Introduction

ADHD and body composition: general overview

Attention-deficit/hyperactivity disorder (ADHD) is a common neurodevelopmental disorder characterized by impaired and persistent symptoms of inattention and/or hyperactivity/impulsivity [1]. Besides other externalizing and internalizing psychiatric disorders, obesity has been reported as a common ADHD comorbidity (Cortese 2019). In general, obesity causes damage to the health of individuals, such as breathing difficulties, dyslipidemia, type 2 diabetes, and certain cancers [2]. In individuals with ADHD, obesity leads to greater impairment in aspects related to binge eating, sleep, psychiatric disorders, and somatic diseases, such as asthma, for example, in which it can impact the concomitant increase in symptoms [3, 4].

The relationship between ADHD and obesity is enhanced by the consequences of ADHD first choice medication (methylphenidate) treatment. Methylphenidate is a stimulant medicine acting on dopaminergic and noradrenergic transmission systems, with an important reduction in ADHD symptoms [5]. It has already been shown to reduce the appetite, to have an anorexigenic effect [5, 6] as well as to be associated with reduced levels of serum adiponectin in children with ADHD after 2 months of methylphenidate treatment [7]. In this sense, the comorbidity between ADHD and obesity has implications in the treatment of both conditions. Many studies suggest that comorbid ADHD negatively impacts the treatment of obesity. On the other hand, ADHD treatment might significantly increase the effectiveness of weight management strategies [5].

Body mass index (BMI) is a good nutritional profile predictor and an indicator of lifestyle widely used in clinical and epidemiological investigations [2], including studies involving obesity and ADHD. The BMI has advantages as a practical, low cost, and easy to interpret method, but it does not reflect the distribution of adipose tissue to discrete compartments within the human body. In general, higher fat mass (FM) is associated with differential risks for the development of cardiovascular and metabolic diseases [8]. Few studies have aimed to explore the relationship between ADHD and other measures of body composition. In children, ADHD symptoms were observed to predict higher FM using the dual X-ray absorptiometry (DXA) technique [9]. In opposite, in another study, higher scores of hyperactivity/inattention were associated with lower percentage body fat using bioimpedance measures [10]. In adults with ADHD, this relationship has not been evaluated so far.

A causal relationship between ADHD and obesity

The bidirectionality in the association between ADHD and obesity was suggested, although the findings are mixed. ADHD was observed to chronologically precede the association with obesity among children/adolescents [9, 11, 12] and adults [13], suggesting that ADHD might contribute to obesity through symptoms of impulsivity and inattention, which could thereafter lead to excessive consumption or difficulties following food patterns [11, 13,14,15]. Disordered eating patterns including emotional overeating and impulsive eating, common among ADHD individuals, have been linked to higher consumption of snack foods and sugar-sweetened beverages, which have in turn been linked to higher adipose tissue in older children and adolescents [16, 17]. On the other hand, the association in the opposite direction has also been demonstrated in children [18] and a recent Mendelian Randomization showed that BMI was able to predict ADHD [19]. According to this hypothesis, ADHD could be a consequence of obesity [19], involving mechanisms related to sleep problems, inflammation, and chronic hyperglycemia-induced obesity [13, 15, 20].

Despite the number of studies showing the association between these conditions, few aimed to explore the different aspects related to the symptoms of inattention and hyperactivity and their relationship with obesity through the life cycle. Among the few studies, hyperactive presentation was previously associated with higher BMI [21]. This presentation is also more engaged in gambling, alcohol addiction, and smoking and binge eating problems [22], suggesting a role of the reward system in the relationship between ADHD and obesity [3]. On the other hand, inattention symptoms could be associated with obesity given the difficulties in the organization to follow eating patterns and schedules [22]. Gaining insight into this link is highly relevant from a public health perspective, given the epidemic of obesity and the increased risk of mortality associated with this condition. Thus, here we aimed to investigate the bidirectional relationship over time between ADHD and body composition measurements (BMI, FM, and fat-free mass [FFM]) in the 1993 Pelotas (Brazil) birth cohort.

