ADHD at Age 7 and Functional Impairments at Age 10

In this study, we report on data from the Children’s Attention Project, a community-based cohort study used to assess children with and without ADHD at 3 time points: baseline, 18 months, and 36 months (ages 7, 8.5, and 10 years, respectively).¹²  Ethical approval was obtained from the human research ethics committees of The Royal Children’s Hospital (31056) and the Victorian Department of Education and Training (2011_001095).

Children were recruited from 43 government schools in Melbourne, Australia, representing diverse socioeconomic communities in 2011 and 2012. A two-stage procedure (screening followed by diagnostic interviewing) was used to ascertain the sample. First, the Conners 3 ADHD Index¹³  was sent to the parents of all children in grade 1 (second year of school). For those children whose parents returned the survey and consented to teacher participation, the child’s teacher was asked to also complete the Conners 3. Children were classified as screening positive for ADHD if both parent and teacher ratings on the ADHD Index were ≥75th percentile for boys and ≥80th percentile for girls.¹¹  A higher cut point was used for girls because our pilot data revealed that this resulted in better correspondence with diagnostic confirmation. Children with a previous ADHD diagnosis were also classified as positive screen results. Children were classified as screening negative if both their parent and teacher ADHD Index scores were <75th percentile for boys or <80th percentile for girls and there was no previous diagnosis of ADHD. Children with discordant parent and teacher ratings were not followed longitudinally in this study.

Next, all children who screened positive for ADHD were matched 1:1 by sex and school to a child who screened negative for ADHD, and all were invited to participate in the longitudinal study. Participation in the longitudinal study involved completion of National Institute of Mental Health Diagnostic Interview Schedule for Children Version IV (DISC-IV)¹⁴  with parents to confirm ADHD status and other mental health conditions, direct child assessments, and parent- and teacher-completed surveys. DISC-IV interviews were conducted by trained research assistants with at least a 4-year degree, blinded to baseline screening status. Children who screened positive and meet diagnostic criteria on the DISC-IV were classified as FS; those who screened positive but did not meet diagnostic criteria on the DISC-IV were classified as ST; and those who screened negative and did not meet diagnostic criteria were classified as controls. The flow of participants into groups is shown in Fig 1.

Children were excluded if they had an intellectual disability, severe medical condition, genetic disorder, moderate-severe sensory impairment, or neurologic disorder. Families who were unable to complete the surveys and interviews in English were also excluded.

Parent and teacher surveys, diagnostic interviews, and direct child assessments were repeated 36 months postrecruitment (2014 and 2015) when the children were 10 years old.

Academic achievement was assessed directly and by teacher ratings. The Wide Range Achievement Test 4 (WRAT 4)¹⁵  word reading (word decoding and recognition) and math computation (counting, number identification, oral problem-solving, and written problem calculation) subtests were used for direct assessment. Standard scores based on age are reported (normative mean = 100; SD = 15) and a composite score was derived (average of the 2 subscales) for use in the predictive analyses. We also collected teacher-reported academic competence (7 items) from the Social Skills Improvement System (SSIS)¹⁶  (age-based standard scores, mean = 100; SD = 15). Emotional-behavioral and social outcomes were measured by using parent- and teacher-completed Strengths and Difficulties Questionnaire (SDQ).¹⁷  Emotional-behavioral problems were measured by using the 20-item SDQ total problems score (range 0–40) summed from the emotional problems, conduct problems, inattention-hyperactivity and peer problems subscales. The 5-item peer problems subscale was also examined separately as our measure of social outcomes (range 0–10).

ADHD symptom severity was measured by using the parent-reported Conners 3 ADHD Index.¹³  Working memory was measured by using the Digits Backward from the Wechsler Intelligence Scale for Children, Fourth Edition (scaled score).¹⁸  Emotional-behavioral and peer problems were measured by using the parent-reported SDQ. ASD symptoms were measured by using the 40-item, parent-report Social Communication Questionnaire Lifetime Form.¹⁹  Child receipt of additional assistance at school (receives specialized services or has an individual learning plan) was reported by teachers at age 7 years.

