Table 4.

Crossover Analysis of the Swanson, Kotkin, Agler, M-Flynn, and Pelham Scale-D Global Scores (Mean Value from Hours 1 to 3 of School) Comparing the Three Methylphenidate Preparations; n is the Number of Attention-Deficit/Hyperactivity Disorder Children and Adolescents with Complete Data on the Relevant Variables

	p-value and effect size estimators (one-sided 97.5% confidence intervala)
Hypothesis tested	ITT-population (n=101)	“Best” case scenario (n=107)	“Worst” case scenario (n=107)	PP-population (n=89)
I. Test for noninferiorityb,c of Medikinet retard with a higher IR component and approximately equivalent daily dose versus Concerta	p<0.0001 (−∞, −0.217)	p<0.0001 (−∞, −0.242)	p<0.0001 (−∞, −0.207)	p<0.0001 (−∞, −0.199)
IIa. Test for superiorityb of Medikinet retard with a higher IR component and approximately equivalent daily dose versus Concerta	0.0009	0.0005	0.0036	0.0004
IIb. Test for noninferiorityb,c of Medikinet retard with similar IR components and a lower daily dose versus Concerta	p<0.0001 (−∞, +0.051)	p<0.0001 (−∞, +0.042)	p<0.0001 (−∞, +0.072)	p<0.0001 (−∞, +0.038)

^aIf the confidence interval is completely negative, Medikinet retard is shown to be superior.

^bAccording to the method of Duchateau et al. (2002).

^cΔ is the noninferiority margin, that is, Medikinet retard is allowed to be slightly worse than Concerta.

ITT=intent-to-treat; PP=per-protocol;