ADHD Diagnosis Disparities Persist After Adjusting for Everything

What the research found

By the spring of eighth grade, about 7 percent of the white children in a national American cohort had received an ADHD diagnosis at some point since kindergarten. For African American children the figure was about 3 percent, for Hispanic children 4.4 percent, and for children of other racial and ethnic groups 3.5 percent.

The cohort was the Early Childhood Longitudinal Study, 17,100 children who entered kindergarten in the autumn of 1998 and whose parents and teachers were surveyed again in first, third, fifth and eighth grade. Paul Morgan and Jeremy Staff at Pennsylvania State University, with colleagues there and at the University of California, Irvine, used those waves to model the timing of a diagnosis rather than its presence at one moment. Their analysis was published in Pediatrics in 2013.

With nothing in the model but race and time, Hispanic children had odds of diagnosis 56 percent lower than white children, children of other racial and ethnic groups 48 percent lower, and African American children 36 percent lower.

The second model added the child’s sex, age, birth weight, the mother’s marital status and age at the birth, household socioeconomic quintile, region of the country, health insurance coverage, whether the survey was conducted in English, and teacher ratings of classroom behaviour and achievement test scores updated at every wave. The sample for that model was 15,100 children. African American children now had odds 69 percent lower than white children, Hispanic children 50 percent lower, and children of other groups 46 percent lower. For African American children the estimated gap roughly doubled once the controls were in place, and the cumulative figures above are drawn from this model, with every other variable held at its sample average.

Several of the controls did predict diagnosis. Boys had about twice the odds of girls. Each standard deviation of teacher-rated acting out in class raised the odds by 46 percent, while attentiveness, organisation and persistence lowered them, as did higher reading and mathematics scores. Household income quintile predicted nothing once behaviour and achievement were accounted for, and children without health insurance were diagnosed less often than insured children, not more.

Among the 780 children already diagnosed by fifth grade, prescription medication use showed the same pattern. After the full set of controls, African American children had odds of taking medication 65 percent lower than white children with the same diagnosis, Hispanic children 47 percent lower, and children of other groups 51 percent lower. The disparity appeared in fifth grade and again in eighth.

What it means for you and your child

What the study recorded was the arrival of a diagnosis. Parents reported whether a professional had evaluated their child for problems with attention, learning, behaviour or activity level, and whether that professional had given a diagnosis, and the researchers could not check those reports against clinical records. The design says a great deal about who gets identified and very little about how many children in each group would meet the criteria if every child were assessed.

Two readings fit the same numbers. Minority children may be underdiagnosed, which is how Morgan and colleagues interpret the direction and how earlier work has interpreted it. White children may be comparatively overdiagnosed. The authors state plainly that further research is required to establish how much of each is occurring, and they note that unmeasured factors may still contribute to the estimate they attribute to minority status. Under either reading, what differs between the groups is the process of identification.

That is a finding about a system, and it makes a poor basis for reassurance about any individual child. A national cohort that entered kindergarten in 1998 and left the study in 2007 cannot tell a particular family what happened at a particular appointment. What it can supply is a fair question, and a reason the question is fair.

The children in this analysis were statistically equated on how their teachers rated their attention and behaviour and on how they scored on reading and mathematics tests, and they were still diagnosed at different rates. Whatever produced that difference operated somewhere between a child’s observable difficulty and the moment a professional was asked to look. Morgan and colleagues call for pediatricians, school psychologists and teachers to solicit concerns from minority parents more often, a recommendation they rest on earlier literature rather than on anything this analysis tested.

So the useful sentence in a meeting is the one that puts the request on the record: I am asking for an evaluation, and I would like the request and the response noted in the file. If your concern has been acknowledged warmly and then left where it was, this study describes the point at which these children’s paths diverged.