Sex and age genetic overlap between body mass index and attention deficit hyperactivity disorder in twins

INTRODUCTION

Childhood obesity is a major and growing public health concern worldwide, and is associated with broad health and developmental consequences across the life course^[1]. According to the World Health Organization, more than 390 million children aged 5-19 years were overweight or obese in 2022, marking a sharp increase over the past few decades^[2]. In China, rapid urbanization and socioeconomic transitions have contributed to a rising prevalence of overweight and obesity among children aged 6-17 years, reaching 19%^[3,4]. As a result, China now has the largest population of children with obesity worldwide, posing a substantial metabolic and public health burden^[5,6].

The etiology of obesity is increasingly recognized as heterogeneous, reflecting complex interactions among biological, behavioral, and environmental factors^[7]. While traditional factors such as diet, physical activity, and lifestyle have long been emphasized, emerging evidence highlights the contribution of neurobehavioral characteristics in shaping individual differences in obesity risk^[8,9]. In recent years, the association between attention-deficit/hyperactivity disorder (ADHD) and obesity has garnered increasing attention^[10-12]. ADHD is among the most prevalent childhood neurodevelopmental disorders. It is characterized by inattention, hyperactivity, and impulsivity, and affects approximately 8.0% of children worldwide^[13] and 6.4% in China^[14]. A systematic review reported a 40% higher prevalence of obesity among children with ADHD^[10], and a recent meta-analysis further supported an increased risk of overweight and obesity (odds ratio 1.56), with higher estimates being observed in Asian populations (3.25)^[11]. This comorbidity is associated with compounded adverse health outcomes and an increased healthcare burden^[15,16].

Despite growing epidemiological evidence supporting the co-occurrence of ADHD and obesity, the etiological mechanisms underlying this association remain incompletely understood. Both conditions are highly heritable, with estimates of approximately 40%-75% heritability for obesity^[17,18] and 30%-80% for ADHD^[19,20]. A recent genome-wide association meta-analysis reported a significant genetic correlation between ADHD and obesity (r_g = 0.22) and body mass index (BMI; r_g = 0.26)^[21], suggesting shared genetic liability. At the neurobiological level, dysregulation of dopaminergic pathways involved in appetite, motivation, and reward processing has been proposed as a potential common mechanism linking ADHD symptoms and obesity risk^[22]. Genome-wide association and candidate gene studies have identified several loci associated with both conditions, including fat mass and obesity-associated (FTO), melanocortin-4 receptor (MC4R), brain-derived neurotrophic factor (BDNF), and synaptosomal-associated protein 25 (SNAP25), implicating pathways related to energy metabolism, neurotransmitter signaling, and synaptic plasticity^[23-26]. Environmental factors, such as family socioeconomic status, parental education level, and parental mental health, also contribute to this comorbidity^[27]. In addition, children with ADHD exhibit distinct dietary characteristics, such as higher consumption of junk food^[28], unhealthy eating patterns^[29], and faster eating rates^[30]. Although hyperactivity is a core feature of ADHD, children with this condition engage in less daily physical activity (21% lower odds)^[31] and face barriers to participation in school activities^[32]. Sleep disturbances, including shorter total sleep time and lower sleep efficiency, are also more prevalent in children with ADHD^[33]. Together, these behavioral factors may contribute to increased BMI and obesity risk.

Twin and family studies provide a powerful framework for disentangling the genetic, shared environmental (C), and unique environmental (E) contributions to complex traits and their co-occurrence^[34]. To date, only a limited number of twin studies have examined the association between ADHD and obesity, reporting modest to moderate genetic correlations^[35-37], with some evidence suggesting stronger associations in females^[38,39]. However, existing research has largely focused on adolescents and adults^[35-38], with limited attention to childhood^[39], a critical developmental period when both ADHD symptoms and obesity risk emerge and evolve. Moreover, genetic and environmental influences on ADHD and obesity may vary across cultural and social contexts^[40-42], underscoring the need for research in Asian populations.

This study aimed to examine the association between ADHD and BMI in a large sample of Chinese twins aged 6-18 years. Specifically, the objectives were to: (1) characterize the phenotypic association between ADHD symptoms and BMI; (2) estimate the relative contributions of genetic and environmental factors to each trait; and (3) assess the shared genetic and environmental influences underlying their overlap, with particular attention to differences by sex and age. By clarifying etiological heterogeneity in the association between ADHD symptoms and obesity risk by sex and developmental stages, this study may help inform more targeted public health prevention strategies.

