However, the rate of intolerable side effects was considerably lower in this trial (5/61, 8%) than in the RUPP Autism trial (13/72, 18%). This could be due to children with autism being more sensitive to the side effects than other children with ID, but it could also be a difference in design. Simonoff et al. used a clinical titration, presumably with dose adjustments weekly. The RUPP trial, in an initial run-in week, gave 2 days of each dose in ascending fashion, and this accounted for six of the dropouts for intolerability, possibly due to the rapid escalation. Then the 4-week crossover had some orders of condition going from low to high dose or from placebo to medium dose. This may also have contributed to the additional seven dropouts in the crossover phase. This difference between studies is encouraging, because the Simonoff titration strategy is more similar to what is actually done (or should be done) clinically. Nevertheless, even 8% is twice the rate of intolerable side effects for typically developing children with attention-deficit hyperactivity disorder (ADHD; MTA Cooperative Group, 1999). It would be interesting to know how many of the five dropouts for intolerable side effects in the Simonoff et al. study had autism. (By the way, it is an interesting coincidence that 18% of the children in the Simonoff study had ‘none’ selected as their final dose in the clinical trial.)
Another interesting contrast is that in this study the effect size for teacher ratings (d = 0.52) is larger than for parent ratings (d = 0.39), a pattern similar to that of typically developing children (albeit at a lower level), whereas in the RUPP Autism study, the effect by parent ratings (d = 0.89) was larger than by teacher rating (d = 0.48). This could be due to the differences in school placement or situation-specific responses by children with autism. One might wonder if the lower teacher ratings of improvement in the RUPP study could be due to a ceiling effect from teachers in the RUPP study rating children slightly less severe at baseline than parents did. However, such is not the case: teachers in the Simonoff study rated children even less severe than either their parents or RUPP teachers did. (Parents in both studies rated children at baseline about 32 on the Aberrant Behavior Checklist hyperactivity subscale, with RUPP teachers rating about 29 and Simonoff teachers rating about 21 at baseline.) Actually, the pattern difference in improvement ratings between the two studies seems to be due to a difference in parent ratings of improvement, not teacher ratings. If teacher ratings of improvement, similar between the two studies, are accepted as ‘gold standard,’ a question arises about parent ratings: does having autism spectrum disorder in addition to ID moderate the home behavior response to methylphenidate? It would be interesting if Simonoff and colleagues could compare effect sizes and response rates (and intolerable side effect rates) between their autistic subgroup and the rest of their sample to explore this issue.
A possible explanation of some of the outcome differences might be the diagnostic inclusion screen. The RUPP Autism study used DSM-IV ADHD criteria while the Simonoff et al. study used ICD-10 hyperkinetic disorder criteria. In the typically developing MTA sample selected for combined type ADHD by DSM-IV criteria (MTA Cooperative Group, 1999), only about one-fourth (145 of 589) also met criteria for hyperkinetic disorder or hyperkinetic conduct disorder (Santosh et al., 2005). Interestingly, this hyperkinetic subgroup had a substantially larger methylphenidate favorable response than the remaining MTA children, which might lead us to expect a larger effect size in the Simonoff study than in the RUPP Autism study, which was not found.
The Simonoff et al. study sample size (122) allowed a moderator analysis by IQ to test the previous suspicion of an association of IQ with benefit. This was not found. However, as the authors point out, this may be due to not having any subjects with IQ 70 or higher, as were included in the previous studies. Carrying this observation a step further, it does appear there is an association of IQ with favorable response, because the effect sizes and response rate that this study confirms for IDD are considerably below those routinely reported for typically developing children of normal IQ with ADHD, confirming the association across those broad categories. It is interesting that the RUPP Autism Network sample, with IQ ranging up to 135 (mean 62.6) showed a considerably larger effect size by parent ratings, but not teacher ratings, than the Simonoff study, with IQ < 70 (mean 53). This difference in parent-rated improvement with similar teacher-rated improvement raises an interesting speculation: could the IQ moderation suspected by Aman et al. based on parent ratings have arisen from parents of brighter children, presumably also brighter themselves, being better able to detect improvement?