Objective: Although reports of childhood status are necessary for making a diagnosis of adult attention deficit hyperactivity disorder (ADHD), systematic investigation of the accuracy of retrospective self-reports has been limited. This study examined accuracy of adult recall of childhood ADHD.
Method: Participants were from a controlled, prospective 16-year follow-up of children with ADHD. At a mean age of 25 years, 176 probands (85% of the 207 subjects in the initial cohort) and 168 non-ADHD comparison subjects were interviewed by clinicians who were unaware of the subjects' childhood status. Subjects were asked about specific childhood ADHD behaviors, and the diagnosis of childhood ADHD was retrospectively established.
Results: Seventy-eight percent of the probands and 11% of the comparison subjects were identified as having childhood ADHD. Six symptoms demonstrated high discriminating power in differentiating the subject groups: distractibility, concentration difficulties, complaints of inattention, acting before thinking, being on the go, and fidgeting/squirming. When findings were adjusted for the prevalence of ADHD in the general population, the power of prediction was low. Positive predictive value was 0.27, i.e., of all adults retrospectively given a diagnosis of childhood ADHD, only 27% would be correctly identified. As expected, positive predictive value increased with increases in the estimated prevalence of ADHD.
Conclusions: Retrospective diagnoses of childhood ADHD made on the basis of self-reports will in most cases be invalid in settings such as epidemiological surveys and primary care facilities. Greater accuracy can be expected in settings in which childhood ADHD is frequent. The results stress the importance of obtaining contemporaneous information on childhood symptoms in establishing a childhood history of ADHD. Future directions and implications for DSM-V are discussed.