Clustering Errors Form a Confusing Thicket in Obesity Research

It is a pain to sort through the errors that find their way into research publications. Even more painful is the experience of having to retract flawed publications. So when the Editor in Chief of Childhood Obesity retracted a fundamentally flawed, cluster-randomized trial, we see a reason to celebrate. Errors that involve clustering designs in obesity research are all too common.

Editor Joseph Skelton did us a favor in his act of editorial integrity this week.

High Alert on Clustering Study Designs

Researchers from the Indiana School of Public Health in Bloomington alerted the journal to errors in the clustering analysis for this research, leading to unsubstantiated claims that “prebiotic fiber can be helpful” for children at high risk for obesity.

Dean David Allison cautions us that cluster-randomized trials are an abundant source of errors that can mislead us in obesity research:

“When we see the phrase ‘cluster randomized trial’ we should be on high alert. We should elevate our skepticism and carefully evaluate whether the design, analysis, and interpretation lead to trustworthy conclusions.”

Use with Care

None of this tells us that cluster-randomized study designs cannot be valuable and produce important insights. They can – as a different example of well-executed cluster randomization does.

Cluster randomization is a tool. Like a hammer, it can help us construct important scientific insights. It can also cause problems when used clumsily. Thus we need to take care when using this tool or relying on the work it supports.

Click here for the retraction, here for the retracted paper, then here and here for further insights on reasons for caution with cluster randomization.

Violet and Gold, painting by Frederick McCubbin / WikiArt

Subscribe by email to follow the accumulating evidence and observations that shape our view of health, obesity, and policy.

August 29, 2024