ADHD: Population-based Estimates of Diagnosis, Treatment, and School Outcomes

Researchers: George Farkas (University of California, Irvine), Paul Morgan and Marianne Hillemeier (Penn State)

Funder: Institute of Education Sciences, U.S. Department of Education (Goal 1, Social and Behavioral Outcomes to Support Learning, National Center for Special Education Research)

Duration: Two Years

Purpose: To use nationally representative data to answer the following key questions: 

  1. What are the age- and grade-specific patterns of ADHD diagnosis among US students in grades 1-8?
  2. Which population subgroups of students are more and less likely to ever receive a diagnosis, and to experience different over-time patterns of ADHD diagnosis?
  3. Among students diagnosed with ADHD, which are more and less likely to receive treatment for this condition?
  4. What is the relative effectiveness of medication, special education and related services, therapy, grade retention or combinations of these treatments in improving the behavior, socio-emotional adjustment, and academic achievement of students diagnosed with ADHD by 8th grade?

Setting: We will analyze the Early Childhood Longitudinal Study – Kindergarten Cohort (ECLS-K) data. These data represent children in kindergarten in 1998-99, who were assessed in fall and spring of kindergarten, fall and spring of 1st grade, and spring of 3rd, 5th and 8th grade.

Sampled Population: The population of U.S. children entering kindergarten in 1998-99. The ECLS-K oversampled Asian and Pacific Islander children.

Predictor Variables and Outcomes: This project seeks to:

(a) identify variables that are risk factors for ADHD among students in 1st-8th grade;
(b) examine over-time patterns of these students’ evaluation and diagnosis for ADHD;
(c) estimate the effectiveness of current treatments for ADHD, including medication, special education and related services, therapy, grade retention or combinations of these treatments, in improving ADHD-diagnosed students’ socio-emotional adjustment, behavior, and academic achievement;
(d) evaluate the effectiveness of these different treatments for varying population subgroups (e.g., girls, racial/ethnic minorities, students from low SES families); and
(e) assess the role of parenting quality and access to health care as mediating variables. 

Contrast Conditions: Natural variation in the ECLS-K allows for the identification of “control” children who

(a) did not display ADHD symptomology (e.g., inattention, hyperactivity) and who were not diagnosed with ADHD,
(b) displayed ADHD symptomology but were not diagnosed with ADHD,
(c) were diagnosed with ADHD, but did not receive services, and
(d) those diagnosed with ADHD who received another type of service or the same service at another intensity level, or in another setting.

The extent of such natural variation is a unique feature of the ECLS-K.

Methodology and Data Analytic Strategy. Our primary research methods are OLS and logistic regression analysis, using weighted adjustments to account for sample clustering. We will also use propensity score matching techniques to better estimate the effects of natural variation in receipt of the aforementioned services. The data analytic strategy involves sequential regression analyses based on the longitudinal nature of the data. We evaluate, in sequence,

(a) which of a range of risk factors contribute to ADHD diagnosis,
(b) how diagnosis of ADHD then leads to the receipt of various intervention treatments, and
(c) whether and to what extent variation in the delivery of such treatments is associated with or predicts improved behavior, socio-emotional adjustment, and academic achievement by students with ADHD, particularly by the end of middle school.