Primary caregiver mental health difficulties were measured by using the 6-item Kessler Psychological Distress Scale.²⁰  Parenting hostility was measured by using the 5-item parenting anger scale from the Longitudinal Study of Australian Children.²¹ 

Child characteristics reported at both age 7 and age 10 include age, sex, cognitive function matrix reasoning subtest from the Wechsler Abbreviated Scale of Intelligence²² ), parent and teacher-reported ADHD symptom severity (Conners 3 ADHD Index), parent-reported ASD symptoms (Social Communication Questionnaire), and parent-reported ADHD medication use. Mental health disorders including ADHD were assessed by using the DISC-IV.¹⁴  Children were classified as having an internalizing disorder if they met criteria for separation anxiety disorder, social phobia, generalized anxiety disorder, posttraumatic stress disorder, obsessive-compulsive disorder, major depression, hypomania, or manic episode and an externalizing disorder if they met criteria for oppositional defiant disorder or conduct disorder. The DISC-IV is used to report ADHD subtype and persistence at age 10. Primary caregiver characteristics reported at both age 7 and age 10 included single-parent status and education level, neighborhood socioeconomic disadvantage (residential postcode classification),²³  and parent mental health (6-item Kessler Psychological Distress Scale).

Summary statistics were used to describe the continuous and categorical sample characteristics. Random-effects linear regression models were fitted to compare academic, behavioral-emotional, and social outcomes at 10 years between children with FS ADHD and controls without ADHD, between children with ST ADHD and controls without ADHD, and between children with FS ADHD and children with ST ADHD (aim 1). The random-effects regression analyses allowed for clustering at the school level. Both unadjusted analyses and analyses that adjusted for potential confounders identified a priori (child age, child sex, single-parent status, parent education level, and socioeconomic disadvantage) are reported. All continuous predictors and outcomes were standardized (mean = 0; SD = 1).

For aim 2, the FS and ST ADHD groups were combined to form a single ADHD group. Random-effects linear regression models were fitted to identify the age-7 predictors of academic, emotional-behavioral, and social outcomes at age 10 for each of the ADHD and control groups (aim 2). We examined predictors of outcome in the control group to contextualize any relationships found for the ADHD group. For each outcome, simple (crude) models in which 1 predictor was used at a time and a single multivariable model including all potential predictors were fitted. The results from the multivariable analyses are considered primary.

All analyses were conducted for families with complete data (nonimputed) and then repeated by using imputation to account for missing data. Data were imputed by using the chained equations method. Forty complete data sets were imputed, which included all 477 children enrolled in the cohort study. Given the similarity in our findings in our complete case and imputed analyses, we focus on the imputed analyses as our primary analyses. All age-7 and age-10 variables included in our analyses were incorporated into our multiple imputation model.

All statistical analyses were conducted in Stata 15.0 (Stata Corp, College Station, TX).

Sample retention by outcome source is illustrated in the participant flowchart (Fig 1), with 78.2% (n = 373) of the baseline sample (N = 477) retained at age 10 years. Retention was defined as having data available on at least 1 key outcome variable. There were no marked differences between responders and nonresponders at age 10 by sex, ADHD symptom severity, ADHD subtype, presence of comorbidities, or primary caregiver high school completion rates (all measured at age 7).

Participant characteristics are described in Table 1. Two-thirds (66.4%) of children in the FS group, 25.8% in the ST group, and 2.8% of controls met full ADHD criteria at age 10. The percentage taking ADHD medication increased from 12.6% at baseline to 21.4% at follow-up in the FS group and from 0% to 6.5% in the ST group. At follow-up, comorbid internalizing disorders were present in 26.4% and 16.7% of the FS and ST groups, respectively, and externalizing disorders were present in 51.2% and 24.3% of the FS and ST groups. The FS and ST groups were socially disadvantaged relative to the controls at baseline (primary caregiver education level and single-parent household status).