Materials/subjects and methods

Participants

This study was performed in Pelotas, a southern Brazilian city (340,000 inhabitants). In 1993, all maternity hospitals in the urban area of the city between January 1 and December 31, 1993, were visited daily and all births were identified. In that year, 5265 births were recorded, and 5249 mothers agreed to take part in the study. Subsamples of the cohort have been evaluated since the perinatal visit, and at 11 years the first attempt to interview the full cohort was made. This study includes data of 11-, 15-, 18-, and 22-year follow-up, with 87.5%, 85.7%, 81.3%, and 76.3% attrition rates, respectively. At the 11-year follow-up, the interviews were carried out in the households, and at 15-, 18-, and 22-year follow-up, the adolescents were interviewed at the headquarters of the study. The height and weight were measured at all ages, and FFM and FM at 18 and 22 years. The variables considered as confounder factors were collected at the cohort inception and 18, and 22 years of age follow-up depending on the analysis. In the current investigation, 4451 individuals were studied at 11 years, 4326 at 15 years, 4106 at 18 years, and 3810 at 22 years. The cohort methodology has been described elsewhere [23, 24] and is presented in Supplementary Fig. 1.

ADHD

Assessments of ADHD at 11 and 15 years were performed using the Strengths and Difficulties Questionnaire (SDQ), which was applied to the parents or caregivers during the follow-up and administered by trained psychologists. The SDQ is considered a screening questionnaire, which measured 25 psychological conditions divided into five scales: inattention/hyperactivity symptoms, conduct problems, emotional symptoms, peer relationship problems, and prosocial behavior. The SDQ was validated and adapted to the Brazilian population of children and adolescents aged 4–16 years [25]. For this study, we used the inattention/hyperactivity subscale, which comprises five questions, and implemented the outcome as several symptoms, ranging from 0 to 10 points.

Assessments of ADHD at 18 and 22 years were performed using the Mini International Neuropsychiatric Interview (MINI) [26], version 5.0, applied by trained psychologists. MINI is a short, structured diagnostic interview validated in Brazil [27, 28] that explores major psychiatric disorders according to diagnostic and statistical manual of mental disorders, version 4 (DSM-IV) and international classification of diseases, version 10 (ICD-10). The diagnosis of ADHD followed the DSM-V [1] criteria, which require a significant impairment of symptomatic severity in the individual’s life for the diagnosis of mental disorders. At 18 years, any subject with two or more positive answers to the first six questions was considered positive for screening and answered 12 additional questions about the 12 remaining ADHD symptoms, and we used symptoms of ADHD ranging to 0 at 18 points. At 22 years, we used ADHD measures as presence or absence, as well as a general symptom scale ranging to 0 at 18 points (for all symptoms). At the same age, we also used the scale of symptoms of inattention and hyperactivity, ranging from 0 to 9 points each.

Body composition

The body composition measures included in this study were BMI for all ages and FFM and FM at 18 and 22 years. At 11 and 15 years, the weight was measured using a SECA digital scale (Birmingham, UK) with 100 g precision. However, at 18 and 22 years the weight was measured through the air displacement plethysmography (BOD POD Composition System, COSMED, Albano Laziale, Italy). Height was measured using a portable aluminum stadiometer with 1 mm precision for all ages. BMI was calculated by taking the person’s weight in kilograms divided by the square of the height in meters squared [(kg)/(m2)]. For ages 11, 15, and 18 years, we used the BMI-for-age score-z. The FFM and FM measured were assessed by DXA and expressed as a percentage. The equipment was calibrated at the beginning of each working day, following the manufacturer’s recommendations. To perform the measurements, participants remained in the supine position, barefoot and wearing light and tight-fitting clothes, without earrings, piercings, or any metallic objects.

Covariates

The inclusion of covariates was based on the literature as being relevant for ADHD and body composition [29,30,31,32]. The following maternal and family characteristics at birth were included: (i) maternal characteristics: pre-gestational BMI (continuous, in grams), maternal smoking and alcohol consumption during pregnancy (no; yes), type of delivery (vaginal; cesarean), and family income (in the minimum wage: 1.0 or less; 1.1–3.0; 3.1–6.0; 6.1–10.0; more than 10); and (ii) individual characteristics: sex (male; female), birth weight (continuous, in grams), and race (Black Brazilians; Other [including mixed-race Brazilians and Asian Brazilians]; White Brazilians). The race or color was self-reported by the participant, according to the criteria of the Brazilian Institute of Geography and Statistics [33]. The Mixed ancestry represents a diverse range of skin colors and ethnic backgrounds with a skin tone darker than white and lighter than black. Also, at 22-years, an education level (continuous, in years) and the presence of other mental disorders (Self Report Questionnaire [SRQ-20]; continuous, ranging 0 to 19 points) were used as possible confounding factors. The SRQ-20 is an instrument that allows the tracking of non-psychotic mental disorders and has already been validated in Brazil [34].

