Attention-Deficit/Hyperactivity Disorder and obesity in US males and females, age 8–15 years: National Health and Nutrition Examination Survey 2001–2004

Abstract

INTRODUCTION

Attention-Deficit/Hyperactivity Disorder (ADHD)and obesity are two common conditions that increase physical and/or mental health risks for youth in the United States.^1–3 ADHD can cause functional impairment by virtue of difficulties in disinhibition, working memory, planning, and sustained attention.^{4, 5} A growing but relatively recent literature has documented a link between ADHD and obesity.^{6, 7} Evidence that individuals with ADHD may have an elevated risk for obesity comes from studies of adults^{8, 9} and youth, in both clinical^10–13 and population-based settings.^14–17

That ADHD would be associated with higher risk for obesity has seemed counterintuitive given that many youth with ADHD have symptoms of hyperactivity. In one of the first investigations of these associations, Holtkamp et al. studied 97 German boys with ADHD to evaluate the hypothesis that they would have a lower prevalence of obesity than boys in the general population, but found instead that it was higher.¹¹ Similarly, many clinical samples of children and adolescents in the US and other countries have reported that obesity is at least as common among youth with ADHD as compared to youth in the population overall.^{6, 10, 12, 13} Several large population studies have also observed a modest association between obesity and ADHD.^14–17 Two analyses of the National Survey of Children’s Health (NSCH) 2003, a large telephone survey conducted with parents of US children, found that youth with ADHD who were not treated with medication were at higher risk of obesity than their non-ADHD counterparts,^{15, 17} and that the strength of this association may be stronger among females than males.¹⁵ In a national study of German youth in which height and weight were measured and ADHD was assessed using a structured diagnostic interview, Erhart et al found that ADHD was more prevalent among obese youth compared to non-obese youth.¹⁴

A number of potential mechanisms, some behavioral and others physiologic, have been hypothesized to account for these associations. From a behavioral perspective, ADHD could potentially increase risk for obesity through deficits in self-regulation and increased impulsivity. These characteristics may contribute to externally-cued eating, eating in the absence of hunger, or binge eating, which evidence suggests are associated with obesity.^{9, 14, 18} Factors such as under-arousal and hypoactivity, associated with the inattentive subtype of ADHD, may contribute to reduced energy expenditure, increased sedentary behavior, or poor sleep.^{15, 19, 20} Some evidence suggests that obese adolescents with sleep problems are more likely to have symptoms of ADHD.^{19, 21} Physiologically, ADHD and obesity may both be associated with hypo-dopaminergic function in the brain. Some research has shown that dysfunction of the dopamine receptor gene DRD2 gives rise to a “reward deficiency syndrome” that is associated with increased risk taking, substance abuse, and eating pathology.²² Overeating may be an attempt at self-medication given that palatable, energy-dense food is also known to activate dopamine pathways.⁷ It is well-established that eating-pathology is more common among females,²³ and this appears to be true as well for females with ADHD.²⁴ Accordingly, we hypothesized that for females with ADHD the risk for obesity might be higher than for males. However, these mechanisms are not fully understood, and are further complicated by the appetite suppressing effect of stimulant medication that is often used to treat ADHD symptomatology.²⁵

To date, only a few studies have investigated possible gender differences in associations between ADHD and obesity, and findings have been mixed.^{14, 15, 26} In the NSCH, Kim et al found that for both boys and girls, obesity was more prevalent among youth with ADHD who were not medicated, compared to youth who did not have ADHD. They also reported that the strength of the association (as represented by the adjusted odds for obesity) was greater for females than for males.¹⁵ Although gender differences were not tested statistically, stratified analyses presented by Erhart et al. in a population-based study of German youth (7–17 years), suggested similar associations between males and females.¹⁴ Another analysis of German youth found that associations between ADHD and obesity were strongest for adolescent females.²⁶

Utilizing nationally representative data from the US National Health and Nutrition Examination Survey (NHANES) 2001–04, we build upon what is understood about ADHD/obesity associations in population studies of youth by examining gender differences. The strengths of these data include direct measurement of height and weight and comprehensive assessment of ADHD using a structured diagnostic interview as well as parent-report. In this report we present analyses of data from NHANES 2001–2004 investigating: 1) the relationship of ADHD to obesity among US youth age 8–15 years; 2) the influence of medication treatment on the association; and 3) how the association may differ based on gender.