Academic, emotional-behavioral, and social outcomes at age 10 were compared among the 3 groups (in pairs; Fig 2, Supplemental Tables 5 and 6). At age 10, with 1 exception (teacher-rated social problems, ST group), children in both the FS and ST groups had more difficulties than those in the control group across all outcome variables after adjustments. The FS group had higher parent-reported emotional-behavioral difficulties compared with the ST group but were similar on all other measures. This pattern was similar to that seen at age 7 (Supplemental Table 5).

The best predictor of academic achievement (reading and math composite) in both the combined (FS and ST) ADHD group and the control group was working memory (Table 2). Emotional problems, conduct problems, and ASD symptoms predicted academic function in the ADHD group in the unadjusted analysis but not in the adjusted analysis.

In the ADHD group, ADHD symptom severity predicted both parent- and teacher-reported emotional-behavioral problems. Emotional problems, conduct problems, ASD symptoms, and parent mental health problems also predicted parent-reported emotional-behavioral problems in the ADHD group. In the control group, the only predictor of parent-reported emotional-behavioral problems was conduct problems; none of the variables examined were predictors of teacher-reported emotional-behavioral problems in the control group (Table 3).

In the ADHD group, although several variables predicted parent-reported social problems in the unadjusted analysis, ASD symptoms were the only variable for which there was evidence of a relationship in the adjusted model. No variables were associated with teacher-reported social problems in the ADHD group in the adjusted model. In the control group, no variables in the model predicted parent- or teacher-reported social problems (Table 4).

In this community-based longitudinal study of children recruited at age 7, children with both ST ADHD and FS ADHD were functioning worse on all outcomes (academic, emotional-behavioral, and social) at age 10 compared with controls, after controlling for demographic variables. The ST group had lower parent ratings on social and emotional problems than the FS group but was similarly impaired to the FS group across most outcome domains by teacher report. For the ADHD group (FS and ST combined) the strongest baseline predictors of outcome at follow-up were working memory for academic functioning, ADHD symptom severity for emotional-behavioral problems, and ASD symptoms for social functioning. Emotional problems, conduct problems, ASD symptoms, and parent mental health symptoms at age 7 also predicted parent-reported emotional-behavioral problems at age 10 in children with ADHD.

The finding of persistently poorer academic performance highlights the importance of identifying academic difficulties in the early school years in children with ADHD and providing appropriate remedial interventions. Furthermore, our findings reveal the reliability of teacher ratings of academic competence in children with ADHD. Using this simple scale to classify children resulted in similar findings to direct academic achievement testing. The practical implication is that in children with ADHD, academic delays can be identified by teachers and remedial supports provided without the need for formal assessment.

The ST group demonstrated functional status between the ADHD group and the controls at follow-up on all outcome measures. This is consistent with the findings from a Swedish longitudinal cohort twin study in which authors found a relationship between ADHD symptoms at age 9 to 12 and psychosocial problems at age 15 across ADHD, ST ADHD, and control groups.²⁴  Our findings for the ST group were also consistent with a cross-sectional Korean study of children aged 8 to 11 years, in which the ST group had parent ratings of academic function and emotional and behavioral symptoms in between those of FS ADHD and controls.²⁵  Consistent with our data, they found that ST ADHD had a more even sex ratio (56% boys) compared with cases of FS ADHD (83% boys). In neither of these studies did authors use direct child academic assessment or teacher-reported outcomes. Our findings extend this work by demonstrating differences in outcomes across settings for the ST ADHD. We found that the ST group’s teacher-reported social and emotional symptom profile and the teacher-reported academic competence were similar to the FS ADHD group. This suggests that children with ST ADHD need similar levels of classroom support as children with FS ADHD. Although lower than the FS group, the ST group had substantial rates of both internalizing and externalizing comorbid disorders, further highlighting the clinically important problems faced by these children. Furthermore, it was notable that although the rate of externalizing disorders was relatively constant between the two time points in both groups and the rate of internalizing disorders was relatively constant in the FS group, the rate of internalizing disorders in the ST group increased substantially. This suggests that in the ST group, secondary emotional effects may accrue over time.

