 |
|
 |
|
 |
 |
|
 |
 |
 |
 |
 |
 |
 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
* Center for Developmental Epidemiology, Department of Psychiatry and Behavioral Sciences, Duke University School of Medicine, Durham, North Carolina
Department of Anthropology, Emory University, Atlanta, Georgia
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMETHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
Methods. White children (N = 991) 9 to 16 years old from the Great Smoky Mountains Study, a representative sample of rural youth, were evaluated annually over an 8-year period for height, weight, psychiatric disorder, and vulnerabilities for psychiatric disorder. Longitudinal analyses on the repeated measures data were conducted using developmental trajectory models and generalized estimating equation models.
Results. Obesity was 3 to 4 times more common than expected from national rates using Centers for Disease Control and Prevention 2000 criteria. Four developmental trajectories of obesity were found: no obesity (73%), chronic obesity (15%), childhood obesity (5%), and adolescent obesity (7%). Only chronic obesity was associated with psychiatric disorder: oppositional defiant disorder in boys and girls and depressive disorders in boys.
Conclusions. In a general population sample studied longitudinally, chronic obesity was associated with psychopathology.
Key Words: obesity • psychopathology • developmental trajectories • depression • conduct disorder
Abbreviations: GSMS, Great Smoky Mountains Study • BMI, body mass index • NHANES, National Health Interview Survey • CAPA, Child and Adolescent Psychiatric Assessment • ADHD, attention-deficit/hyperactivity disorder • SPMM, semiparametric mixture model • GEE, Generalized Estimation Equation • CDC, Centers for Disease Control and Prevention
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
Overweight and obesity, or perceptions thereof, may also affect self-esteem, body image, and social mobility,5,8 and it is well-known that adults who are overweight or obese are at an increased risk for psychological disorders.9,10 Less is known about whether childhood obesity is associated with concurrent psychopathology, or with specific types of psychiatric disorder. Most of the published research on obesity and psychopathology in children has used clinical samples recruited because of obesity11–14 or depression.15 The risk of psychiatric disorder in obese children in these referred samples may not reflect what is seen in the general population.
Among the structural, familial, and environmental vulnerabilities associated with both obesity and psychopathology in childhood are being from a poor16,17 or single-parent household,17,18 having harsh or abusive parents,17,19,20 and, for some psychiatric disorders, being female.21 However, there is little research that establishes the direction of effect, or the role of these factors as mediators or moderators of the link between obesity and psychopathology.
Weight fluctuations are common in childhood, and even more so in adolescence.22 Because most studies of childhood obesity and psychopathology are cross-sectional or retrospective, we do not know whether there are distinct patterns of obesity throughout childhood, nor do we know whether certain patterns are associated with particular outcomes. This study provides repeated measures of height and weight taken annually over 8 years, together with annual information about family characteristics and mental health. Longitudinal data permit identification of distinct developmental trajectories of obesity, separating the transiently obese from the chronically obese, and those whose obesity is confined to childhood or adolescence.
The goals of this study are 1) to determine the number and type of distinct obesity trajectories in a general population sample of white children 9 to 16 years old; 2) to examine whether trajectory membership is associated with family vulnerabilities such as gender, poverty, single-parent family, parenting style, parental history of mental illness or drug abuse, and life events; and 3) to test the relationship between obesity trajectory membership and psychopathology, controlling for a range of potential mediators.
|
METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
We used a 2-phase process to select the final sample for the longitudinal study. A screening questionnaire was administered to a parent (usually the mother) of the first stage random sample (N = 3896; 95% of those contacted). The questionnaire consisted mainly of the externalizing (behavioral) problems scale of the Child Behavior Checklist,26 and was administered by telephone or in person. All children scoring above a predetermined cut point (the top 25% of the total scores, in this case it was a score of 20) plus a 1-in-10 random sample of the remaining 75%, were recruited for detailed interviews. Eighty percent of those recruited agreed to participate. The contribution of each participant was weighted by the inverse of their selection probabilities, stratified by age and gender, to provide accurate prevalence estimates for the population of the study area.27 Families were reinterviewed annually until the child was 16, and every 2 to 3 years thereafter. The data presented here, based on the first 8 annual waves of the study (1993–2000), consist of 4600 interviews with 991 non-Hispanic white participants and their parents.
