Randomised clinical trials: advantages and limitations of using parallel-group design or cross-over designs in the field of methylphenidate for ADHD




Poster session 2


Monday 24 October 2016 - 15:30 to 16:00


All authors in correct order:

Krogh HB1, Storebø OJ1, Kielsholm ML1, Nielsen SS1, Simonsen E1, Gluud C2
1 Psychiatric Research Unit, Denmark
2 Copenhagen Trial Unit, Denmark
Presenting author and contact person

Presenting author:

Ole Jakob Storebø

Contact person:

Abstract text
Objectives: Methylphenidate is the most commonly used drug for attention-deficit hyperactivity disorder (ADHD) in children and adolescents. Several randomised clinical trials have assessed the effects of the drug using either parallel-group or cross-over designs. We investigate advantages and limitations of these two designs for the assessment of methylphenidate for ADHD in children and adolescents. Furthermore, we investigate the risks of carry-over effect when using the cross-over design.

Methods: The methods used followed the Cochrane Handbook. Data from randomised clinical trials were included. Authors of cross-over trials where only end-of-period data were available were contacted to obtain data from all intervention periods. Meta-analyses were conducted and the results were presented in forest plots, and Chi² and I² were applied as tests for statistical heterogeneity.

Results: A total of 147 (n = 7134) cross-over trials and 38 (n = 5111) parallel-group trials were included. Differences in intervention effects were tested by comparing parallel-group trials to the first-period of cross-over trials and by analysing the first-period of cross-over trials pooled with parallel-group trials and comparing them to end-of cross-over trials (Chi² = 3.67, df = 1; P = 0.06; I² = 72.8%; 75 trials). The risk of carry-over effect was tested by comparing first-period to the end-of-trial period in cross-over trials (Chi² = 2.47, df = 1; P = 0.12; I² = 59.6%; 4 trials). Parallel-group trials are closer to the real world scenario, and offer better evaluation of the benefits.

Conclusions: Based on the results, both parallel-group trials and cross-over trials are suitable for investigating methylphenidate for children and adolescents with ADHD. The choice of design is, however, important to consider as the parallel-group design offers clear and more realistic benefits. Furthermore, data from cross-over trials are more difficult to include in systematic reviews.