The Children’s Attention Project: a community-based longitudinal study of children with ADHD and non-ADHD controls

There is a substantial literature addressing the aetiology and treatment of ADHD, with studies examining the course of ADHD over time mostly focussed on children selected through clinical settings [9‐13]. These studies of clinical samples have demonstrated that both boys and girls with ADHD are at risk for a range of poorer outcomes in adolescence and adulthood. For example, in terms of mental health, children with ADHD have higher rates of externalising behaviour disorders (e.g., oppositional defiant disorder, conduct disorder) than non-ADHD controls [14, 15], with mixed evidence regarding risks for substance use and mood and anxiety disorders [13, 16‐18]. Children with ADHD also have increased risk for poor educational (e.g., grade repetition, suspensions, lower attainment and achievement), social (e.g., fewer friends, trouble maintaining friendships) [13, 14, 19], and occupational outcomes (e.g., poorer job performance, higher employer-rated ADHD symptoms) [14]. The parents of children with ADHD also have poorer outcomes over time, including increased psychological distress and poorer family functioning than the parents of non-ADHD controls [9]. ADHD symptoms persist into adolescence and adulthood for approximately 50% of children diagnosed with ADHD in childhood [20]. However, apart from evaluations of specific medication, behavioural or educational interventions, little is known about the factors associated with symptom remission. Although there may be a decline in ADHD symptoms as children progress into adolescence and adulthood, the impairment associated with the disorder tend to persist [20].

While previous research provides valuable information about the outcomes for clinic-referred children with ADHD, it tells us little about children not accessing clinical services [21, 22]. Australian population studies show cross-sectionally that as few as 1 in 5 children and adolescents with mental health difficulties had accessed professional services within the six months prior to being surveyed [23]. Compared with children with ADHD in the general population, clinical samples over-represent children with more severe ADHD, males, and those with comorbid mental health conditions [21, 22]. In turn, clinical samples under-represent those with ADHD-I and females with ADHD [21, 22].

Community-based longitudinal designs have the potential to address knowledge gaps regarding the course of ADHD and provide findings that are more applicable to the general population of children with ADHD. However, most existing community-based studies are limited in their assessment of ADHD status. For example, one common approach is to ask parents whether their child has ever received a diagnosis of ADHD. This approach is likely to contain measurement error arising from parents who indicate ‘yes’ based, for example, on professional opinion rather than formal diagnostic assessment. Furthermore, this approach does not assess whether symptoms are current, and as with the studies of clinical samples of children with ADHD, this approach will only include children who do get referred and assessed for ADHD. The second common approach is to define ADHD status using parent and/or teacher-reported ADHD symptom checklists [24, 25]. The limitation of this approach is that it does not assess two key aspects required for ADHD diagnosis – namely, the persistence of symptoms for a period of at least six months, and significant functional impairment.

To date, there have been few community-based longitudinal studies of children with diagnostically confirmed ADHD and controls [26‐29]. Barberesi and colleagues report on the outcomes of a population-based birth cohort of ADHD incident cases (n = 379) and non-ADHD controls (n = 758) in Rochester, Minnesota [26, 29]. Children with ADHD were found to have poorer educational and psychiatric outcomes than non-ADHD controls [26, 29]. However, the study employed a retrospective design and direct assessment of ADHD, mental health disorders and educational outcomes was not possible. Rasmussen and Gillberg [28] followed a Sweedish community sample of children with ADHD (combined with Developmental Coordination Disorder) into adulthood. Their sample of children with ADHD-only was too small at follow-up for meaningful analyses (n = 11). Bussing and colleagues report 8 year outcomes for 5 to 11 year old North American school-children with ADHD (n = 95), subthreshold ADHD (n = 75), and a comparison group (n = 163) [27]. At follow-up, children with ADHD had poorer mental health, quality of life, and increased contact with the juvenile justice system compared with non-ADHD controls. The wide age range at baseline and the large gap between baseline and outcome assessment precluded detailed examination of how children’s symptom trajectories changed over time.

Further research in this area would benefit from prospective community-based designs that include both direct assessment of children’s functioning and multi-source (parent, teacher, child) data on potential risk and protective factors for poor long-term outcomes. Regular follow-up assessments are required to enable examination of subtle changes in children’s symptoms. This is important in understanding the developmental trajectories for children with ADHD and the predictors of poor versus better outcomes in this population.

Socio-ecological models that highlight the multitude of individual and contextual factors shaping children’s health and development provide a useful framework for conceptualising the factors likely to influence the persistence of ADHD and associated outcomes. The socio-ecological model of children’s health that underpins the selection of constructs to be measured in our study is shown in Figure 1.

Macro-environmental (e.g., social, economic, policy factors) and distal social environments (e.g., neighbourhood, school, community factors) are important factors influencing children’s developmental trajectories, but little attention has been given to these factors within existing longitudinal studies examining outcomes for children with diagnostically-confirmed ADHD. Key proximal influences are also critical and may include parent–child interactions, parental mental health, and children’s relationships with their teachers and peers. Longitudinal studies of children with ADHD have identified that family context [30‐32], child factors (e.g., IQ, comorbid mental health problems; ADHD severity/subtype) [15, 31, 33, 34], homework management [35] and interventions (education programs; stimulant medication) [36] may influence outcomes. However, findings have been mixed and studies have yet to examine a comprehensive range of predictors within the one study. Furthermore, the existing research examining predictors of outcomes is largely characterised by the sampling limitations noted for previous cohort studies, and to date there has been little attention to the potential risk or protective roles of a broad range of early school factors such as child-teacher relationships.