Fewer than 10% of the area residents and the sample were African American. Because racial/ethnic differences in adiposity and its correlates are well-documented,28,29 non-white children were excluded from these analyses as power was not adequate for comparisons.
The study protocol was approved by the Duke University School of Medicine Institutional Review Board.
Procedures
Two interviewers visited the family each year, either at home or in a location convenient for them. Before the interviews began, parent and child signed informed consent forms. They were then interviewed in separate rooms. Each parent and child was paid $10 after the interview.
Measures
Obesity
Height and weight were both assessed at 2 different points during the interview, and the 2 averaged for these analyses. Participants wore normal clothing without shoes, socks, and belts. Height was measured to the nearest 0.1 cm using a stadiometer (CMS Weighing Equipment, London, United Kingdom) and standard techniques.30 The measurement was repeated if the first 2 measurements differed by >0.5 cm, and the nearest 2 of the resultant 3 were averaged. Participants were weighed twice to the nearest 0.1 kg on a portable digital scale (Soehnle, Murrhardt, Germany); a repeat measure was taken if the first 2 differed by >0.2 kg, and averaged as for height. Body mass index (BMI) was calculated using the standard wt/ht2 formula.
For adults, obesity is generally defined as BMI >30 kg/m2; but no equivalent standard exists for children.30,31 In this study we use age- and sex-specific 95th percentiles from the revised (2000) growth charts of the Centers for Disease Control and Prevention (CDC), which are based on data from the National Health Interview Survey (NHANES).32 The reference values are based on US national survey data and are intended for US children and adolescents.33 Because the GSMS children were weighed clothed, and the NHANES children were weighed wearing only a gown, undergarments, and foam slippers, we subtracted 1 kg from weight before calculating the BMI.
Psychopathology
The psychopathology outcomes were assessed using the Child and Adolescent Psychiatric Assessment (CAPA), a psychiatric interview for children 9 to 17 years old. The CAPA is an interviewer-based interview.34 The goal of interviews using this format is to combine the advantages of clinical interviews with those of highly structured epidemiologic interview methods.35 While using a highly structured format of questions and probes, the interviewer-based approach trains the interviewer to ensure that the parent or child being interviewed understands the construct under review, and provides enough detail and examples for a clear rating of the clinical severity of each symptom to be made. A detailed glossary provides the operational rules for identifying a clinically significant level of severity for each symptom.
The CAPA interviews parent and child separately, using different interviewers. The presence of a symptom can be determined on the basis of information from a single respondent, or using the either/or rule common in clinical practice. For these analyses, we counted a symptom as present if reported by either parent or child or both. The time frame of the CAPA for determining the presence of most psychiatric symptoms is the past 3 months. Diagnoses were based on Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition criteria.36 For these analyses we combined psychiatric diagnoses into 7 groups: conduct disorder, oppositional defiant disorder, depressive disorders (including major depression, dysthymia, and depression not otherwise specified), anxiety disorders (including separation anxiety, generalized anxiety disorder, simple phobia, social phobia, agoraphobia, and panic), bulimia, substance abuse, and attention-deficit/hyperactivity disorder (ADHD).
Background and Family Variables
Potential mediators and moderators of the relationship between obesity and psychopathology included in these analyses were gender; family income (coded 0 = $0–$25 000, 1 = $25 001–$45 000, and 2 = 45 001+); 1 or both parents with less than an 11th-grade education; single parent family; parental history of treated mental illness, drug abuse, or criminal conviction; harsh or overprotective parenting style; lax supervision; and traumatic life events.
Season
We included in the trajectory models a variable to account for the season when the child was weighed and measured. All participants were interviewed each year on a day as close as possible to their birthday. Small seasonal variations in weight gain have been noted in 2 European samples,37,38 and studies relying on self-reports of weight have found that healthy adults believe their weight increases by 
5 pounds in the winter,39–41 although clinical research has failed to find support for seasonal weight gain.41 There is little literature on seasonal effects on children’s weight.
