Gillman indicated that infants with low birth weight might be especially vulnerable to excess fatness in adulthood at any attained body mass index (BMI). Data he presented indicated that low-birth-weight newborns were at increased risk of developing central obesity and the disease conditions associated with it.
Several studies examined the effects of high birth weight on later health outcomes, with attained adult BMI being statistically controlled, and their findings are intriguing. For example, Dr Gillman presented data from the Nurses’ Health Study that showed that, when attained BMI was controlled, the risk of cardiovascular disease, stroke, hypertension, and type 2 diabetes mellitus all decreased with increasing birth weight, through the range of birth weights from <5.0 to >10.0 lb. Two studies1,2 showed that birth weight was also inversely related to insulin resistance, when adjustment was made for attained BMI.
Dr Kramer also considered the effect of high birth weight, but without control for attained BMI. He showed that high-birth-weight newborns remained above mean weight in later childhood and that women were >1.8 times more likely to be overweight in early pregnancy if they had been large for gestational age (LGA) (generally >90th percentile of weight for gestational age) at birth. It would be useful to have data on health outcomes other than weight that are usually associated with weight for such women.
The 2 speakers did not resolve the question of whether controlling for attained BMI improves or confounds the interpretation of the risk of high birth weight. High birth weight is associated with an increased risk of high weight or high BMI later in life. However, the finding that, at any given adult BMI, health outcomes are better among those born with high birth weight is worth understanding. Studies suggest that infants born LGA have a greater proportion of lean tissue relative to body fat, whereas small for gestational age (SGA) infants are disproportionately high in fat.3 At 47 months, the weight status of LGA infants was +0.50 SD and length and head circumference were +0.43 SD above values for the reference population.3 A recent report showed that a 1-SD increase in birth weight was associated with a 0.9- to 1.4-kg increase in fat-free mass among adolescents but was not associated with an increase in fat mass.4 A study of female twins showed that the twin who was heavier at birth was taller (3.3 cm/kg birth weight) and slightly heavier (1.1 kg/kg birth weight) at 18 to 34 years of age.5 The heavier twin also had more lean mass, less subcutaneous and abdominal fat, and a lower waist/hip ratio as an adult, whether adjusted for adult body mass or not. Findings were similar in a corresponding study of male twins.6 Therefore, at the same attained BMI, an adult who was born SGA might maintain a higher fat content and thus have a higher health risk than an adult who was born LGA.
Both Dr Kramer and Dr Gillman showed that the mean birth weight has been increasing during the past decade, with concurrent increases in LGA births and decreases in SGA births. Dr Kramer showed that this trend could be explained by several known factors, including higher maternal prepregnancy weight. If higher weight infants become higher weight adults, then the trend for increased birth weight is likely to continue.
Evidence presented by Dr Gillman showed that, despite the relationship of high birth weight to increased weight and BMI later in life, being in a high BMI group at birth is not associated with increased prevalence of high blood pressure later in life. A group with low BMI at birth shows such an association.7 Leon et al8 noted a decrease in systolic blood pressure among men 50 years of age as birth weight increased from <3250 g to ≥4250 g. Similarly, being in the highest birth-weight tertile was associated with a reduced risk of insulin resistance in adulthood across adult BMI tertiles.9 Therefore, the association of high birth weight with later weight and BMI may not be paralleled by an increased risk of other, more-proximal, disease outcomes.
EARLY POSTNATAL LIFE
Increasing evidence indicates that rapid postnatal weight gain, particularly during the first 1 year of life, increases the risk of high adult BMI, particularly for individuals born SGA. Drs Gillman, Kramer, and Ness reviewed others’ data or presented their own data showing that individuals who are small at birth and who gain most rapidly in the first several months of life have a particularly increased risk of high BMI later in childhood and insulin resistance, hypertension, and coronary artery disease later in life.
In addition, Drs Ness and Gillman presented data indicating that an early “adiposity rebound,” ie, the time in childhood when BMI reaches a nadir and starts to increase with age, also increases the risk of later overweight. These findings are relevant to the relationship of breastfeeding or formula feeding to later obesity development, because the 2 feeding methods may promote different patterns of growth.
