Body Mass Index and Attention-Deficit/Hyperactivity Disorder in U.S. youth: a reverse J-shaped association

Technical Summary: Body Mass Index and Attention-Deficit/Hyperactivity Disorder in U.S. Youth

Problem Statement
Attention-deficit/hyperactivity disorder (ADHD) is a prevalent neurodevelopmental disorder affecting approximately 9% of U.S. children and adolescents. While metabolic and inflammatory pathways are increasingly implicated in ADHD pathophysiology, the relationship between Body Mass Index (BMI) and ADHD risk remains contradictory. Prior epidemiological studies have yielded inconsistent findings, ranging from positive associations between obesity and ADHD to null or inverse relationships. The authors posit that these inconsistencies stem from methodological limitations, specifically the reliance on linear dose-response models that fail to detect non-linear inflection points or thresholds where the direction or magnitude of risk changes. Furthermore, the biological mediators linking adiposity to ADHD, particularly those involving inflammation and iron metabolism, remain poorly characterized.

Methodology
This study utilized a nationally representative dataset from the National Health and Nutrition Examination Survey (NHANES), pooling cycles from 1999–2004 and 2013–2018. The analytic sample consisted of 18,383 children and adolescents (<18 years) with complete data for BMI and ADHD status.

• Exposure and Outcome: BMI was calculated from standardized physical examinations and treated as both a continuous variable and categorized into tertiles. ADHD status was ascertained via two distinct, validated proxy measures depending on the survey cycle: parental/self-report of a clinical diagnosis (1999–2004) and documented use of ADHD-specific prescription medications (2013–2018).
• Statistical Analysis: The authors employed survey-weighted logistic regression to estimate odds ratios (ORs) while adjusting for a comprehensive set of covariates, including age, sex, race/ethnicity, poverty-income ratio (PIR), physical activity, secondhand smoke exposure, maternal smoking during pregnancy, and WIC participation.
• Non-linearity and Threshold Detection: To move beyond linear assumptions, the study utilized Generalized Additive Models (GAM) with restricted cubic splines (RCS) and two-piecewise logistic regression. These methods were used to identify a data-driven inflection point (threshold) where the association between BMI and ADHD changes.
• Mediation Analysis: The study quantified the proportion of the BMI-ADHD association mediated by Red Cell Distribution Width (RDW), a biomarker for systemic inflammation, oxidative stress, and iron deficiency, using a nonparametric bootstrap approach.
• Robustness Checks: Sensitivity analyses were conducted using alternative ADHD definitions (questionnaire-only vs. medication-only) and stratified analyses across demographic subgroups (sex, age, race/ethnicity) to test for effect modification.

Key Results

• Inverse Dose-Response: In fully adjusted models, higher BMI tertiles were associated with lower odds of ADHD compared to the lowest tertile (High vs. Low: OR = 0.64; 95% CI 0.52–0.78; P-trend < 0.001).
• Reverse J-Shaped Threshold: GAM and piecewise regression revealed a significant non-linear, "reverse J-shaped" relationship with an inflection point at BMI = 21.03 kg/m².
  • Below Threshold: For every 1 kg/m² increase in BMI below 21.03 kg/m², the odds of ADHD decreased by 9% (OR = 0.91; 95% CI 0.88–0.95).
  • Above Threshold: Above 21.03 kg/m², the association between BMI and ADHD was null (OR = 1.00; 95% CI 0.98–1.02).
  • The threshold model provided a significantly better fit than a simple linear model (LRT P < 0.001).
• Mediation by RDW: RDW mediated approximately 11.4% (95% CI 0.1–47.8%, P = 0.044) of the total association between BMI and ADHD, suggesting that inflammation-iron pathways contribute to the observed relationship.
• Subgroup Heterogeneity: Significant effect modification was observed by sex and race/ethnicity.
  • Sex: The protective threshold effect was pronounced in males (OR = 0.89 below threshold) but absent in females, who showed a modest positive association.
  • Race/Ethnicity: Non-Hispanic White and Other Hispanic youth exhibited significant threshold effects, whereas Non-Hispanic Black and Mexican American subgroups showed null associations.
• Sensitivity: The threshold pattern remained consistent across different ADHD definitions and sensitivity analyses.

Key Contributions

1. Identification of a Non-Linear Threshold: This is the first study to identify a specific BMI threshold (21.03 kg/m²) characterizing a reverse J-shaped association between BMI and ADHD in a nationally representative U.S. pediatric sample. This finding offers a potential explanation for the contradictory linear results in prior literature.
2. Mechanistic Insight: The study provides evidence that Red Cell Distribution Width (RDW) mediates a portion of the BMI-ADHD link, implicating inflammation and erythrocyte homeostasis as novel mechanistic targets in the metabolic-neurocognitive pathway.
3. Demographic Specificity: The research highlights that the BMI-ADHD relationship is not uniform, varying significantly by sex and race/ethnicity, which challenges the application of a single linear model across diverse populations.

Significance and Claims
The authors claim that their findings challenge the prevailing linear paradigm in obesity-mental health research. They suggest that the "protective" effect of modest adiposity (below the threshold) may reflect improved nutritional status or neurodevelopmental resilience, while the null association above the threshold suggests that the adverse neuroinflammatory consequences of excess adiposity counterbalance these benefits.

The study posits that RDW-mediated inflammation-iron pathways represent a modifiable biological mechanism, potentially offering targets for intervention (e.g., iron supplementation, anti-inflammatory diets). However, the authors maintain a modest stance regarding the mediation effect, noting the wide confidence interval (0.1%–47.8%) and acknowledging that approximately 90% of the association operates through pathways not captured by RDW.

The paper concludes that while pediatric BMI screening may inform mental health risk stratification, it must be applied with precision, recognizing specific adiposity ranges and demographic subgroups. The authors emphasize the need for future longitudinal studies to confirm causality and mechanistic research to clarify why excess adiposity ceases to be protective beyond the identified inflection point.