Statistical analysis

First, we investigated the bidirectional relationship over time between ADHD and obesity using four follow-up assessments (11-, 15-, 18-, and 22-year follow-up) in a cross-lagged panel model (CLPM) analysis. This analysis tests to what extent mean exposure scores (e.g., ADHD symptoms) at time 1 predicts a certain attitude or behavior (e.g., BMI) at time 2 (a cross-lagged effect) and vice versa. A significant cross-lagged effect is often interpreted as an indication of causality of media exposure at time 1 to behavior at time 2 (and vice versa). Crude and adjusted analyzes were performed using ADHD in continuous form and considering the points at 11 and 15 years, using SQD score and BMI information, and at 18 and 22 years, using MINI symptoms and BMI variables. For the adjusted model, we standardized using a linear regressing model, the effects of pre-gestational BMI, maternal smoking and alcohol consumption during pregnancy, type of delivery, family income, sex, birth weight, and race as covariables, and their residuals were standardized and incorporated in this model, and the results are expressed in the beta coefficient (β) and their respective 95% confidence interval (95% CI).

To further explore the findings of CLPM analysis, we tested the association of the main findings of ADHD in childhood and body composition in adulthood. For this, we performed crude and adjusted multiple linear regression using several symptoms of ADHD (SDQ-hyperactivity/inattention subscale) at 11 years as exposure and BMI and body composition at 18 and 22 years as an outcome. For this analysis, we used the BMI as a continuous data (kg/m2), and for the adjusted model we considered the maternal and individual characteristics at birth for both analyses. The results are expressed in β and their respective 95% CI.

To test the association between ADHD and body composition at 22 years of age, we performed a cross-sectional analysis using ADHD as exposure and BMI (original information; kg/m2) and body composition measures as an outcome. In this analysis, the presence of ADHD (binary form) and scales of general, inattention, and hyperactivity symptoms assessed by MINI were considered as exposure variables. We use the maternal and individual characteristics at birth and at 22 years of age (i.e. education level and the presence of other mental disorders) as covariables in the adjusted model. The results were expressed through the β (with β < 0.05 being considered significant) and their respective 95% CI. Data analysis was performed using Stata (Stata/SE 14.0) and the lavaan package in R for the CLPMs approach.

Ethics committee

The 1993 Pelotas (Brazil) birth cohort study obtained ethical approval from the Medical School Ethics Committee of the Federal University of Pelotas, and full informed consent was provided by the cohort members or by their parents when individuals were younger than 18 years.

Results

Sample characteristics

Maternal and individual characteristics of the sample at birth and 11–22 years are presented in Table 1. The association between the covariables and ADHD and body composition measures are shown in Supplementary Tables 1 and 2. At 22 years, the means of BMI, FFM, and FM were 25.2 kg/m2, 69.7%, and 30.0%, respectively. The median of the number of ADHD symptoms was 4.0 (interquartile range [IQR] 7.0; 2.0), 4.0 (IQR 6.0; 1.0), 1.0 (IQR 7.0; 1.0), and 4.5 (IQR 8.0; 2.0) at 11, 15, 18, and 22 years, respectively.

Table 1 Description of the sample according to maternal characteristics at birth, and individual characteristics at 11, 15, 18, and 22 years of age among participants in the 1993 Pelotas (Brazil) birth cohort.
Full size table

Cross-lagged panel models (CLPM)

After the inclusion of covariables in the analysis, ADHD-SDQ scores (β = 0.573, 95% CI [0.552, 0.595]) and BMI (β = 0.835, 95% CI [0.826, 0.845]) remained stable over 11–15 years. Longitudinal results showed that higher ADHD scores at age 11 slightly predicted higher BMI at age 15 (β = 0.055, 95% CI [0.036, 0.072]) (Fig. 1a). For 18–22 years, the ADHD-MINI symptoms remained stable (β = 0.374, 95% CI [0.343, 0.405]), as well as BMI values (β = 0.836, 95% CI [0.826, 0.846]) (Fig. 1b).

Fig. 1: Effect size (beta coefficient) of the bidirectional association tests between ADHD symptoms and body mass index estimated by cross-lagged panel models among participants in the 1993 Pelotas (Brazil) Birth Cohort.
figure 1

a Using data of 11- and 15-year of age follow-ups and b using data of 18- and 22-year of age follow-ups. BMI body mass index, ADHD attention-deficit/hyperactivity disorder, SDQ Strengths and Difficulties Questionnaire (ranging from 0 to 10 points), MINI Mini International Neuropsychiatric Interview (ranging from 0 to 18 points). Adjusted for maternal characteristics at birth (pre-gestational BMI, maternal smoking and alcohol consumption during pregnancy, type of delivery and family income) and individual characteristics (sex, birth weight, and race).