Most breastfed infants in the United States receive some formula, many are only partly breastfed, and many are breastfed relatively briefly. Therefore, any protective effect of full breastfeeding would be underestimated in analyses that group breastfed infants into 1 or 2 feeding groups or limit comparisons to dichotomous breastfeeding variables (yes or no). In such cases, brief and partial breastfeeding would be combined with exclusive breastfeeding and compared with no breastfeeding. Dr Gillman presented such data from 8 studies. He showed that any breastfeeding was associated with a reduction in the odds of later obesity in all of the studies, although not all showed a statistically significant effect. He noted that confounding factors could be determining the observed association. However, dilution of the actual effect by partial or token breastfeeding is also a problem. Data from a sibling-pair analysis of the Growing Up Today Study cohort that he studied prospectively showed a reduced odds of overweight in adolescence with increased breastfeeding duration.10 However, because the associations are generally weak and may become insignificant after adjustment for the many differences, other than infant feeding practices, between breastfeeding and formula-feeding mothers, the findings are not always consistent or completely persuasive. In other words, it is recognized that unknown, suspected, and established confounders might create a spurious association between breastfeeding and reduced obesity risk.
It was also noted that part of the difficulty of determining whether breastfeeding and reduced adult obesity risk are causally related is attributable to the paucity of detailed data on exclusiveness and duration of breastfeeding. This is particularly true when adult outcomes are considered in relation to feeding practices of many years ago, because recall of breastfeeding duration and exclusiveness may be inaccurate. There is also little consistency in defining terms such as “mostly breastfed,” “partly breastfed,” and even “exclusively breastfed.” Very detailed data, gathered prospectively, are needed to establish the degree of breastfeeding or the proportion of the infant’s intake that is provided by breast milk or its substitutes.
Data from a national survey discussed by Dr Devaney showed that breast milk provided only 30% of daily energy for children 4 to 6 months of age, whereas recommended practices would have that be close to 100%. In Dr Bentley’s study of African American mothers in North Carolina, only ∼15% of children were reported to be exclusively breastfed for 3 months, although >60% of the women had initiated breastfeeding.
Dr Mennella presented data that might bear on how breastfeeding could affect long-term diet. She drew attention to the very early exposure of the child, through both amniotic fluid in utero and breast milk, to the flavors of the foods the mother consumes. Many food flavors can enter these body fluids. Early exposure to these flavors was proposed to be 1 way in which infants learn to recognize and accept suitable foods that are typically consumed in their culture. Studies discussed by Dr Mennella showed that food flavors consumed by pregnant women are more readily accepted by the newborn infants. Similarly, vegetables consumed by lactating mothers were preferred by the children at weaning. Therefore, in addition to the potential effect of breastfeeding on later health, breastfeeding might affect health in childhood by contributing to early food preferences. Whether greater acceptance of vegetables by a child would result in a lower obesity risk later in life was not demonstrated.
In general, the speakers were supportive of breastfeeding because of its many health benefits, but some cautioned that its possible protection against obesity development was too uncertain, and likely small, to warrant the use of that benefit as a reason for promoting breastfeeding. Although it is true that confounding characteristics might contribute to the association of breastfeeding with lower obesity risk, they might not explain all of the risk reduction. In that regard, the positive results of sibling-pair analyses are reassuring. Until we develop methods to collect high-quality data on breastfeeding duration and exclusiveness, we probably cannot know with certainty whether breastfeeding is protective against obesity.
The focus on alternative explanations (confounding factors) for the reduced obesity risk associated with breastfeeding underemphasizes the possible value of breastfeeding. If we considered this issue from the standpoint that formula use is associated with increased obesity risk (even because of unknown or confounding reasons), we might be more willing to advise women that they might decrease their infants’ risk of later obesity by avoiding formula use. Because all agree that full breastfeeding should be supported for its other health benefits to the mother and child, this appears to be a useful and truthful public health approach.
