The perception of ADHD as a strictly childhood-onset disorder has been the subject of discussion in recent years. Psychiatric data collected over a period of 15 years in the Multimodal Treatment of ADHD (MTA) study demonstrated that 2.1% of individuals without ADHD in childhood presented with ADHD symptoms after the age of 12 years (Sibley et al. 2018), suggesting that late-onset ADHD is a valid phenomenon. However, late-onset ADHD remains poorly understood, and it is unclear whether the late-onset phenotype shares its aetiology with childhood-onset ADHD. Birth cohort studies have suggested that late-onset ADHD cases appear to show typical development during childhood, but are then phenotypically indistinguishable from childhood-onset ADHD cases by early adulthood, suggesting that developmental and environmental factors specific to adolescence may be critical to understanding the aetiology of late-onset ADHD.
Neurocognitive models of ADHD outline two pathways that contribute to cognitive and behavioural ADHD symptoms:
- A “cool” executive functioning (EF) deficit associated with mesocortical dopamine circuits, resulting in impairment in cognitive control, including working memory, inhibitory control and planning capabilities (i.e. ‘inattentive’ symptoms).
- A “hot” rewards processing deficit associated with cortical–striatal dopamine loops, resulting in impairment in delay discounting, risky decision-making and motivation (i.e. ‘hyperactive/impulsive’ symptoms).
This study evaluated the impact of adolescence-specific factors on EF and impairment in the context of these models, with the aim of proposing three potential aetiological hypotheses for late-onset ADHD:
- Increased cognitive load exacerbates “cool” EF vulnerabilities – academic demands, workload and requirements for independent learning typically increase in adolescence, and this increased cognitive load may result in impairments in concentration, memory and self-regulation, particularly in individuals with cognitive vulnerabilities such as below-average IQ.
- Adolescent rewards response mimics “hot” rewards-processing deficits – adolescent rewards processing is reported to differ from that of children and adults, with inability to delay gratification associated with impulsive and risk-taking behaviours in adolescence. Adolescents with particularly heightened rewards responses may present with motivational and behavioural impairments that mimic ADHD symptoms or worsen existing cognitive vulnerabilities.
- Exposure to traumatic stress produces ADHD-like deficits in self-regulation – traumatic stress causes disruption of self-regulatory processes related to cognition, behaviour and emotion, and this may be amplified in the developing adolescent brain, resulting in elevated sensitivity and vulnerability to psychopathology. Trauma-related disruptions to the “cool” EF or “hot” rewards processing neural pathways may result in presentation of ADHD-like symptoms in adolescents.
To investigate these hypotheses, 50 participants were recruited from the 9th grade of two public high schools in a culturally diverse region in the eastern United States (ADHD group*: n = 32 [childhood-onset ADHD, n = 17; late-onset ADHD, n = 15]; non-ADHD group†: n = 18. Mean [standard deviation (SD)] age: 14.59 [0.79] years, 14.87 [0.74] years and 14.50 [0.79] years, respectively). ADHD diagnosis was based on the Diagnostic and Statistical Manual for Mental Disorders – 5th Edition (DSM-5TM), according to combined parent and adolescent reports on the Schedule for Affective Disorders and Schizophrenia for School-Age Children-Present and Lifetime Version (K-SADS-PL). Age at onset was determined from independent parental and adolescent reports of the age at which each symptom on the K-SADS-PL presented, enabling construction of a comprehensive timeline of symptom onset and progression. Late-onset ADHD was defined as a diagnosis of ADHD with ≤2 symptoms of both inattention and hyperactivity/impulsivity before age 12, according to both parental and adolescent report.
Cognitive vulnerability was assessed using a variety of validated methods to measure six aspects of EF: working memory; cognitive flexibility; response inhibition; processing speed; metacognition (ability to initiate, plan, organise, self-monitor and sustain working memory); and IQ. Rewards processing was assessed using validated methods to measure the following: risky decision-making; delay discounting; and academic motivation. Trauma exposure was assessed using the K-SADS-PL Post-Traumatic Stress Disorder Screener. Environmental demands were measured from parent reports of parental academic demands and extracurricular demands. At 1-year follow-up, adolescents, parents and teachers completed the DSM-5TM ADHD checklist and the Impairment Rating Scale, allowing assessment of ADHD symptom persistence/desistence, and rates of progression into the school’s 10th grade were obtained from official school records. The results of the study were as follows:
- There were no differences in the clinical profiles of the childhood- and late-onset ADHD groups, including symptom scores, medication use and previous diagnoses.
- Hypothesis 1: there were no significant differences between adolescents with late-onset vs childhood-onset ADHD in any of the cognitive vulnerability measures; whereas, compared with the non-ADHD group, the late-onset ADHD group showed significantly more metacognition problems (mean [SD] score 42.06 [7.11] vs 56.32 [14.46]; p = 0.001) and a lower full-scale IQ (101.07 [15.52] vs 91.02 [11.94]; p = 0.049). Additionally, the late-onset group faced significantly higher parental academic demands compared with the childhood-onset ADHD group (mean [SD]: 3.79 [0.41] vs 3.11 [0.78]; p = 0.004), whereas there was no significant difference in environmental demands between the late-onset and non-ADHD groups. This suggests that, in line with hypothesis 1, increased cognitive load in adolescence may result in ADHD-like symptoms in students with cognitive vulnerabilities.