Despite the lack of consensus on seasonal weight variation in the literature, we noted a substantial seasonal weight variation in our sample. On average, 29% of children measured in the winter months (January, February, and March) were obese, compared with 20% during the remaining months, with a mean BMI difference of 1. To correct for this, we included a 4-category variable for season (January to March, April to June, July to September, and October to December) in the trajectory model as a time-dependent covariate.
Interviewers and Interviewer Training
Interviewers were residents of the area in which the study took place. All had at least bachelor’s level degrees. They received 1 month of training and constant quality control, maintained by postinterview reviews of each schedule, notes, and tape recordings by experienced interviewer supervisors and study faculty. Interviewers were trained by Department of Social Services staff in the States requirements for reporting abuse or neglect.
Data Management and Analysis
Scoring programs for the CAPA, written in SAS software (SAS Institute, Inc, Cary, NC), combine information about the date of onset, duration, and intensity of each event and symptom to create scale scores and diagnoses. Prevalence rates were calculated using Stata 7.0 (Stata Corporation, College Station, TX), taking the survey design into account through weights and clustering. Significant differences were calculated using a design-based Pearson 
2 with a second-order correction converted to an F statistic.
Obesity trajectories were determined by fitting a semiparametric mixture model (SPMM) to the data, using PROC TRAJ in SAS.42 SPMMs identify distinct groups of individual trajectories within the population. This approach to modeling growth curves is different from traditional latent growth curve modeling in that the latter assumes the random parameters to be bivariate normal distributed. In other words, all individuals belong to a single class of individuals who vary continuously on a latent trait. In contrast, the group-based method employed here assumes a number of discrete classes, each having a specific intercept and age slope and an estimated population prevalence.42 Because we suspected that individuals do not vary continuously on obesity, but rather that there are a distinct number of obesity trajectories, the group-based method was the most appropriate.
The logit model was used to model the presence or absence of obesity predicted by age. By specifying the sampling weights, we invoked the robust variance estimator (ie, sandwich type estimator) to adjust the standard errors of the parameter estimates to account for the 2-phase sampling design. Thus, the classes that were identified, and their estimated prevalences and correlates, relate to the entire population of white children in the 11-county area from which participants were sampled.
Children were classified into their most probable obesity trajectory class by the use of posterior probabilities. A posterior probability is the probability of each individual belonging to each group.42 Thus, individuals are assigned to the group to which they have the highest probability of belonging. Once children were assigned to a trajectory class, we conducted analyses in 3 stages.
First, we computed bivariate statistics to obtain prevalence rates of various background, familial, and psychopathological characteristics by obesity trajectory. An F test for each variable tested whether there was any difference among obesity groups. Second, we fit multinomial logit models to determine how the background and familial characteristics identified as significant in the first stage distinguished children on the different obesity trajectories from the comparison group of never-obese children. Third, we fit Generalized Estimation Equation (GEE) models to estimate the relationship between the different obesity trajectories and psychiatric disorders, controlling for the background factors identified in the second stage. The GEE models were run using binary regression with logit link function for dichotomous outcomes, such as the presence or absence of a psychiatric diagnosis, using the Stata program xtgee. The use of multiwave data with the appropriate sample weights capitalized on the availability of multiple observation points over time, while controlling for the effect on variance estimates of repeated measures on the same child, for overlapping cohorts, and for design effects.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
18 for 9 year olds, and rising to 20 by age 12, 22 by age 14, and 24 by age 16.
|
20% were obese. Generally, the prevalence of obesity increased with age. These results show that this rural, Southern sample was substantially heavier than the general population of children and adolescents in the United States, based on the CDC criteria, as 20% of the sample was above the 95th percentile by 16 years.
|
|
Differences Among Obesity Trajectories
Table 2 shows the prevalence of a range of child and family characteristics in each obesity group. There were no overall differences among the groups in gender, family structure, parenting style, family history of mental illness, drug abuse, crime, or traumatic events. Childhood and chronic obesity, but not adolescent obesity, were associated with having uneducated parents and lower income.