Dr Mennella called attention to the role of the individual child, who, in responding to the tastes and smells of various foods, may influence what is fed. These child behaviors may be determined or influenced by familiarity and by the individual child’s intrinsic preferences for the various flavor classes (sweet, sour, salt, bitter, and umami), which might be partly determined by genetics and manifested in differing sensitivities. Dr Mennella referred to studies indicating a developing repertoire of taste sensitivity in infancy, which results in a different and varying spectrum of preferences, compared with adults. Sweet preference seems to be innate. Children prefer a higher level of sweetness than adults do. This preference is functional, because breast milk has a predominantly sweet taste. Nevertheless, it has worrisome implications for diet quality, because highly sweetened foods and beverages have become widely available and are served to children at very young ages, as shown by Dr Devaney. It may be an even greater concern in the United States, where a child-centered feeding pattern is common, as discussed by Dr Wardle.
Dr Mennella indicated that salt preference is also strong in infancy. In contrast to the allure of sweetness and salt, bitter flavors seem to be innately aversive to infants. However, studies of children raised on protein hydrolysates indicate that early exposure to bitter and sour flavors increase their acceptance. This observation may be relevant for children’s acceptance of vegetables and other foods that are low in sugars and salt.
Salt enhances the flavors of other foods, modifying bitterness as well as off-tastes in food, and preferences for saltiness increase with experience with salty foods. Therefore, the high salt content of many fast foods may promote later food and salt preferences that are detrimental to adult health. An understanding of the role of such preferences in determining the acceptance of healthful or less healthful foods may be important for recommendations regarding the order of introduction to infants of different food groups (eg, fruits, meats, and vegetables). Such an understanding might also support policies and education aimed at delaying the introduction of certain table foods (high-salt or high-sugar foods).
Dr Devaney’s presentation demonstrated the extent to which such foods have become a standard part of the diets of infants and toddlers. She presented data from a survey of a national sample of infants and toddlers that indicated that high-salt fast foods and high-sugar snacks and drinks are commonly consumed by individuals 1 to 2 years of age. The data also showed that the reported energy intake for these infants exceeded their estimated energy needs by 23% at 7 to 11 months of age and by >30% at 1 to 2 years of age. Because of the preferences of young children for foods high in sugar and salt, the introduction of such foods may be particularly likely to lead to overconsumption at these early ages. Furthermore, nearly 30% of children had received solids before 4 months of age, which contributed to the low percentage of energy (30%) provided by breast milk between 4 and 6 months of age. Dr Bentley showed that nearly 80% of a group of African American mothers participating in the Special Supplemental Nutrition Program for Women, Infants, and Children had fed their infant solids by 8 weeks of age.
A special characteristic of obesity development in childhood, as in infancy, is its dependence on the actions of others to create the circumstances that cause it or cure it. This is so whether we think of passive environmental factors (such as the presence of a television in the home) or active parental interventions (such as restricting consumption of certain foods). Therefore, to elucidate how the changing circumstances surrounding the feeding of children in the past few decades might have influenced changing obesity rates, parenting characteristics and feeding styles were considered.
Drs Wardle and Bentley particularly focused on these issues. Dr Wardle discussed data from France and the United States that supported the idea that restrictive or controlling feeding styles, which are more common in France and among groups with higher socioeconomic status in several countries (including the United States), might help prevent obesity. Other US studies reported associations between restrictive parental feeding styles and child overweight and proposed that restrictive styles might impair the development of self-regulation of intake and might result in the development of overweight. In the African American cohort she studied, Dr Bentley found that the obese mothers were less restrictive than the lower weight mothers. They also paid less attention to the quality of their children’s diet.
On balance, the findings presented seemed to indicate that parents try to control their children’s diet when they perceive a health need. In the Twins Early Development Study in the United Kingdom, Dr Wardle found that, among twins of discordant weight, more control was exercised toward the twin perceived to be underweight. Controlling feeding styles might be used by some parents to optimize children’s weight, either to prevent obesity development or to increase the weight of low weight children. Intervention studies described by Dr Wardle indicate that parents can be taught methods that alter the foods accepted by their children. Parents can use these methods to increase acceptance of healthful foods, rather than undesirable junk food. The utility of such training might be limited, however, if high-risk families have a low probability of changing their feeding approaches.