METHODS

We analyzed data for youth aged 8–15 years from the 2001–2002 and 2003–2004 waves of the National Health and Nutrition Examination Surveys (NHANES).^{27, 28} These are the most-recent waves of NHANES in which the mental health of youth was assessed using a structured diagnostic interview administered to parents.²⁹ In this report, we use ‘parent’ to refer to a youth’s primary caregiver, whether parent or guardian. NHANES is representative of the civilian non-institutionalized US population and a probability sample is selected using a complex, multistage, stratified, clustered design.³⁰ The response rate for youth in the 2001–2002, and 2003–2004 waves was approximately 85%.³¹ Participants in NHANES are examined and interviewed in a mobile examination center. Our analyses utilize survey weights that adjust for non-response, unequal selection probabilities, and under coverage of the target population.³⁰ NHANES restricts access to data on the mental health status of youth to researchers who have received approval from the National Center for Health Statistics (NCHS) Research Data Center and requires that researchers access restricted-use data through a secure system which they administer.³² Data collection for NHANES was approved by the NCHS Ethics Review Board. Analysis of restricted-use data through the NCHS Research Data Center is approved by the NCHS Ethics Review Board.³² The institutional review board of the Ohio State University determined that our analysis did not require additional review.

RESULTS

Of the 3907 youth aged 8–15 years who participated in NHANES 2001–2004, 3739 (96%) had measured height and weight. Six hundred and eighty nine (689) youth were excluded from our analyses because they did not have complete data on ADHD. The 689 youth excluded for missing information on ADHD did not differ statistically significantly from youth included in our analytic sample (n=3050) relative to gender (p=0.19) or the prevalence of obesity (p=0.22). However, youth who were excluded from the analytic sample due to missing data tended to be younger (mean age 11.7 years vs. 12.1 years, p = 0.04), poorer (mean PIR 2.09 vs. 2.57, p < 0.001), and more likely to belong to a minority racial/ethnic group (53.2% vs. 62.8% non-Hispanic white, p = 0.02) compared to those in the analytic sample.

Characteristics of the sample and population are presented in Table 1. Overall, 40.5% of youth were in families that had a PIR at or below 185% of the poverty level, but most youth were covered by health insurance. The prevalence (95% CI) of obesity (BMI-for-age ≥95^th percentile) for US youth ages 8–15 years in 2001–04 was 18.1% (15.5 – 20.6). As has been reported elsewhere,² the prevalence of obesity differed in the population relative to racial/ethnic group. High PIR was associated with lower obesity prevalence, and maternal smoking during pregnancy was associated with a higher prevalence of obesity. The prevalence of obesity among the 2% of youth with MDD was much higher than among youth who did not have MDD (34.7% vs. 17.7%, p=0.01).

Overall, 412 youth (15.6%) were classified as having ADHD; of these, 94 were identified by both the DISC-IV and parent-report, and 124 were identified by the DISC-IV only. Twice as many males compared to females had ADHD (20.3% vs. 10.5%, Table 1). The prevalence of ADHD was also higher for youth whose mothers smoked during pregnancy compared to youth whose mothers did not. Youth with MDD or conduct disorder were much more likely than youth without these comorbid disorders to have ADHD (Table 1).