- Hypothesis 2: the late-onset ADHD group showed poorer performance than the childhood-onset group on the delay discounting task (mean [SD] 278.07 [51.01] vs 322.76 [64.45]; p = 0.047), and also demonstrated lower levels of academic motivation than the non-ADHD group (3.94 [1.07] vs 4.61 [0.54]; p = 0.016).
- Hypothesis 3: There was no significant difference in trauma exposure between the late- and childhood-onset ADHD groups; however the late-onset group experienced a significantly higher rate of multiple trauma exposure than the non-ADHD group (86.7% vs 33.3%; p = 0.010; odds ratio 11.82). Traumas reported by the late-onset group were experienced before ADHD symptom onset in all cases except one, whereas in the childhood-onset group, the reported traumas predominantly occurred after ADHD symptom onset. This may indicate a complex, bidirectional relationship between ADHD and trauma, which may be regulated by distinct pathways in childhood- and late-onset cases.
- 1-year follow-up: DSM-5TM ADHD symptoms persisted after 1 year in 50.0% of late-onset ADHD cases, and 67.7% of cases still experienced clinically significant impairment with elevated ADHD symptoms at follow-up. The late-onset ADHD group were also more likely than the childhood-onset group to be reassigned to an alternate high school for 10th grade due to persistent course failure (26.7% vs 0.0%; p = 0.023), indicating increased likelihood of academic disengagement in late-onset ADHD cases.
This study had a number of limitations. Firstly, it relied on retrospective reports of age at ADHD symptom onset, and only psychiatric history was used to rule out other disorders as the cause of late-onset symptoms. Additionally, the researchers could not be certain that rewards processing deficits in the late-onset group were related to adolescent development, as they may have been present but undetected in childhood due to a reduced requirement for self-regulation and self-motivation in lower school settings. Furthermore, the sample size was low (N = 50) and the study was only powered to detect large effects; participants were also primarily of ethnic minority groups and from low-income homes, meaning that these results may not be generalisable to adolescents in the wider US population.
The authors concluded that the late-onset ADHD phenotype may have heterogeneous aetiologies, with multiple factors potentially contributing to late-onset symptoms. They suggested that future work should further investigate how environmental and cognitive risk factors may be exacerbated in the context of adolescence, leading to an adolescence-specific form of ADHD. Longitudinal studies are also required to gain further insight into trajectories of symptom persistence/desistence among cases of late-onset ADHD. It is possible that clarifying these characteristics of late-onset ADHD may enable refinement of the DSM-5TM nosology to aid the clinical diagnosis of individuals presenting with ADHD symptoms for the first time in adolescence.
*Participants in the ADHD group were recruited from a larger trial assessing high school interventions for students with ADHD. For that trial, 9th-grade teachers were asked to nominate students who displayed ADHD symptoms; following parental consent, teachers completed a DSM-5TM ADHD checklist and provided measures of academic impairment for each nominated student. Adolescents were eligible for inclusion if they: displayed ≥4 DSM-5TM symptoms of inattention or hyperactivity/impulsivity; scored at least one D or F grade in a core academic class; failed to return ≥20% of academic assignments; scored ≥3 on the academic impairment item of the 0–6 teacher Impairment Rating Scale; were rated as having elevated academic problems on the teacher Adolescent Problems Checklist; and had an IQ of ≥70 on the Wechsler Abbreviated Scale of Intelligence – 2nd Edition (WASI-II). Parents of the 48 participants enrolled in the larger trial were contacted regarding the option of recruitment into the current study; of these, 36 consented to participate, of whom 4 were excluded due to not meeting diagnostic criteria for ADHD (this was not an eligibility criterion for the larger trial)
†Students from all classrooms with at least one participant in the ADHD group were eligible for inclusion in the non-ADHD group study if they: had an IQ ≥70 on the WASI-II; had ≤3 current symptoms of both inattention and hyperactivity/impulsivity according to combined parent and adolescent K-SADS-PL reports and teacher ratings; and had no evidence of ADHD in childhood. Participants were permitted to have academic impairment and non-ADHD mental disorders. Recruitment of the non-ADHD group was deliberately staggered relative to the ADHD group, such that an additional criterion for inclusion in the non-ADHD group was that the participant must increase the group’s demographic similarity to the ADHD group based on school, gender and ethnicity
Sibley MH, Rohde LA, Swanson JM, et al. Late-onset ADHD reconsidered with comprehensive repeated assessments between ages 10 and 25. Am J Psychiatry 2018; 175: 140-149.
Sibley MH, Ortiz M, Graziano P, et al. Metacognitive and motivation deficits, exposure to trauma, and high parental demands characterize adolescents with late-onset ADHD. Eur Child Adolesc Psychiatry 2020; 29: 537-548.