As we reported on Friday, a small, 17-person study published by researchers from the University of Illinois suggested they had the perfect treatment for kids with attention deficit disorder (ADHD) — a walk in the park.
We’re all for the most natural, simple treatments available. And this one really sounded good! What could be better than telling parents, “Hey, take your kid for a walk in the park and your child’s ADHD will be less severe.”
So why does this study make us hold our noses?
Besides the small sample size (which should raise warning bells for anyone who thinks this could apply to everyone), the other red warning flag to me was the use of the Digit Span Backwards test as the sole measure for the subjects. Why would the researchers choose this measure when there are perfectly good and well-accepted ADHD research measures available?
Good question, so let’s find out.
I first looked into the literature of the use of the Digit Span Backwards (DSB) test, because while I’m familiar with it, I never heard of it being used as the sole type of measure for severity of ADHD. It is most commonly used to assess neurological or cognitive deficits in people with neurological or cognitive problems (such as people with dyslexia, brain damage, alcohol drinking, dementia, and so forth). It’s also used to test memory performance under a wide range of conditions (such as testing, traumatic brain injury, long-term depression, etc.). In 90 citations I reviewed in PsycINFO, only 9 were found where this measure was used in connection with ADHD or with attention issues in children. Another 9 were found (some duplicates) when I also reviewed the citations the researchers referred to in their study.
In using Digit Span Backwards as a measure to test the performance of ADHD, one study found no deficit for the 26 children in the study (O’Donnell, 2005). This was after the recommendation cited (Hale, Hoeppner, & Fiorello, 2002) by the University of Illinois researchers was published, suggesting that the findings from the older study might not be robust enough to replicate for a small study (especially where sensitivity would be important). Other studies found a significant effect for only one type of ADHD — the combined type not the inattentive type. These differences and issues are largely glossed over by the current researchers.
Here’s the real kicker, however, left out of every media report published on the study:
Three other measures were administered after DSB: the Stroop Color-Word Test (Dodrill, 1978), Symbol
Digit Modalities (SDM) (Smith, 1968), and the Vigilance Task of the Gordon Diagnostic System Model 111-R (VT) (Gordon, McClure, & Aylward, 1996). Although all four measures have been widely used to test attention and impulse control in ADHD and other populations in between subjects designs, Stroop displayed practice effects (demonstrated in a repeated measures ANOVA comparing scores by session number) F(2,17) = 16.90, p < .0001, and Stroop, SDM, and VT displayed low power (power = .05, .28, and.07, respectively, in repeated measures ANOVAs comparing scores by condition). Thus, these measures are not further described here.
So the researchers had to throw out 3 of the 4 measures they were using to measure their results. This was after the study was completed because 8 of the original 25 subjects were dropped from the study (for reasons of non-compliance, errors in following directions or simply not completing the study — do any of these things sound familiar?). So you’re already starting off with a drop-out rate of 32%, meaning this technique will be useless in 1 out of 3 children it’s tried on. But then you add the insult of dropping 75% of the measures you were using and are left with a single remaining measure that has only 0.75 power and isn’t commonly used as a sole treatment measure for ADHD?