Full size image

ADHD in childhood and body composition in adulthood

ADHD symptoms at 11 years and body composition in adulthood are presented in Table 2. After adjustment for maternal and individual characteristics at birth, we found that the increase of ADHD symptoms at 11 years is associated with higher BMI at 18 (β = 0.04, 95% CI [0.03, 0.05] and 22 years β = 0.17, 95% CI [0.11, 0.23]) and FM at both 18 (β = 0.11, 95% CI [0.02, 0.21]) and 22 years (β = 0.12, 95% CI [0.06, 0.29]). We also observed a decrease in FFM at 18 (β = −0.12, 95% CI [−0.21, −0.02]) and 22 years (β = −0.12, 95% CI [−0.28, −0.05]) according to the higher number of ADHD symptoms at 11 years (Table 2).

Table 2 Multiple linear regression results for crude and adjusted models testing the association between ADHD symptoms at 11 years and BMI and body composition at 18 and 22 years of age among participants in the 1993 Pelotas (Brazil) Birth Cohort.
Full size table

Cross-sectional analysis

In the adjusted analysis at 22 years, we observed a decrease in the FFM percentage (β = −1.72, 95% CI [−3.36, −0.09]), and the increase in the FM percentage (β = 1.79, 95% CI [0.07, 3.51]) among individuals with ADHD (Table 3). Using the scales of symptoms for ADHD, we found an increase in BMI with an increase in general (β = 0.06, 95% CI [0.004, 0.12]) and hyperactivity (β = 0.15, 95% CI [0.05, 0.25]) symptoms, after adjustment. Inattention symptoms of ADHD were not associated with any measure of body composition (Table 3).

Table 3 Multiple linear regression results for crude and adjusted models testing the association between the presence of ADHD, ADHD symptoms (general, inattention, hyperactivity scales), and BMI and body composition at 22 years of age among participants in the 1993 Pelotas (Brazil) birth cohort.
Full size table

Discussion

In this study, we found evidence of the association between ADHD and body composition measures using several approaches. Our results also suggested that ADHD symptoms at 11 years may predict higher BMI in adolescence (i.e., at 15 years), but an opposite relationship was not found. An increased number of symptoms at this age may be a determining point for body composition, with an increase in BMI and FM percentage, as well as a decrease in FFM percentage in adulthood (i.e., at 18 and 22 years). In adulthood, ADHD presence and ADHD symptoms were associated with worse scores of all evaluated measures and suggested that hyperactivity symptoms may have a higher effect on obesity.

We observed that the presence of ADHD at 22 years of age was associated with lower FFM (which considers muscle mass in conjunction with bone) percentages, as well as higher FM (adipose tissue) percentages in adults. Although several studies demonstrated the association between ADHD and obesity in a cross-sectional approach between childhood [31] and adults [3, 29, 35,36,37] in healthy samples, few studies seek to understand how different aspects of body composition can affect this relationship. The few studies reporting different measures of body composition have shown that, in childhood, higher scores of hyperactivity/inattention were associated with lower percentage body fat (β = –0.20, 95% CI [0.37, –0.03], p = 0.02) using bioimpedance in a cross-sectional design performed in 450 preschool children at 4–6 years of age [10] and, in the opposite, in another study, ADHD at 6 years of age increased 0.22 kg of FM at 9 years of age (95% CI [0.11, 0.28], p < 0.001) in children belonging to the Generation R birth cohort, using the DXA [9]. As far as we know, this is the first study that aims to investigate the association and ADHD and other measures of body composition in addition to BMI in adults, partially limiting the comparison of our findings. The results are more similar to the latter, showing that ADHD is associated with worse body composition indexes, mainly with higher FM and lower FFM. It is suggested that the distribution of adipose tissue to discrete compartments, as well as the type of measure, is associated with differential risks for some cardiovascular and metabolic diseases [8], then the association observed by us may also indicate that ADHD may also be a differential risk for metabolic diseases.

Interestingly, even though we did not find an association between ADHD presence and BMI at 22 years of age, the number of symptoms of ADHD was associated with higher BMI, suggesting that probably the severity of the disease might influence the BMI in this age. Our results also suggest that the subtypes of ADHD may have different influences on BMI, with a greater impact on symptoms of hyperactivity on body composition at 22 years. The main role of hyperactive and impulsive symptoms was also demonstrated by a recent study including individuals from 10 to 22 years in the Netherlands, in which the authors found an association of BMI with hyperactivity and impulsivity throughout adolescence, but not with inattention [38]. A study carried out with 110 college student women between 25 and 46 years old, using a structural equation model, observed that the symptoms of ADHD and impulsivity correlated with eating behaviors, including binge eating and emotional eating, leading to the consequent increase in BMI [3]. Similar results were observed in a replication study carried out in adult men aged 25–50 years old [35]. It has been suggested that compulsion related to eating disorders could contribute to or manifest itself as the impulsivity common to ADHD or could trigger symptoms of hyperactivity among these individuals [39]. Binge eating, as well as substance abuse and gambling, could reflect a biological change in the common reward system among individuals with ADHD, mainly those with hyperactive presentation [40]. Thus, the association observed here may reflect an alteration in the reward system; however, it does require further studies to better understand the relationship between ADHD presentation and higher risk for obesity.