Parental obesity was shown by several speakers to be highly predictive of children’s overweight/obesity. This observation suggests (as stated by one of the speakers) that improving infant feeding practices might be at least partly dependent on improving the diet of the entire family.
Data on activity patterns of children were also presented, and several speakers indicated the importance of this part of the energy balance equation. Dr Devaney noted that the data she presented, which showed children’s energy intake exceeding estimated energy requirements, might be attributable to maternal overreporting of the child’s intake but also might indicate that energy expenditure was lower than assumed in the new dietary reference standard.11 Whatever the primary factor, weight trends show that energy intake and expenditure are often out of balance.
Dr Kramer estimated that an excess of only 65 kJ per day between energy intake and energy requirements would result in an 8-lb weight gain in 8 years, the gain most frequently observed among adults 20 to 40 years of age in 2 National Health and Examination Surveys, 8 years apart.12 A persisting excess of 210 kJ per day would result in a 40-lb gain during that period. Dr Kramer also listed the very high energy values of many common fast foods, in the range of 840 to 2500 kJ, and the long periods of standard exercises that would be required to expend that much energy. A recent summary of the dietary intake data for adults 20 to 70 years of age from the National Health and Nutrition Examination Survey13 indicated that energy intake might have increased by >670 kJ per day among men and by 1400 kJ per day among women between 1971 and 2000. However, most of the increase occurred before 1988; changes in the 24-hour diet recall methods, which occurred in 1988, may account for some or all of the increase.
Although individual food items have energy contents that can be expended only with long periods of exercise, long exercise periods may not be needed to prevent the observed problem. An attendee, Dr S. Blair, commented that programmed exercise is often the focus of energy expenditure recommendations but the reductions in output have likely come from “the little bits of calories here and there that have been whittled out of daily existence.” His point is well taken. If an excess of only 65 to 210 kJ per day would account for weight gains of 8 to 40 lb in an 8-year period, then daily activities that added that much to the energy expenditure side of the balance equation could be very helpful. The support that several speakers gave to efforts to reduce television watching by children by 30 minutes per day might be viewed in that light. A recent New York Times opinion piece14 noted the results of a survey in several US cities. A large proportion of parents used strollers for 3-year-olds (75%) and 4-year-olds (39%), instead of having them walk. Surely this represents a valuable missed opportunity for physical activity for the child, probably sufficient to increase expenditure by >210 kJ per day.
It is generally agreed that, although genetic predisposition or susceptibility may play a role in determining who becomes overweight or obese, genetic factors have not changed during the period of increasing obesity rates and these factors cannot have caused the upsurge. Furthermore, genetic factors are not generally amenable to intervention. Therefore, increased understanding of genetic vulnerability, although it may be helpful in identifying those at highest risk, cannot direct us to the causes or treatments of interest. The conference speakers did not focus on genetic risk factors.
As a result of the conference, several areas were identified in which gaps and inconsistencies in the current literature indicate that additional research is needed. Some of this research could probably be implemented with existing data sets, used with more analytic and sophisticated conceptual frameworks. Other areas would necessitate almost completely new research approaches. The following section summarizes the research needs identified by the editors after review of the conference materials.
Determine Body Composition of High-Birth-Weight, LGA, and SGA Infants
Obesity is not fully defined by measures of weight and length, among infants, children, or adults. Increased and consistent use of measures that are not technologically complex, including measures of length, abdominal circumference, and skinfold thickness, if recorded rigorously, would allow more complete understanding of how fat contents at birth and in early childhood influence later obesity and health.
Better Define How Breastfeeding and Formula Feeding Influence the Patterns of Weight Gain in the First Year of Life and How These Patterns Are Related to Later Risk of Obesity
Encouragement of rapid growth in infancy has been a traditional approach. Undue concern about a slower growth rate among breastfed infants may contribute to the abandonment of breastfeeding. Newer growth standards developed with fully breastfed infants might moderate this tendency. Striking the right balance to avoid both acceptance of inadequate growth and promotion of excessive growth remains a challenge.