METHODS

Statistical analysis

Means and standard deviations for ADHD symptoms and BMI were calculated by sex and age group (children aged 6-11 years and adolescents aged 12-18 years). ADHD symptoms were natural-log-transformed before analysis because of their skewed distribution. Group differences were examined using independent‑samples two-tailed t-tests, and P < 0.05 was considered statistically significant. Phenotypic residuals were derived from whole‑sample linear regression models that adjusted for continuous age and sex and were used in all subsequent analyses. Accordingly, sex‑specific and age‑stratified twin models reflected residual variation after adjustment for overall age and sex effects. Generalized estimating equations were used to estimate the association between BMI and ADHD symptoms while accounting for the non-independence of twin observations. Descriptive and regression analyses were conducted using IBM SPSS Statistics version 27.

Twin correlations for ADHD symptoms and BMI were calculated by sex, age group, and zygosity. Higher correlation in MZ than in DZ twins indicated a genetic influence on the trait’s variance. Quantitative genetic analyses were conducted to estimate additive genetic (A), C, and E contributions to each trait using twin modeling approaches^[49]. Univariate ACE models were fitted to estimate variance components for ADHD symptoms and BMI. Bivariate Cholesky decomposition models were applied to assess the extent to which genetic and environmental influences were shared between ADHD symptoms and BMI [Figure 1]. This approach was selected as a flexible, exploratory method to decompose the covariance between traits without imposing strong a priori assumptions about the underlying structure. This was particularly suitable given the limited prior evidence on sex- and age-specific genetic overlap in this population. Quantitative genetic analyses were performed using the structural equation modeling package Mx^[50], adjusting for age and sex. Full ACE models were retained to avoid potential bias associated with model reduction^[51].

Figure 1. Bivariate Cholesky decomposition twin model for ADHD symptoms and BMI. A, C, and E represent the additive genetic, shared environmental, and unique environmental variance components, respectively. a, c, and e denote the corresponding path coefficients. ADHD symptoms and BMI are two phenotypic measures in a single twin. r_g, r_c, and r_e indicate the genetic, shared environmental, and unique environmental correlations between the two traits, respectively. For clarity, only one twin is shown. ADHD: Attention-deficit/hyperactivity disorder; BMI: body mass index.

RESULTS

A total of 1,774 same-sex twins aged 6 to 18 years (mean age, 10.96 ± 3.04 years) were included in the analysis. Descriptive statistics for BMI and ADHD symptoms assessed by the CBCL are presented in Table 1, stratified by age group and sex. Overall, boys showed higher ADHD symptom scores than girls, whereas BMI was higher in boys and adolescents than in children.

Intraclass correlation coefficients for ADHD symptoms and BMI, stratified by sex, age group, and zygosity, are presented in Table 2. Across all sex and age groups, intraclass correlations were consistently higher in MZ twins than in DZ twins for both ADHD symptoms and BMI. In addition, DZ correlations for ADHD symptoms exceeded half of the corresponding MZ correlations across groups.

Univariate ACE models were fitted to estimate the genetic and environmental contributions to BMI and ADHD symptoms [Table 3]. For BMI, A influences accounted for a large proportion of the variance [0.88, 95% confidence interval (CI): 0.72-0.90], whereas E influences accounted for the remaining variance (0.12, 95%CI: 0.10-0.13), with shared environmental estimates close to zero. For ADHD symptoms, A influences accounted for approximately 46% of the variance (0.46, 95%CI: 0.37-0.57), while shared environmental influences accounted for a comparable proportion (0.47, 95%CI: 0.36-0.56), and E influences accounted for a very small proportion (0.07, 95%CI: 0.06-0.08). Similar patterns were observed across sex and age strata, with modest variation in the point estimates.