|
Predictors of Obesity Trajectory Membership
We fit multinomial regression models to determine which factors predicted group membership (Table 3). Taking the never obese as the referent group, chronically obese children were significantly more likely to be in the lower and middle income groups, and to have uneducated parents. The childhood obesity group was also more likely to have uneducated parents. There were no other significant differences between the never-obese group and children on any of the other trajectories.
|
Obesity Trajectories and Psychopathology
We tested for increased risk of any of 7 psychiatric disorders in the 3 obesity groups, relative to the nonobese group. Age, sex, and income were included in each model, as well as the other psychiatric disorders, to control for comorbidity. Oppositional disorder was more common in chronically obese boys and girls (odds ratio: 2.5; 95% CI: 1.36–4.61) and depression in chronically obese boys (odds ratio: 3.7; 95% CI: 1.27–10.2) but not girls. There were no significant associations between obesity trajectory membership and bulimia, ADHD, substance use, conduct disorder, or anxiety, controlling for comorbid psychiatric disorders. As shown in Table 3, the risk of both oppositional disorder (both sexes) and depression in boys were increased in youth with low incomes, and the risk of oppositional disorder was increased in youth with middle incomes as well. Testing for mediational effects, we found no mediational effects of income or any other family variables to explain the link between obesity trajectories and psychopathology.
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
In our sample, the prevalence of obesity was higher among the boys than the girls, although the difference was not significant. Several recent studies found obesity to be more prevalent among boys than girls,43–45 with rates of obesity increasing faster in boys than in girls.1 However, gender patterns may vary by race; therefore, any gender differences may depend on the race of the sample. Gordon-Larsen et al44 examined prevalence of overweight and obesity by sex and race from the National Longitudinal Survey of Adolescent Health and found that boys had a higher prevalence of overweight and obesity than girls among whites and Asians, but girls had a higher prevalence among Hispanics and blacks.
Familial and environmental risk factors for chronic obesity were surprisingly few; of the >20 examined, only income and lack of parental education remained significant in multivariable analyses. Children with chronic or childhood obesity were more than twice as likely as other children to come from families where 1 or both parents left school before the 11th grade. Lack of parental education remained significant after controlling for having a single, unemployed, or teenage parent, and other correlated risk factors (Table 3). An inverse relationship between income and BMI has been observed in several studies.46–48 The present study found a correlation between BMI and income of between -.20 and -.25 across each of the 8 annual assessments. However, this link between income and overweight may be specific to white children in the United States,1 and is clearly a phenomenon of developed economies. This study found an inverse relationship between chronic obesity and socioeconomic status in the combined sample, contrary to several previous studies, which find a stronger association among females than males.49 However, in our sample, different types of obesity had different associations with income so the apparent inconsistencies may be attributed to the difference between examining trajectories of obesity and obesity as a monolithic category.
Using developmental trajectory modeling, we were able to show that chronic and childhood-only obesity were more common in the lowest income group (annual family income below $25 000) than in the middle or upper income groups. Chronic obesity was also more common than expected in children from the middle income group ($25 000–$45 000). This inverse relationship between weight and poverty, beginning in childhood and, in the case of the poorest children, persisting into adolescence, reflects earlier studies.16,50 It is particularly troubling in the light of evidence from the 1946 British birth cohort that the inverse association between socioeconomic status and BMI predicted higher BMI through to middle age48 and was impervious to the child’s later academic or economic achievements.
The present study shows first and foremost that there are different types of childhood obesity. Additionally, this study shows that of the 4 basic weight trajectories—never obese, chronically obese, increasing risk, and decreasing risk—only chronic obesity was associated with a statistically significant increase in the risk of psychiatric disorder. The group with the next-highest prevalence of psychiatric disorder was in the small group who were obese in childhood but became less so as they moved through adolescence. Adolescent-onset obesity was not associated with increased risk of psychopathology.