Formula feeding and breastfeeding may require different considerations for evaluation of growth. Infants born LGA, SGA, or appropriate weight for gestational age and infants of different gestational ages (eg, term versus preterm) may require different early trajectories for optimal long-term health. Future research should explore the optimal patterns of weight gain for infants of different gestational ages, weights, and composition at birth, according to the method of feeding. Suitable animal models might be helpful for providing direction or addressing some of these issues.
Develop and Promote the General Use of Consistent Standardized Measurements and Definitions for a Number of Key Variables
Breastfeeding classifications should be standardized, so that new studies collect feeding data in consistent and more detailed ways. Terms that define breastfeeding intensity and duration and that are compatible with the way women think about and report their practices need to be developed for routine use in research. Research questions that might have been answered decades ago remain unanswered because of repeated failures to distinguish and monitor important differences in breastfeeding practices. Classifying infants as breastfed or not breastfed is not sufficient for detection of small but important effects.
Just as we have moved from measures of weight to measures of BMI and then to measures of body fat in adult studies, studies of infants and children need to advance beyond measures of height and weight to measures that are specific, clearly defined, and more directly related to outcomes of interest. In some cases, more explicit definitions and more consistent use of current terminology would be helpful. LGA and high birth weight have different meanings and may have different implications for later disease. The literature often uses “macrosomic” as an equivalent term, but with a clearly negative implication. These terms should not be conflated. Similarly, high-birth-weight infants born to women with diabetes should be distinguished from high-birth-weight infants born to nondiabetic mothers. Children born SGA also are not a homogeneous group, and measures that would help distinguish different risk status conditions and different causal mechanisms (such as ponderal index, head circumference, and waist circumference) require additional testing.
Clarify the Effects of Parental Feeding Styles on Later Obesity Risk
Significant research has examined the effects of “controlling” feeding styles on later child behavior and weight. Data comparing different cultures suggest that control is inversely associated with weight, but an absence of prospective data collected within the United States makes it impossible to determine the direction of causality. Is the greater degree of parental control observed among parents of obese children contributing to the obesity or a response to it? Furthermore, a high degree of parental control in infant feeding may be beneficial, whereas the same degree of control with an older child may be detrimental.
Decisions about feeding must be made whatever the quality and certainty of the information we have. We need to develop guidelines even when information is provisional. There is currently too much reluctance to base guidance on the knowledge available and too much concern about adverse effects of controlling the child’s diet.
The conclusions from studies of controlled feeding styles should not be extended to aspects of control of the food environment (availability of or access to sweet, salty, or nutrient-poor snacks, for example). In the current food and marketing environment, parental control of infants’ and children’s food options may be essential for good health and may not result in the same problems as control over meal patterns, food choices within a meal, or total intake. Research should be directed at identifying parenting patterns that are conducive to healthful eating and family harmony throughout the stages of childhood. Many parents crave greater direction about feeding in their efforts to rear healthy children.
Issues regarding parental control of activity are similar. Attempts to control or force activity on children may have negative consequences. In contrast, changing the environment by limiting time spent watching television and using the computer or increasing family activity may be more positive.
Continue to Explore the Effects of Exposure to a Range of Flavors Prenatally and in Breast Milk on Later Dietary Choices and Patterns
Such a determination may require longitudinal studies that accurately quantify the duration and frequency of breastfeeding, as well as the maternal diet. Mothers would be interested to know whether varied food choices during pregnancy and lactation could influence their children’s later food preferences, and this knowledge might change their behavior.
Determine the Effects of Timing and Order of Exposure of Infants to Various Food Groups on Their Food Acceptance and Body Weight
It is important to establish whether early exposure to vegetables, which is associated with an increased acceptance of vegetables in childhood, also lowers the risk of obesity. Independent of the timing of exposure, the order of exposure of the infant to different food groups may affect food preferences, which may in turn affect obesity risk. Therefore, the effect of order of exposure also should be studied.
Clarify the Mechanisms by Which Maternal Obesity Influences Child Obesity
Although genetic factors may play a role, the stronger relationship of child obesity to maternal obesity than to paternal obesity, as observed in some studies, suggests a behavioral component. Genetic factors cannot be changed but behavioral factors can be, if we can identify them. We need a better understanding of the nongenetic mechanisms linking maternal and child obesity.
