An Exploration of Possible Moderators in an Addressing Disparities Comparative Effectiveness Trial for Elementary Students with ASD or ADHD

Background: As we grapple with the importance of determining patient-centered outcomes, those of us conducting intervention research with individuals who have neurodevelopmental disorders need to explore the possible moderating effects of important variables such as age, IQ, language(s) spoken in the home, income level and ethno-racial identity.

Objectives: This study presents the exploratory analyses of possible moderating variables from a larger comparative effectiveness research (CER) trial that may be used for hypothesis generation for future studies.

Methods: The trial compared Unstuck and On Target (UOT) to Contingency Behavior Management (CBM) in 21 Title I elementary schools in three school districts for students with ASD or ADHD. Both interventions targeted executive functioning skills in the classroom. The primary outcome for these analyses are academic classroom observations on the following behaviors: demonstrates social reciprocity, follows rules, transitions appropriately, gets stuck, displays negativity, and participates in class (total raw scores ranging from 0 (worst) to 6 (best)). IQ was measured by the WASI-II. Family income ranged from $8,796-400,000 and was split into tertiles (<35,000; 35,000-100,000; >100,000).

Results: Diagnosis is an important moderator, with ADHD students making significant treatment gains in both interventions (d=.45-.70), while ASD students only showed a positive effect in UOT (d=.42-.60). Importantly, age (r=-.033; p=.69), IQ (r=.165; p=.055) and income (r=.062; p=.495) do not significantly correlate with change in classroom behavior. This lack of significance remains true when analyzed separately by ADHD and ASD. There are three categories for language spoken at home: English only, Spanish only, and More than one language spoken. To be included in the study, parents needed to be able to understand parent trainings, handouts and parent questionnaires in either English or. Overall, language spoken in the home does not moderate the effectiveness of UOT as measured by blinded classroom observations (F=.001), but it does a little bit more with CBM, with the children whose families spoke English only in the home making the least amount of progress, though again, this change was not statistically significant (F=.191; see Table 1).

When looking at potential descriptive differences by race/ethnicity and intervention, it appears that Other (primarily reported as mixed race) students change the most, followed by Black students in UOT, Latino students in either intervention, with White students in both interventions and Black students in CBM making the smallest positive changes. None of these results were statistically significant on t-tests.

As stated above, there is not a significant linear relationship between income and change. However, because income was the specific disparity that this CER trial was designed to address, further exploration is warranted. Descriptively, high- and low-income students changed the most, though these differences are not significant (F=.185).

Conclusions: Overall, there may be some important moderators of effectiveness that could be explored in future large-scale effectiveness trials. It is particularly interesting that the hypotheses generated from these descriptive analyses include speaking languages other than English at home, being non-White and not from a middle-income family may predict greater change in hypothesis-, treatment-, and diagnosis-naïve classroom observations.