Associations between ADHD symptoms and BMI were examined, and estimates from the regression and bivariate twin models are presented in Table 4 and summarized in Figure 2. In the full sample, ADHD symptoms were positively associated with BMI (β = 0.29, 95%CI: 0.08-0.50). Subgroup analyses indicated that this association was more pronounced among girls and adolescents, with the strongest association observed in girls aged 12-18 years (β = 0.92, 95%CI: 0.25-1.58). Bivariate twin modeling estimated a modest genetic correlation between ADHD symptoms and BMI in the full sample (r_g = 0.13, 95%CI: -0.01-0.27). Higher and statistically significant genetic correlations were observed in girls (r_g = 0.23, 95%CI: 0.03-0.41) and in children (r_g = 0.17, 95%CI: 0.01-0.33). In adolescents, both genetic and E factors contributed to the association between ADHD symptoms and BMI. The estimated E correlations were relatively larger than those in younger children, although they did not reach statistical significance.

Figure 2. Sex- and Age-Specific Associations of ADHD Symptoms With BMI. Points represent effect estimates, and horizontal bars indicate 95%CI. Phenotypic associations are expressed as regression coefficients (β), while genetic and unique environmental correlations are expressed as r_g and r_e, respectively. Statistical significance is inferred when the 95%CI does not include zero. Sample sizes for each subgroup are provided in Table 1. ADHD: Attention-deficit/hyperactivity disorder; BMI: body mass index; CI: confidence interval.

DISCUSSION

This study investigated the genetic and environmental factors underlying the association between ADHD symptoms and BMI in a large sample of Chinese twins aged 6-18 years. We observed a modest genetic overlap between ADHD symptoms and BMI in the overall sample, with stronger and statistically significant genetic correlations in girls and younger children. These findings are consistent with previous evidence suggesting shared genetic liability between ADHD and BMI and extend the literature by highlighting the importance of sex and developmental stage in shaping this association in a non-Western population.

Our findings provide a detailed characterization of the association between ADHD symptoms and BMI, highlighting variation across sex and developmental stage. We report a statistically significant positive phenotypic association between ADHD symptoms and BMI, consistent with a growing body of literature linking ADHD to obesity and related metabolic outcomes^[10,41,52]. Previous studies have reported mixed evidence regarding sex- and age-specific patterns in this association. For example, meta-analyses by Cortese et al. did not identify significant sex differences^[10], whereas Nigg et al. reported stronger associations in females and adolescents^[53], and Zhu et al. observed a more pronounced association in males, particularly among unmedicated individuals^[11]. In the present study, regression analyses indicated that the ADHD-BMI association was more pronounced in females and adolescents, supporting the presence of sex- and age-related heterogeneity at the phenotypic level. These patterns might partly reflect developmental and psychosocial factors. Weight-related concerns tend to increase during adolescence, particularly among females^[54], and may interact with ADHD-related self-regulatory difficulties, potentially contributing to emotional eating and disordered eating behaviors that reinforce the association between ADHD symptoms and BMI^[42,55]. More broadly, the observed sex differences also suggest that the ADHD-BMI association might be shaped less by the level of physical activity alone than by the balance between energy expenditure and dysregulated energy intake. In this context, the stronger association observed in females may reflect a greater contribution of eating-related and psychosocial pathways, including emotional eating and weight-related concerns. These findings also suggest that hormonal and behavioral mediators might be important targets for future investigation.

Bivariate twin analyses suggest a modest genetic overlap between ADHD symptoms and BMI, with higher and statistically significant genetic correlations observed in girls. This female-specific pattern of shared genetic liability is consistent with prior twin research, including studies of Swedish adolescents that reported stronger genetic overlap in girls^[38], as well as evidence that genetic correlations are present in females across childhood and adolescence^[39]. These sex-specific genetic patterns provide population-level support for hypothesized neurobehavioral mechanisms linking ADHD and obesity risk. Prior molecular genetics and neuroimaging studies have suggested partial overlap in polygenic risk and neural circuits involved in reward processing and self-regulation linking ADHD-related impulsivity with BMI^[37]. Pubertal hormonal changes in females, particularly fluctuations in estrogen, may further interact with these genetic susceptibilities to influence attention regulation, emotional processing, and reward sensitivity, thereby increasing vulnerability to impulsivity-related eating behaviors and obesity risk^[56,57]. By contrast, higher levels of hyperactivity in males may be associated with greater energy expenditure, potentially attenuating weight gain during certain developmental periods^[23]. Together, these findings help contextualize the observed sex differences and highlight specific neurobehavioral pathways that warrant further investigation.

