Objective: To test the hypothesis that the clinical severity of subtypes paralleled a gradient of familial severity.
Method: One hundred forty children with attention-deficit/hyperactivity disorder (ADHD) and 120 normal control children and their biological relatives were studied: Because these data had been collected prior to the publication of DSM-IV, DSM-III-R symptoms were used to approximate DSM-IV subtypes using a method the authors had validated in prior work.
Results: The first prediction from the hypothesis was true: rates of ADHD among relatives of each subtype group were greater than rates among relatives of controls. But the second prediction did not hold: rates of ADHD were not significantly higher among relatives of combined-typed probands compared with relatives of other probands. The "gradient model" also predicted that subtypes would not "breed true" (i.e., the subtype of the relative would not be the same as that of the proband). The prediction of nonspecificity was refuted for the inattentive and combined subtypes, but hyperactive-impulsive ADHD was found almost exclusively among relatives of hyperactive-impulsive probands.
Conclusions: Although the results are limited by some small subsamples along with the use of a DSM-III-R-ascertained sample, they provide little evidence for the idea that DSM-IV subtypes of ADHD correspond to familially distinct conditions. They also do not confirm the idea that the subtypes fall along a gradient of familial severity. Instead, they suggest that symptom differences among subtypes are due to nonfamilial, environmental causes.