One-hundred and eighty-five youth (7.4%) were taking medication for ADHD symptoms and 11.7% of these youth were obese (Table 1). Taking medication for ADHD was associated with a lower prevalence of obesity in comparison to all other youth (p = 0.04, data not shown); youth with ADHD who were not medicated (n=227) had a somewhat higher prevalence of obesity than youth without ADHD (21.6% vs. 18.3%) but this difference did not reach statistical significance (p=0.15)(Table 1). In logistic regression analyses combining males and females, the adjusted odds of obesity among youth medicated for ADHD were 0.62 (95 % CI = 0.35 – 1.08) times that for youth without ADHD. The adjusted odds of obesity among youth with ADHD who were not medicated were 1.18 (95 % CI = 0.70 – 2.00) times that for youth who did not have ADHD.

Stratified by gender, males with ADHD who were medicated were less likely than males without ADHD to be obese (adjusted OR = 0.42, 95% CI = 0.23 – 0.78) (Table 2). Males with ADHD who were not medicated were as likely as males without ADHD to be obese (adjusted OR = 1.02, 95% CI = 0.43 – 2.42). The odds of obesity for females taking medication for ADHD did not differ statistically from those of females without ADHD (adjusted OR = 1.21, 95% CI = 0.52 – 2.81). Females with ADHD who were not medicated had adjusted odds of obesity 1.54 times those for females without ADHD; however, the confidence interval was wide and ranged from a somewhat protective association (lower bound of 95% CI = 0.79) to a quite strong association (upper bound of 95% CI = 2.98).

DISCUSSION

We found that the prevalence of obesity among youth with ADHD who were not medicated was as high or higher than youth without ADHD, and that these associations may differ for males and females. Although not statistically significant at an alpha level of 0.05, females with ADHD who were not medicated had odds of obesity that were 50% higher than those of females without ADHD. For both males and females with ADHD the prevalence of obesity was lower among youth treated with medication and this was strongly so for males. Our analyses are representative of US youth aged 8–15 years in 2001–2004 and utilize anthropometric measurements to assess obesity and a structured diagnostic interview, medication-use, and parent-report to assess ADHD.

Our results from NHANES, particularly in females, are consistent with evidence suggesting ADHD may increase risk for obesity.^{9, 14, 15, 17} In analyses of >60,000 youth studied in the NSCH 2003, unmedicated US youth aged 5 to 17 years with ADHD had odds of obesity 1.51 times those of youth who did not have ADHD.¹⁷ Also in NSCH 2003, gender-stratified analyses by Kim et al. found that the odds of obesity for females (6 to 17 years) with unmedicated ADHD were 1.85 times those of females without ADHD, whereas the odds ratio for males was 1.42.¹⁵ NSCH 2003 is a nationally representative cross-sectional study of US youth. Parents (or guardians) of children under age 18 years are interviewed by telephone, and the parent or guardian most knowledgeable about the child’s health is asked to respond to a series of interview questions that take approximately 25 minutes to administer.³⁹ Weight status is based on parent-report of the youth’s height and weight. The accuracy with which parents can estimate their growing child’s height and weight is uncertain, but particularly for young children error is likely.⁴⁰ The large sample size of the NSCH results in greater precision in estimates as compared to NHANES, but NHANES has a higher response rate than NSCH and its methodology may be less subject to measurement error.^{28, 39} Our analyses of youth studied in NHANES 2001–2004 complement what is known about the associations between ADHD and obesity from analyses of NSCH 2003.^{15, 17} Both NSCH and NHANES are nationally representative, but anthropometric measurements are taken directly in NHANES, reducing the potential for misclassification of youth’s weight status. In addition, NHANES administered a DSM-IV structured diagnostic interview to determine the mental health status of children and adolescents. We found, as did Froehlich et al.,¹ that among youth who met diagnostic criteria for ADHD by the DISC-IV, approximately half were reported by their parent to have been previously diagnosed with ADHD.

