A pooled estimate can favor treatment while the contributing studies differ considerably. A 2015 pediatric lisdexamfetamine meta-analysis illustrates why heterogeneity belongs beside the headline result.
The evidence base
Maneeton and colleagues included five placebo-controlled randomized trials totaling 1,016 children and adolescents with ADHD. The trials lasted four to 11 weeks and tested oral lisdexamfetamine doses of 30–70 mg/day. Two used crossover designs.
The review synthesized existing trials; it did not recruit another independent 1,016 participants. Its conclusions also concern diagnosed ADHD populations, not healthy cognitive enhancement or illicit methamphetamine exposure.
The symptom estimate was favorable but heterogeneous
Across four studies contributing to the pooled symptom-change analysis, the weighted mean difference was −15.20 ADHD rating-scale points, with a 95% confidence interval of −19.95 to −10.46.
The corresponding I² was 94%. That statistic indicates substantial inconsistency in observed study effects beyond sampling variation under the model. It is not a statement that 94% of patients were heterogeneous or that the treatment worked in 94%.
A random-effects analysis accommodates variation statistically. It does not explain why the trials differed or guarantee that the pooled average applies to a new setting.
Which differences may matter?
Design, observation period, dose handling, participant selection and outcome procedures can affect comparability. Crossover and parallel studies also structure observations differently. The review’s favorable average cannot establish an identical effect at every dose, in every age group or for every child.
Nor should one pooled average be treated as a clinical cutoff or a prediction of the score change an individual will experience.
Safety results need their own interpretation
Overall discontinuation did not differ significantly: risk ratio 0.78, 95% confidence interval 0.46–1.31. Adverse-event discontinuation was also nonsignificant, with a wide interval around its reported ratio.
Pulse increased by a pooled mean of 4.96 beats per minute versus placebo, with a 95% confidence interval of 2.77–7.15. Pooled systolic and diastolic blood-pressure changes were not significant. These distinct results should not be collapsed into either “no physiological effects” or universal cardiovascular harm.
What are the limits of five studies?
The small study count limited assessment of publication bias, and the review identified unclear risk-of-bias features in some trials. Short treatment periods also limit long-term inference.
The defensible historical conclusion retains both the favorable symptom estimate and its substantial heterogeneity. It does not establish uniform efficacy, equivalent safety to placebo or individual functional and driving outcomes.
Scientific source
Maneeton B, Maneeton N, Likhitsathian S, et al. Comparative efficacy, acceptability, and tolerability of lisdexamfetamine in child and adolescent ADHD: a meta-analysis of randomized, controlled trials. Drug Design, Development and Therapy. 2015;9:1927–1936. doi:10.2147/DDDT.S79071. Original commentary on the supplied paper.
Leave a Reply