Is there evidence of a causal link between childhood maltreatment and attention deficit/hyperactivity disorder? A systematic review of prospective longitudinal studies using the Bradford‐Hill criteria

Paraskevi Bali,Edmund Sonuga‐Barke,Christina Mohr‐Jensen,Ditte Demontis,Helen Minnis
DOI: https://doi.org/10.1002/jcv2.12169
2023-05-28
JCPP Advances
Abstract:Background Studies report an elevated risk of maltreatment in children with attention deficit/hyperactivity disorder (ADHD), and elevated levels of ADHD in people who suffered childhood maltreatment (CM). However, the direction(s) of causality between CM and ADHD remain unclear—does ADHD create a context for CM, does CM cause ADHD, or both? Objective This study systematically reviews and qualitatively synthesizes the research evidence relating to this question using Bradford‐Hill criteria for establishing causality—strength, temporality, dose‐response and plausibility. Methods We conducted a systematic review, following PRISMA guidelines, of prospective longitudinal studies examining both CM and ADHD. We then used Bradford‐Hill criteria to assess the quality of evidence for a causal link between CM and ADHD. Results All 11 included studies demonstrated an association between CM and ADHD. Seven included evidence for temporality: five suggesting that CM precedes ADHD in the lifespan; two suggesting ADHD precedes CM. Four studies demonstrated a dose response relationship in which greater CM exposure was associated with elevated risk of ADHD. Studies presented a range of plausible mechanisms, including CM causing ADHD through biological programming, versus ADHD causing CM through parental stress. Conclusions The high quality prospective longitudinal studies reviewed confirm the association between ADHD and CM, but present conflicting evidence about the direction of causality and mechanisms underpinning this association. To better understand the complex interplay between CM and ADHD, more studies using new research designs will be required that can partition effects by type of CM and account for bidirectional effects and other complexities.
What problem does this paper attempt to address?