Chronically obese children had significantly higher rates of oppositional defiant disorder, and (for boys) depression. This is, to our knowledge, the first study to show an association between chronic obesity and behavioral problems in children and adolescents. Several studies have linked child and adolescent obesity to depression, but the majority of these are based on clinic samples, referred either because of obesity11–14 or depression.15 We cannot rule out the possibility that the co-occurrence of the 2 conditions increased the likelihood of a clinical referral.51
Among epidemiologic studies, the evidence linking obesity and depression is much less clear. The New York longitudinal study52 found an inverse relationship between adolescent depression and young adult obesity in males, and no relationship in females. One study53 argued that the association between overweight and depression may be caused by the effects of dieting, or of poor health. Another54 demonstrated that any association between depression and BMI in a sample of third graders was explained by concern about being overweight. Of course, this may be true across the whole population but not true for the smaller group of obese youth.
Some studies have found that obesity is associated with lower levels of psychopathology, as opposed to higher levels.55 Friedman et al55 note that inconsistencies in prior studies of obesity and psychopathology may be attributed to methodological and sampling limitations. Additionally, we would add that inconsistencies in prior findings may be caused, in part, by the failure to consider developmental trajectories. In our sample, being obese at 1 point in time was not associated with an increased risk in psychopathology, rather being in the chronically obese trajectory was.
One limitation of this study is that it does not address the issue of causality. Studies such as this beg the question, does obesity increase the risk of psychopathology or does psychopathology increase the risk of obesity? DiPietro et al56 rely on epidemiologic data to examine changes in weight as a function of changes in depressive symptoms and find that depression plays a role in weight change. Younger men (<55 years old) who were depressed at baseline gained more weight during the follow-up period than those not depressed, while young women who were depressed at baseline gained less weight than the nondepressed. In a small prospective case-control study, Pine et al15 found that depression during childhood was positively associated with BMI during adulthood. Conversely, Pine et al52 found an association with current but not prior depression in a large, population-based sample. These inconsistencies between prior depression and current BMI may be due to differences in type of sample, measurement of psychopathology, and the use of self-report versus measured height and weight.
In this study, we examine the association between obesity and psychopathology; however, we draw no conclusions about causal ordering. If psychopathology increases the risk of obesity, we would expect an association in the chronic and adolescent-onset groups. If obesity increases the risk of psychopathology, we would expect an association in the chronic and childhood-limited groups. Our findings show that psychopathology is most common in the chronically obese group first, and the childhood-limited group second. Therefore, these results could be suggestive of obesity increasing the risk of psychopathology; however, that conclusion is purely speculative. Future research investigating the sequence of events is necessary to draw any conclusions about causation. It is also possible that rather than 1 increasing the risk of the other, they are both associated with the same physiologic mechanisms.
Evidence, largely from adults, links obesity and depression through underlying neuro-endocrine substrates. Specifically, activity of the hypothalamo-pituitary adrenal axis appears to be altered in both depressed57–59 and obese60,61 adults, although evidence from depressed obese adults is contradictory,62 and little is known about children. The trajectories of obesity identified here may manifest in part sequelae of suboptimal conditions for early development. For example, a growing body of research documents the role of maternofetal stress in later mental and physical health risk. Prenatal stress followed by postnatal overnutrition leads to childhood overweight63 and adult obesity64 as well as adult depression65 and increased hypothalamo-pituitary adrenal activity.66,67 Overall, relevant findings are more suggestive than definitive, leaving etiologic interrelationships between obesity and mood or behavior disorders and their possible common neuroendocrine mediators a subject for further investigation.
Although chronic obesity and bulimia were linked in bivariate analyses, there were very few cases of bulimia in this sample, and all but 2 were comorbid for oppositional disorder and/or depression. Interestingly, bulimia was only associated with obesity when it was comorbid with a psychiatric disorder. Bulimia is clearly an important clinical problem, and the fact that in this population of 9 to 16 year olds it was seen almost exclusively in children with other psychiatric disorders argues for vigilance on the part of clinicians to identify these comorbid children early.