In addition to sex-specific genetic patterns, the relative contributions of genetic and environmental factors also varied by developmental stages. During childhood, genetic factors accounted for a larger proportion of the association in both sexes, whereas environmental contributions were relatively modest. In adolescence, genetic influences remained evident in girls, while E contributions became relatively more pronounced in boys. This sex-specific developmental pattern is consistent with longitudinal evidence indicating that higher ADHD symptom scores in girls were associated with steeper increases in BMI trajectories from childhood into late adolescence^[58]. Developmental changes during adolescence, including increased autonomy and shifting environmental exposures, may interact with neurobehavioral traits related to ADHD and obesity risk. Such interactions occur within a broader context of shared genetic and early-life environmental liability linking ADHD and obesity^[27,35]. Together, these developmental patterns are consistent with broader evidence that hyperactivity-related behaviors in males tend to decline after childhood, whereas impulsivity and inattention in females may persist into adolescence. Such sex-specific developmental differences may coincide with periods of heightened behavioral and metabolic susceptibility^[59], potentially contributing to heterogeneity in obesity risk.

Consistent with previous literature, boys exhibited higher levels of ADHD symptoms than girls, reflecting well-established sex differences in symptom presentation^[59]. Concomitantly, boys and adolescents had higher BMIs, in line with normative developmental patterns of weight gain^[60]. Twin correlations and univariate genetic analyses confirmed substantial genetic contributions to both ADHD symptoms and BMI, consistent with prior twin studies^[61,62]. The heritability estimate for ADHD symptoms in this study was modestly lower than that in our previous report^[43]. This may reflect differences in sample composition, including questionnaire-based zygosity determination and the restriction to same-sex twin pairs. Shared environmental influences were evident for ADHD symptoms but minimal for BMI, suggesting that family and school environments may shape ADHD-related behaviors more strongly than body weight. This pattern may partly reflect contextual influences in collectivist settings such as China, where shared social environments may exert a stronger influence on behavioral traits^[40]. In addition, because ADHD symptoms were assessed using parent-reported CBCL data, with the same parent often rating both twins, the relatively high shared environmental variance for ADHD symptoms may have been inflated by rater effects, such as parental response style or reporting bias, a concern particularly relevant in single-informant designs^[63,64].

Several limitations should be noted. First, ADHD symptoms were assessed using parent-reported CBCL data, which, although widely validated, may be subject to reporting bias and does not provide clinician-based diagnostic confirmation. Therefore, the findings should be interpreted as reflecting dimensional ADHD-related symptoms rather than clinically diagnosed ADHD, which may limit generalizability to clinical populations. In addition, 11.6% of the BMI data were self-reported. Despite a high correlation with measured anthropometrics (r = 0.82), self-report may underestimate BMI because of underreported weight or overreported height, potentially attenuating the observed association. Second, ADHD subtypes were not distinguished, limiting insights into subtype-specific mechanisms. In addition, information on ADHD treatment status was not available, precluding evaluation of treatment effects on the association between ADHD symptoms and BMI. Third, epigenetic processes, gene-environment interactions, and gene-environment correlations were not examined. Therefore, the estimated variance components may not fully capture the complex interplay between genetic liability and environmental exposure, particularly for ADHD symptoms. Fourth, important environmental and behavioral factors related to BMI, such as physical activity and dietary patterns, were not directly evaluated, limiting interpretation of the underlying mechanisms of the association between ADHD symptoms and BMI. Fifth, analyses were restricted to same-sex twin pairs, precluding direct tests of sex differences in genetic architecture using opposite-sex comparisons and potentially limiting the generalizability of the sex-specific findings. Sixth, the sample was drawn from Beijing and Shandong and therefore may lack national representativeness; it did not include urban-rural information, which may limit the broader applicability of the findings. Finally, although this study represents one of the largest twin samples of Chinese children and adolescents to date, the sample size, especially in stratified analyses, may have limited statistical power to detect more nuanced effects.

In summary, this large twin study conducted in an Asian population advances understanding of the association between ADHD and BMI by delineating etiologically heterogeneous pathways shaped by sex and developmental stage. Genetic influences predominated in childhood, whereas in adolescence, genetic effects remained stronger in females, while environmental influences appeared more prominent in males. These findings underscore the importance of incorporating sex and developmental stage into etiological models of childhood obesity risk and support more differentiated prevention strategies. Interventions addressing environmental and behavioral factors may be particularly relevant for boys, whereas strategies emphasizing neurobehavioral regulation may hold greater promise for girls.