Marjolein Luman

Background: Childhood-onset attention-deficit/hyperactivity disorder (ADHD) persists in 28% to 58% of individuals, and is associated with adverse long-term outcomes, like additional psychiatric diagnoses and lower educational attainment. Knowing early life predictors for adverse long-term outcomes might help develop preventive measures, but too few have been identified using conventional statistical models. The current study investigated the potential of machine learning in identifying predictors for ADHD persistence and long-term functioning of adults with childhood-onset ADHD. Method: A total of 133 adults ( M _age = 29.53, SD _age = 2.83, 75% males) from an 18-year prospective cohort study were assessed on ADHD persistence and long-term functioning. Machine learning models were developed using cross-validation, with and without additional resampling, and compared to a conventional logistic/linear regression model. With SHapley Additive exPlanations (SHAP) we explored model behavior. Childhood predictors encompassed ADHD symptoms and treatment, other psychopathology, cognition, somatic characteristics, ADHD polygenic risk score, and parental demographics and psychopathology. Results: Performance of all models for ADHD persistence and functioning (quality of life) was low. Best performing models included the logistic regression model with resampling for ADHD persistence (AUC’s _{all models} = .52–.58; AUC _{logistic regression, resampling} = .58), and the ridge regression model for long-term functioning (R ² ’s _{all models} = -0.36–0.02; R ² ’s _{ridge regression} = .02, RMSE’s _{all models} = 2.07–2.42; RMSE’s _{ridge regression} = 2.07). Conclusions: Findings highlight the complexity of predicting long-term outcomes of ADHD. Model improvement is crucial before models can inform clinical practice, e.g., by increasing sample size and investigating additional predictors.

The effectiveness of behavioral parent training (BPT) techniques for children with attention-deficit/hyperactivity disorder (ADHD) may vary across individuals. Using a microtrial, we studied whether parent-ratings of child “hot” and “cool” executive functions moderated effects of stimulus-control (antecedent-based) versus contingency-management (consequent-based) techniques on child behavioral difficulties. Parents of 92 children with ADHD were randomly assigned to antecedent-based training, consequent-based training or waitlist. Results indicated that consequent-based techniques were particularly effective for children with lower arousal regulation and higher reward responsivity, while antecedent-based techniques were most effective for children with lower sensitivity to aversive stimuli. Possible explanations for these findings are discussed.

Stimulant medication is the first-line pharmacological treatment option for children with attention-deficit/hyperactivity disorder (ADHD). However, long-term effects remain unclear. Therefore, we explored long-term (>12 years) outcomes of childhood stimulant medication use, using data from the Dutch NeuroIMAGE cohort. Childhood (up to Mage=16.14, SDage=3.48) stimulant treatment trajectories were detected using K-means clustering based on age at initiation, stop age, total and maximum daily dose, and duration and variability of stimulant use in 324 children. The trajectories were compared across seven domains of adult outcomes (controlling for multiple testing) including psychiatric status, behavioral and emotional functioning, academic and professional functioning, adaptive functioning, neurocognitive functioning, physical health, and healthcare service use. At follow-up, 143 adults with childhood ADHD (Mage=28.34, SDage=3.74, 65% males) were part of three stimulant treatment trajectory groups: early-and-intense (n=46), late-and-moderate (n=63), and stimulant-naïve (n=34). The three groups did not differ on any of the 51 adult outcomes and findings did not change after controlling for age and sex, except that a significant group difference emerged for callous and unemotional traits (behavioral and emotional functioning domain). More specifically, the early-and-intense and late-and-moderate groups had more favorable scores compared to stimulant-naïve group, although this may partly reflect pre-existing group differences. The main findings did not change after controlling for ADHD medication use at wave IV. Overall, stimulant medication use in childhood did not predict long-term outcomes in those with childhood ADHD in our sample, suggesting that there may be no lasting effects of stimulants on many long-term functional outcomes.

Background: Attention-deficit/hyperactivity disorder (ADHD) is a prevalent neurodevelopmental disorder, with increased risk for negative adult long-term outcomes. These outcomes may differ between males and females with ADHD, but whether these differences are more pronounced in ADHD or similar to those in the general population remains unclear. Method: In an 18-year follow-up of 283 children ( M _age = 28.48) we assessed if sex moderated 51 long-term outcomes across seven domains: psychiatric status, behavioral/emotional problems, academic/professional functioning, adaptive functioning, neurocognition, physical health, and healthcare service use in those with ADHD ( n _females = 50, n _males = 104), and without ADHD ( n _females = 77, n _males = 52). Additionally, we aimed to explore the moderating role of gender expression. Results : Sex interacted with group for social functioning: males with ADHD were more likely to show lower social functioning compared to females with ADHD and controls. There were no sex differences on the other outcomes. As sex and gender expression were highly similar ( r = .94), specific effects of gender expression could not be studied. Conclusions. Generally, sex did not have a more pronounced effect on long-term outcomes of childhood ADHD in individuals with ADHD compared to the general population. Gender expression could not be studied in our predominantly cisgendered sample.

