Abstract / Summary
Abstract Background. ADHD is conceptualised as a neurodevelopmental disorder with childhood onset, but prospective studies have documented adolescents and adults who meet criteria for ADHD yet had no childhood diagnosis, or only non-clinical levels of symptoms in childhood. What these cases represent (residual childhood ADHD, delayed recognition, or something else) is an important diagnostic question. Objective. To compare five prospective longitudinal cohorts and examine how ascertainment of childhood symptoms, subthreshold symptoms, impairment, comorbidity, substance use and repeated assessment affect the detection of apparent late-onset ADHD. Unlike earlier reviews, we focus on how ascertainment procedures shape apparent late-onset estimates. Methods. A targeted, purposive comparative narrative synthesis of published reports from the Dunedin Multidisciplinary Health and Development Study, the E-Risk Longitudinal Twin Study, the 1993 Pelotas Birth Cohort, the Avon Longitudinal Study of Parents and Children (ALSPAC) and the Multimodal Treatment Study of ADHD Local Normative Comparison Group (MTA LNCG). Heterogeneity in populations, instruments, ages and definitions precluded meta-analysis. Numerators and denominators were extracted from each original report. Results. Estimates differed markedly between cohorts: Dunedin: about 3% (31 cases) met adult ADHD criteria at age 38; about 90% lacked a childhood ADHD diagnosis on prospective assessments at ages 11, 13 and 15. E-Risk: 8.1% (166/2,040) met criteria at age 18; about two-thirds (about 5.5% of the cohort) did not meet criteria at any childhood assessment. Pelotas: 12.2% (492) met symptom criteria at 18–19 years; 84.6% of them (10.3% of the cohort) had no childhood ADHD at age 11. The estimate fell to 6.3% after comorbidity exclusions. ALSPAC: of 75 apparent late-onset cases, 74.7% (56) had elevated hyperactivity scores earlier in childhood; 19 (about 0.4% of the sample) formed a stringently defined late-onset group. MTA LNCG: about 95% of screen-positive participants were excluded after comprehensive assessment; 2 of 239 (0.8%) were judged adult-onset, both with complex psychiatric histories. Conclusions. Estimates of apparent late-onset ADHD varied substantially with how childhood symptoms, impairment, comorbidity and follow-up were defined. Absence of a documented childhood diagnosis is not definitive evidence of adult onset, and the evidence does not establish that all such cases are unrecognised childhood ADHD. Developmental history and consideration of alternative explanations remain essential. Keywords: ADHD; adult ADHD; late-onset ADHD; longitudinal cohort; developmental psychopathology; diagnostic validity; differential diagnosis