These findings extend our understanding of the critical relationship between working memory and academic functioning. Visuospatial working memory has been found to mediate the relationship between inattention and poorer math achievement one year later in early elementary schoolchildren generally.²⁶  Consistent with Reenie et al,²⁷  we found an association between working memory deficits and math achievement function longitudinally in children with ADHD. Authors of several intervention studies have investigated the effects of cognitive training primarily targeting working memory in children with ADHD. Unfortunately, although performance on laboratory tests of working memory can be improved, there is little evidence of changes in ADHD symptoms or academic performance.²⁸  Training children to improve the neuropsychological deficits underpinning their functional difficulties²⁹  is conceptually attractive, but the clinical utility of this approach remains to be demonstrated. Perhaps combined approaches whereby working memory training occurs alongside academic remediation could be a fruitful approach in this population.

Our study demonstrates the critical influence of ASD symptoms on social and emotional functioning in children with ADHD. ASD symptoms have been found cross-sectionally to be associated with higher social and emotional impairments in children with ADHD.³⁰  In the current study, we extend this by demonstrating that comorbid ASD impacts function over time in children with ADHD. A potential implication of this result is that early identification and treatment of elevated ASD symptoms may help reduce negative outcomes in children with ADHD. Our results suggest that a broad clinical approach is needed to manage ADHD, which includes not only ADHD symptom management but also identification and management of comorbid conditions such as ASD³¹  and internalizing and externalizing disorders.

This study had a number of design strengths. We recruited and carefully phenotyped children within a narrow age band and retained three-quarters of the sample to follow-up. We sampled boys and girls across the sociodemographic spectrum and included children with all comorbidities, resulting in a real-life sample with mixed developmental vulnerabilities. We therefore believe our findings are generalizable to the population of children with ADHD in the community. Finally, we examined a broad range of functional outcomes, generating a rich understanding of the difficulties faced by these children and their families.

Our study also had some limitations. First, there were some potential sample biases on recruitment. Families excluded because of incomplete screening data were relatively socially disadvantaged compared with participating families, and the rate of consent in our negative screening control group was lower than in our cases. Second, our design did not enable the examination of the influence of internalizing and externalizing disorders on outcomes at 10 years because we treated comorbidities as outcomes. This design was chosen to align with developmental pathways research regarding the sequence of emergence of mental health comorbidities.³²  Third, the definition of ST differs between cohorts.³³  We believe ours is robust and clinically meaningful, but the definition needs to be considered when comparing results with other studies. Finally, although we have reported previously on the prevalence and predictors of medication use in this cohort,³⁴  only a minority of participants had been treated with medications, so we were unable to comment on treatment effects.

ADHD symptoms in early elementary school are robust markers of developmental and mental health vulnerability. This is true irrespective of whether children meet diagnostic criteria. Therefore, clinicians should monitor children with ADHD symptoms even when they fall below the diagnostic threshold. The strongest modifiable risk factors were poor working memory (academic outcome), ADHD symptom severity (emotional-behavioral), and ASD symptoms (emotional-behavioral and social). This information should help to inform the development of intervention models used to improve outcomes in ADHD.

ADHD

attention-deficit/hyperactivity disorder

ASD

autism spectrum disorder

DISC-IV

Diagnostic Interview Schedule for Children Version IV

FS

full-syndrome

SDQ

Strengths and Difficulties Questionnaire

SSIS

Social Skills Improvement System

ST

subthreshold

WRAT 4

Wide Range Achievement Test 4