There is debate about the accuracy of BMI as a measure of the prevalence of obesity.68,69 For example, Frankenfield et al69 compared BMI to percent body fat as measures of obesity, and found that BMI underidentified obesity. Defining obesity as a BMI 
30 or at least 25% body fat for men and 30% for women, 30% of men and 40% of women who had BMI <30 had obesity-level body fat. In adolescents, however, Malina68 found BMI to be a highly sensitive and specific marker of obesity, if not of overweight.
A technical limitation of this study is that the children were weighed and measured clothed except for their shoes, socks, and belts. A constant 1 kg was subtracted from their weight before BMIs were calculated. The seasonal difference in weight observed in this study may have been caused in part by extra clothing worn in the winter, although American homes tend to be kept to a fairly even temperature throughout the year. Another limitation is that although the GEE models used repeated measures, they did not deal with the relative timing of obesity and psychiatric disorder. Thus, we could determine if an association existed between the obesity trajectories and Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition diagnoses, but not whether an association was more likely to occur at a particular age or if the association existed over time.
|
CONCLUSIONS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
|
|
ACKNOWLEDGMENTS |
|---|
We are most grateful to the people of western North Carolina for their collaboration in the GSMS, to Dan Nagin and Mike Ezell for their methodological assistance, and to the reviewers for their helpful comments.
|
FOOTNOTES |
|---|
Address correspondence to Sarah Mustillo, PhD, Center for Developmental Epidemiology, Duke University Medical Center, Box 3454, Durham, NC 27710. E-mail: smustillo{at}psych.mc.duke.edu
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
This article has been cited by other articles:
 |
|
 |
|
  L. Gray and A. H. Leyland Overweight status and psychological well-being in adolescent boys and girls: a multilevel analysis Eur J Public Health, July 28, 2008; (2008) ckn044v1. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 D. M. Janicke, J. S. Harman, K. J. Kelleher, and J. Zhang The Association of Psychiatric Diagnoses, Health Service Use, and Expenditures in Children with Obesity-related Health Conditions J. Pediatr. Psychol., June 3, 2008; (2008) jsn051v1. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J. C. Lumeng, D. Somashekar, D. Appugliese, N. Kaciroti, R. F. Corwyn, and R. H. Bradley Shorter Sleep Duration Is Associated With Increased Risk for Being Overweight at Ages 9 to 12 Years Pediatrics, November 1, 2007; 120(5): 1020 - 1029. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 S. E. Anderson, P. Cohen, E. N. Naumova, P. F. Jacques, and A. Must Adolescent Obesity and Risk for Subsequent Major Depressive Disorder and Anxiety Disorder: Prospective Evidence Psychosom Med, October 1, 2007; 69(8): 740 - 747. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. Franklin, G. Denyer, K. S. Steinbeck, I. D. Caterson, and A. J. Hill Obesity and Risk of Low Self-esteem: A Statewide Survey of Australian Children Pediatrics, December 1, 2006; 118(6): 2481 - 2487. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 S. E. Anderson, P. Cohen, E. N. Naumova, and A. Must Association of Depression and Anxiety Disorders With Weight Change in a Prospective Community-Based Study of Children Followed Up Into Adulthood Arch Pediatr Adolesc Med, March 1, 2006; 160(3): 285 - 291. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 R. M Viner and T. J Cole Adult socioeconomic, educational, social, and psychological outcomes of childhood obesity: a national birth cohort study BMJ, June 11, 2005; 330(7504): 1354. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
H. Fonseca and M. Gaspar de Matos Perception of overweight and obesity among Portuguese adolescents: an overview of associated factors Eur J Public Health, June 1, 2005; 15(3): 323 - 328. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
A. Datar and R. Sturm Childhood Overweight and Parent- and Teacher-Reported Behavior Problems: Evidence From a Prospective Study of Kindergartners Arch Pediatr Adolesc Med, August 1, 2004; 158(8): 804 - 810. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. De Civita and P. L. Dobkin Pediatric Adherence as a Multidimensional and Dynamic Construct, Involving a Triadic Partnership J. Pediatr. Psychol., April 1, 2004; 29(3): 157 - 169 |