The causal relationship between these two traits has been the focus of several studies [9, 11,12,13, 38]. Although many approaches were used to detect the direction of this association, the results do remain inconclusive so far, with evidence for both directions, or lack of association [38]. In the present study, using CLPM, we observed a causal effect of higher ADHD symptoms at 11 years of age on higher BMI at 15 years. Importantly, we also showed symptoms of ADHD at 11 years leading to higher BMI and FM at 18 and 22 years. This direction has been demonstrated in several longitudinal studies. A study conducted in Finland showed that ADHD at 8 years of age was directly related to a general obesity increase at 16 years [11]. Following the same direction, Schwartz et al. [12] through telephone records showed higher BMI averages over time among children and adolescents (from 3 to 18 years old) followed up annually, which corroborated by our findings. However, Kase et al. [38] related no longitudinal direct effects were found between ADHD symptoms and BMI.

The mechanisms involved in the relationship between ADHD and BMI seem to be complex and multifactorial. Several mechanisms linking both etiologies are suggested and demonstrated. For instance, it has been seen that there is a genetic sharing between traits, and modifications common to both conditions may underlie the association, such as impairment in the reward system [19, 41]. Besides, there is a suggestion that excessive food intake, difficulty in adhering to a healthy eating pattern, and organizational and attention difficulties among individuals with ADHD could trigger compensatory mechanisms leading to binge eating and reduction of caloric expenditure. Our results suggest that ADHD in early life can be considered a critical point for body composition programming. Probably, when it occurs in childhood, ADHD may model the eating patterns that affect late body composition. Although research on ADHD and early life endocrine function are scarce, this finding could partially be due to underlying associations between ADHD symptoms and endocrine dysregulation, such as the underproduction of adiponectin (i.e., appetite-reducing hormone synthesized in fat tissue) [42] as well as disparate conditions of insulin resistance [43]. It has been reported that serum adiponectin levels are decreased in children with ADHD after 2 months of methylphenidate treatment, suggesting a possible involvement of adiponectin in neurobiological mechanisms related to eating disorders or gaining weight in children with ADHD [7]. Other study showed that adiponectin level decreases in the case of obesity, and this alteration might be related to an increase in appetite and weight gain [7].

Our results should be interpreted considering some limitations. Although we used data from a large base population cohort, we had a limited number of ADHD cases due to the population source and the nature of the disease. We could not test the bidirectional association over time (i.e., follow-up at 11 and 22 years) through CLMP, because ADHD was not measured in the same way at all follow-ups, preventing formal assessment of the direct impact of the association between childhood and adulthood (i.e., between 11 and 22 years old). Furthermore, the diagnostic process in population studies differs from the gold standard for diagnosing ADHD, that is, clinical evaluation. To minimize this effect, a structured interview was applied by trained psychologists at the age of 22, which has been demonstrated to improve diagnostic accuracy. Moreover, we also performed our analysis using dimensional analyses, using the number of symptoms to try to minimize the recall bias. As for 18 years of age, there was a screening, and therefore, the full assessment was not questioned for the total population, we also tested the counting of the number of screening symptoms (0–6 points). There were no statistically significant changes compared to the analyses including 0–18 points (data available upon request). Finally, the use of ADHD medication may be considered a confounder for the association analysis in this study. However, the information concerning these variables was not assessed in the cohort.

Among the strengths of this study, we should highlight that our findings are based on analysis with a longitudinal design, which allowed us to explore the bidirectional pathways involved in the relationship between ADHD and obesity. Moreover, we investigate the association with other body compositions beyond BMI, mainly the adult population, still little explored in other studies. Furthermore, we use a Brazilian-based population cohort with high follow-up rates, which has comprehensive data on standardized and well-collected means of mental health and body composition, allowing us to better explore this relationship.

In conclusion, we found a positive association between ADHD and body composition in this cohort, with a greater effect of hyperactivity symptoms in this relationship and an important contribution to other body composition measures besides BMI. In addition, our data demonstrated evidence that ADHD predicted a higher BMI, and higher scores on the ADHD scale at age 11 years may a critical point for body composition in the future.