Implement Mechanistic Studies of Obesity Development Designed to Establish Causal Pathways and Clarify Which Associated Factors Are Not Causal
Epidemiologic studies can reveal associations. Other types of studies are needed to evaluate which associations might actually prevent later obesity development, to suggest targets for intervention. Good data are needed as we move forward. New data sets should include information on maternal diabetes mellitus during pregnancy and the degree to which it was controlled, on many of the physical dimensions of the newborn, on body composition or body fat at birth, on the physical characteristics (eg, weight, height, and skinfold measurements) of the mother and father, and on postnatal growth and diet. Data for many of these items may need to be collected more rigorously and more consistently than is typical in the clinical setting. Because of the expense of longitudinal studies, currently planned or future longitudinal studies of pregnancy and infancy that are primarily focused on other issues should be encouraged to collect information on factors associated with later obesity.
Implement Studies to Evaluate Interventions That Prevent or Reduce the Risk of Childhood Obesity or Improve Body Weight Among Children
Although studies of causation can suggest appropriate intervention approaches, only research can establish what works in specific settings.
PUBLIC HEALTH ISSUES
The conference also suggested important clinical and public health messages that require increased emphasis. These represent areas in which the underlying scientific findings are reasonably well established and the health consequences are significant, so that the messages can be promoted without additional research. However, they often receive short shrift in health care settings. Health messages are most effective when they are consistent, repeated, and focused. These messages deserve increased attention, particularly from pediatricians, dentists, and other health care practitioners who see parents and children regularly. 1) The frequency and duration of exclusive breastfeeding should be increased. 2) Nothing should be served in an infant’s bottle except formula (or breast milk), ie, no cereal, soda, or juice. 3) Solid foods are frequently introduced too early; appropriate timing needs more support. 4) Foods served to infants and toddlers are in many cases unhealthful. Food choices suitable for adults are not necessarily suitable for young children. 5) Creating a home environment characterized by healthful food choices and regular physical activity minimizes the need for undesirable patterns of parental control and improves the health of the whole family. 6) Parents should model for their children eating and activity behaviors that improve health. 7) Marketing to young children and serving them manufactured products that are high in salt or sugars should be discouraged.
There are many aspects of the data related to obesity development among children that hamper interpretation and the drawing of firm conclusions. Almost all areas discussed are plagued by a lack of standardization or of rigorous evaluation of the utility of the outcome measures used. For example, in some studies, early life predictors were associated with changes in adult BMI in the normal range. However, the effects on risk of increasing BMI within that range might be trivial.
LGA infants can be of any weight, depending on gestational age. A 2% increase in LGA births does not mean a 2% increase in infants with high birth weights. Infants with birth weights with high absolute values (eg, >4000 g) would virtually all be LGA infants born at term. Because fat is laid down predominantly in the last month of pregnancy, LGA infants born at term might differ in body composition (higher percentage of fat) and in later obesity risk from LGA infants of lower weight born before term.
Furthermore, in many data sets, groups of truly high-birth-weight infants (>4000 or >4500 g) might include a disproportionate number of children of diabetic mothers, whose composition and obesity risk might be very different from those of children not exposed to a diabetic environment but born with high birth weight. This might be particularly true in older data sets, in which the mean birth weight was lower than present values and diabetic diagnosis and control were incomplete. In general, caution should be used in extrapolating the risks of high birth weight, term infants to all LGA infants or in extrapolating to all high-birth-weight infants the risks identified among infants born to diabetic mothers.
In addition, the same weight outcomes resulting from different starting points might not mean the same thing, as might be the case when the same adult BMI is observed for individuals born SGA or LGA. Higher weight or BMI does not necessarily mean proportionately higher body fat. To truly understand these differences, differences in body composition and in more-proximal health outcomes (morbidity or death) must be determined. Future studies, and possibly interventions, will require information about early life circumstances, good measurements of birth length and weight, data on maternal weight gain, and measurements of maternal and paternal heights. It would be useful to have clinicians play a role in ensuring the quality of the data from which future policies will be developed.