The present study will establish a community-based cohort of 200 6–8 year old children with diagnostically-confirmed ADHD and matched non-ADHD controls. The cohort will be tracked across the early years of primary (elementary) school to determine the mental health, academic and social outcomes for children with and without ADHD and to identify the factors associated with differential outcomes within the ADHD group. Within a community-based sample, the primary aims of this study are to:

The Children’s Attention Project is a longitudinal cohort study conducted in accordance with STROBE guidelines [37]. The project commenced in early 2011, with the cohort to be screened and recruited from two consecutive years of Grade 1 enrolments (cohort 1 in 2011, cohort 2 in 2012). Baseline, 12, 24 and 36 month follow-ups will be conducted across 2011 to 2015. The anticipated flow of participants for the two cohorts combined is shown in Figure 2, based on data from a pilot study (n = 345). The study has been funded by the National Health and Medical Research Council of Australia (project grant number: 1008522) and internal funding from the Murdoch Childrens Research Institute. It has been granted approval by the Human Research Ethics Committees of the Royal Children’s Hospital (#31056) and the Victorian Department of Education and Early Childhood Development (#2011_001095).

Participants are Grade 1 (which is the second year of formal schooling in Victoria) children and their parents and teachers from 43 government primary schools in Melbourne, Victoria. Grade 1 was chosen as the recruitment point, as this is a developmental stage at which children with ADHD often begin to manifest impairments related to the functional impact of their symptoms. Earlier sampling would result in missed cases. Although later sampling would increase the yield of cases, opportunities to reliably record and then track key early influences on developmental trajectories would be lost.

Exclusion criteria include: intellectual disability; serious medical conditions; genetic disorders; moderate-severe sensory impairment; neurological problems; and parents with insufficient English to complete interviews/questionnaires.

Measures are summarised in Table 1. Parent and teacher reported measures will be collected from baseline (T1) to 36 month follow-up (T4). Parent interviews and direct child assessments collected at T1 will be repeated at T4. Measures of child outcomes are all reliable, valid scales, with excellent psychometric properties, sensitivity to intervention and widely used in Australia and internationally. Measures of possible determinants also have robust psychometric properties, are brief, and are mostly drawn from the Longitudinal Study of Australian Children (LSAC) [41, 42]. Family demographic information collected at baseline (and repeated for those variables that can change over time) include parents’ age, employment, education, mental health diagnoses (including family history of ADHD), income, family composition and whether the child has ever lived for an extended period of time out of parental care. Items (mostly from LSAC) are also collected on school absences, and child and parent sleep quality. School demographics are sourced from administrative data.

In order to detect a 0.3 SD clinically meaningful difference between the ADHD group and controls on each of our continuous primary outcomes, with 80% power and at a significance level of 0.05, we need to recruit 200 children in the ADHD group and at least 200 children in the control group. Based on data from the pilot study, we need to distribute ADHD screening questionnaires to approximately 5,300 parents in order to have data on 175 children with ADHD at follow-up. This assumes an initial 70% response rate to the combined parent/teacher screening questionnaires of whom 10% screen positive for ADHD, and an 80% consent rate to the diagnostic confirmation and baseline assessment, of whom 70% of children screening positive then meet criteria for ADHD on the DISC-IV. We have allowed for 15% attrition over the three subsequent follow-ups.

Children whose screening status differs from their subsequent diagnostic status will be followed as an ‘at risk’ sample. Based on data from our pilot study, we anticipate that approximately 30% of the sample who screen positive for ADHD and 3% of those who screen negative, will fall into the at risk group. Based on the sample size in the present study this will equate to approximately 98 children at baseline.

We will compare the proportions of children meeting diagnostic criteria for other DSM-IV mental health diagnoses (DISC-IV) between the ADHD and control groups using the Chi-squared test for each of the diagnostic categories and the Mann Whitney test to compare symptom counts (oppositional defiant disorder/conduct disorder, mood and anxiety disorders, autism spectrum disorder). We will compare mean differences in parent and teacher reported total behavioural and emotional problems, social functioning, parent-reported child quality of life, measured academic achievement, and teacher-reported classroom performance between the two groups using t tests (or Mann–Whitney tests, depending on the data distribution). Furthermore, we will compare mean differences in parents’ levels of depression and family quality of life between the two groups using t tests.

We will then conduct regression analyses (linear regression for continuous data and logistic regression for binary data) for each outcome, adjusted for potential confounders identified a priori (e.g.,. child gender, family income, family structure, mother’s age, comorbidity). We will also repeat all of the above to explore differences between ADHD, at risk, and control groups.

We will conduct multivariable regression analyses (linear or logistic as appropriate) within the ADHD, at risk and control groups separately to determine the predictors of poorer (or better) functioning in each of our primary outcome domains (academic achievement, mental health problems, quality of life and social functioning). Predictors will include (depending on the outcome domain being examined): ADHD subtype, ADHD symptom severity, comorbidities, demographics, cognitive functioning, executive functioning, communication skills, parenting and couple factors, parental mental health, family QoL, teacher-child relationships, teacher/school characteristics and treatment factors. Tests of interaction between ADHD status and these predictors will be used to identify those that uniquely predict poorer or better outcomes within the ADHD group.