Background: Behavioural parenting programmes are effective for many children with attention-deficit/hyperactivity disorder (ADHD) and oppositional defiance disorder (ODD) or conduct disorder (CD) problems, but treatment effects vary widely. We aimed to identify key moderators of intervention effects on child behaviour difficulties using individual participant data from randomised controlled trials (RCTs). Methods: We pooled individual data from 48 RCTs (5729 families) of behavioural parenting programmes for children with (sub)clinical characteristics of ADHD and/or ODD/CD. Fifteen parent-reported child, parent, and family characteristics at baseline were examined as potential moderators. The primary outcome was parent-reported child behavioural difficulties (e.g., hyperactivity-impulsivity, defiance, irritability, aggression, rule-violation) at the first post-intervention assessment, harmonised across instruments and trials. We used a machine learning method, the Least Absolute Shrinkage and Selection Operator (LASSO) regularization to select robust moderators and then tested selected moderators in a standard regression model. Findings: Two significant moderators were identified: Baseline child internalising problems (e.g., anxiety, depression, withdrawal) and baseline non-constructive parenting (e.g., harsh or inconsistent discipline). Moderating effects were driven by increases in child behavioural difficulties among families in control condition with higher baseline levels of child internalizing problems and non-constructive parenting, rather than larger improvement among intervention recipients. For non-constructive parenting, this increase was specific to waitlist conditions. Interpretation: Our findings suggest that children with higher levels of internalising problems and/or families reporting higher non-constructive parenting are at greatest risk of deterioration in behavioural difficulties without support. Results may inform triage: to identify children with co-occurring internalizing problems who may benefit from timely access to behavioural parenting programs and minimize waiting times for families with higher non-constructive parenting.

Background. Attention-deficit/hyperactivity disorder (ADHD) is a prevalent neurodevelopmental disorder, with increased risk for negative adult long-term outcomes. These outcomes may differ between males and females with ADHD, but whether these differences are more pronounced in ADHD or similar to those in the general population remains unclear. Method. In an 18-year follow-up of 283 children (Mage = 28.48) we assessed if sex moderated 51 long-term outcomes across 7 domains: psychiatric status, behavioral/emotional problems, academic/professional functioning, adaptive functioning, neurocognition, physical health, and healthcare service use in those with ADHD (nfemales = 50, nmales = 104), and without ADHD (nfemales = 77, nmales = 52). Additionally, we aimed to explore the moderating role of gender-expression. Results. Sex interacted with group for social functioning: males with ADHD were more likely to show lower social functioning compared to females with ADHD and controls. There were no sex differences on the other outcomes. As sex and gender-expression were highly similar (r = .94), specific effects of gender-expression could not be studied. Conclusions. Generally, sex did not have a more pronounced effect on long-term outcomes of childhood ADHD in individuals with ADHD compared to the general population. Gender-expression could not be studied in our predominantly cisgendered sample.

Self-help parenting programs that are aimed at reducing child behavioral difficulties may be effective and accessible alternatives to face-to-face parenting programs. Insight into the mechanisms of change for such programs is important to enhance their effectiveness. The goal of this study was to examine whether changes in supportive and unsupportive parenting (measured by the Parenting Practices Interview; PPI) and parenting sense of competence (measured by the Parenting Sense of Competence scale; PSOC) mediated the effects of a self-help parenting program on child behavioral difficulties. A total of 110 families were randomized to a 15-week self-help program or waitlist. Outcomes and mediators were measured at pre-, mid- and post-measurement. Outcomes were parent-rated child behavioral difficulties, assessed by the Eyberg Child Behavior Inventory, Intensity Scale (ECBI-I), and ecologically momentary assessments (EMA) of child behavior problems. Results showed that reductions in unsupportive parenting significantly mediated reductions in child behavioral difficulties assessed with the ECBI-I. We did not find this mediating effect for the EMA. Supportive parenting and parenting sense of competence did not mediate changes in child behavioral difficulties. Although future research into self-help parent training should provide insight into possible reciprocal interactions between changes in parents and children, reducing unsupportive parenting practices may play a key role in reducing child behavioral difficulties.

Background. Adults with attention-deficit/hyperactivity disorder (ADHD) diagnosed in childhood show higher healthcare utilization, but their medication use remains understudied while this can clarify the long-term health burden of childhood ADHD. Methods. In this 18-year prospective cohort study, pharmacy dispensing records of ADHD, other psychotropic, and somatic medication of the last year were compared between adults with childhood ADHD (n=125, Mage=29, 68% males), their siblings (n=109, Mage=28, 35% males) and controls (n=101, Mage=30, 42% males) without childhood ADHD at study entry. Group differences in demographics and psychiatric status were included as covariates (sensitivity analyses). Results. The childhood ADHD group was more likely to be prescribed ADHD medication, more different ADHD medications, non-steroidal anti-inflammatory drugs (NSAIDs), and neurological medications (specifically analgesics) compared to their siblings and controls (OR range=3.25-8.35, β range=0.61-0.65). Compared to controls, the childhood ADHD group was also more likely to be prescribed other psychotropic medication (OR=3.82) and used more different types of psychotropic medication (β=0.38). Siblings and controls did not differ. Findings remained after controlling for demographic differences, except that the childhood ADHD group no longer differed from controls on analgesics. After controlling for psychiatric status differences (ADHD, major depressive disorder), the childhood ADHD group still differed on ADHD medication use, more different ADHD medications, and NSAIDs, and from controls on more different psychotropic medications. Conclusions. Adults with childhood ADHD, but not their siblings, show elevated long-term medication use. The elevated medication use is not merely explained by shared familial factors but is partly attributable to current psychiatric status.