Several speakers noted the generally limited intake of fruits and vegetables by children and explored possible determinants of this pattern. We should not assume, however, that increased intake of these more healthful foods would reduce obesity risk, because intake of sweet, salty, and high-fat foods might not change if more healthful foods were also eaten.
Finally, readers should be aware that, in general, the unintended consequences of interventions that might reduce obesity risk were not discussed at any length by any of the speakers. Some of the reported findings suggest possible interventions, for example, those aimed at promoting exclusive and prolonged breastfeeding, increased physical activity, or decreased consumption by children of high-sugar and high-fat foods, limiting the growth of SGA newborns in the first months of life, or encouraging more restrictive feeding styles. However, some of these interventions could have other, more important, negative effects that would offset their potentially beneficial effects on obesity development. For example, there was passing mention of the potential harm that might result from ill-conceived attempts to reduce postnatal catch-up growth among SGA infants or to delay the adiposity rebound to reduce later obesity risk. Possible efforts to increase fetal growth to reduce the incidence of SGA births were criticized because of the potential effects on increasing LGA births and associated obesity. This caution may be misplaced. First, relatively few infants are born at very high birth weights (eg, 2% at >4500 g but 13.7% at 2500–2999 g among adult white mothers15). Second, infant and neonatal mortality rates are more than twice as high in the lower weight group than in the higher weight group.15 This early outcome difference could easily take precedence over effects on later obesity rates. Other positive outcomes associated with increased birth weight would also need to be evaluated. Analogous considerations of other health consequences must be made before we recommend other interventions to reverse obesity trends, if we are to improve overall health.
The conference illustrated that there is much for us to learn about what causes obesity to develop among children. As we attempt to intervene to prevent obesity development, we must be careful that we do not create other serious health problems; that is the challenge in the coming years.
- ↵Yarbrough DBS, Barrett-Connor E, Kirtz-Silverstein D, Wingard DL. Birth weight, adult weight, and girth predictors of the metabolic syndrome in postmenopausal women: the Rancho Bernando study. Diabetes Care.1998;21 :1652– 1658
- ↵Byberg L, McKeigue PM, Zethelius B, Lithell HO. Birth weight and the insulin resistance syndrome: association of low birth weight with truncal obesity and raised plasminogen activator inhibitor-1 but not with abdominal obesity or plasma lipid disturbances. Diabetologia.2000;43 :54– 60
- ↵Hediger ML, Overpeck MD, Maurer KR, Kuczmarski RJ, McGLynn A, Davis WW. Growth of infants and young children born small or large for gestational age: findings from the Third National Health and Nutrition Examination Survey. Arch Pediatr Adolesc Med.1998;152 :1225– 1231
- ↵Singhal A, Wells J, Cole TJ, 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
- ↵Loos RJ, Beunen G, Fagard R, Derom C, Vlietinck R. Birth weight and body composition in young women: a prospective twin study. Am J Clin Nutr.2002;75 :676– 682
- ↵Adair LS, Cole TJ. Rapid child growth raises blood pressure in adolescent boys who were thin at birth. Hypertension.2003;41 :451– 456.
- ↵Leon DA, Koupilova I, Lithell HO, et al. Failure to realise growth potential in utero and adult obesity in relation to blood pressure in 50 year old Swedish men. BMJ.1996;312 :401– 406
- ↵Food and Nutrition Board, Institute of Medicine. Dietary Reference Intakes for Energy, Carbohydrate, Fiber, Fat, Fatty Acids, Cholesterol, Protein, and Amino Acids (Macronutrients). Washington, DC: National Academy Press; 2002
- ↵Hill JO, Wyatt HR, Reed GW, Peters JC. Obesity and the environment: where do we go from here? Science.2003;299 :853– 856
- ↵Crain W. No free ride for toddlers. New York Times. Sep 6, 2003;A11
- ↵Rees JM, Lederman SA, Kiely JL. Birth weight associated with lowest neonatal mortality: infants of adolescent and adult mothers. Pediatrics.1996;98 :1161– 1166
- Copyright © 2004 by the American Academy of Pediatrics