We chose to combine information from the DISC-IV and parent-report to identify youth with ADHD. This combined method includes youth with current or past-year symptoms that would qualify them for an ADHD diagnosis but who have not been previously diagnosed. It also includes youth who have been diagnosed with ADHD but whose symptoms have been effectively managed through treatment. As a result, the overall prevalence estimate of 15.6% that we report, which includes current and previously diagnosed ADHD, is higher than prevalence estimates based only on the DISC-IV¹ or only on parent-report.¹⁷

In these analyses we have investigated the influence of medication-use and gender on the association between ADHD and obesity. Although for both males and females the prevalence of obesity among youth who are medicated for ADHD is lower than the prevalence for youth with ADHD who are not taking medication, the strength of this association is stronger for males. Males treated with medication for ADHD were less than half as likely as males without ADHD to be obese (9.2% vs. 19.3%), and this estimate was not sensitive to adjustment for covariates. Among females with ADHD, the prevalence of obesity in those who were medicated was lower than for those who were not, but the difference was less pronounced than for the males (17.8% vs. 24.0%).

There is some evidence that the association of ADHD with obesity may depend upon gender. In our analyses, males with unmedicated ADHD had a similar prevalence of obesity to males without ADHD (20.3% vs. 19.3%, adjusted OR=1.02), whereas females with unmedicated ADHD were more likely to be obese than females without ADHD (24.0% vs. 17.3%, adjusted OR=1.54). These results are not conclusive given the wide confidence intervals, but they suggest that ADHD may be differently associated with obesity in male and female youth. Analyses of German youth also support the possibility of gender differences in associations between ADHD and obesity.²⁶ In their gender-stratified analyses of NCHS 2003, Kim et al. found that although the prevalence of obesity was higher for males than females overall, the odds of obesity associated with having ADHD and being unmedicated compared to not having ADHD were higher for females than for males.¹⁵ This may be related to the higher prevalence of the inattentive subtype of ADHD observed in females versus the hyperactive-impulsive subtype which is more prevalent among males.⁴¹ Persons with the inattentive type of ADHD are often described as “hypoactive”, potentially contributing to reduced energy expenditure.²⁰ As has been suggested previously, the association between ADHD and obesity may be a result of genetic and environmental influences (e.g., prenatal maternal smoking) and/or dopamine insufficiency.²² As reviewed by Davis,⁷ both obesity and ADHD have been found to be associated with impulsivity and deficits in self-regulation. How these and other factors manifest differently in females and males warrants further investigation.

Our study has a number of limitations. As a cross-sectional study, it does not permit conclusions to be drawn about temporality or causality. We analyzed data from the most recent waves (2001–04) of NHANES in which youth ADHD was assessed using a structured diagnostic interview; it is possible that associations between ADHD and obesity in the US adolescent population have changed. We categorized youth reported by their parent to have taken medication during the past year “for being overactive, being hyperactive, or having trouble paying attention” as ADHD-medicated. This group includes 20 youth who were not reported by their parent to have been diagnosed with ADHD and who did not meet diagnostic criteria for ADHD on the DISC-IV. We did not investigate the type(s) of medication that youth with ADHD were taking, although the most commonly prescribed medications for ADHD in youth are stimulants.^{25, 42} Also, the age at which children were diagnosed with ADHD or began taking medication for ADHD is not known; some children who either recently began taking medication or recently stopped could have been misclassified. It is possible that males might receive medication for ADHD at an earlier age than females which could explain why the difference in obesity prevalence between medicated and unmedicated youth with ADHD was larger for males than females. Although the precision of our estimates was limited by the clustered survey design, and particularly among females, the sample size, the magnitude of the association we observed between ADHD and obesity for unmedicated females (OR = 1.54) is meaningful and supports the need for further study. Perhaps because ADHD is more prevalent among males than females,⁴ few studies of the association of ADHD with obesity have focused on gender differences. Future research examining the unique risk factors that females with ADHD experience may broaden our understanding of the complex interactions between ADHD and weight status.

Acknowledgments

Abbreviations

Footnotes

References