We read with great interest the article by Muckelbauer et al (1), in
which the authors assessed whether a combined environmental and
educational intervention (aimed at promoting water consumption) reduced
the prevalence of overweight among children in elementary schools. The
authors claimed that their study was a “randomized, controlled cluster
trial” including 32 elementary schools (defined as the cluster units), for
a total of 2950 children. However, a scrutiny of this paper led us to
wonder why the authors claimed that this trial was randomized, and we
think the title is misleading for Pediatrics readers.
The authors indeed specified that “randomization was performed at the
city level,” but actually, only two cities were considered. Even for a
cluster randomized trial, the number of randomized units has to be higher.
Readers may also be confused by the school selection process the authors
used. Indeed a random sample of schools from each city was selected to be
included in the study. However in a randomized trial, randomization is
defined as “the process of randomly allocating participants into one of
the arms of a controlled trial,” (2) not as the process of selection of
participants. At the end, the authors’ “randomized trial” is really a non-
randomized controlled study: “[a] quantitative study estimating the
effectiveness of an intervention (harm or benefit) that does not use
randomisation to allocate units to comparison groups” (2). In that sense,
Figure 1, which does not depict randomization, is accurate. The level of
evidence of such a design is somewhat different from the level of evidence
of a randomized trial, and at least, authors should have discussed this
point as a limitation of their study.
To design the study as a cluster randomized trial, schools should
have been randomly allocated to the intervention or the control group,
after considering the city as a stratification factor. Such an approach
was already used by James et al (3) in conducting a similar study. Doing
so may also have prevented schools from declining participation, because
consent of the school’s guardians would have been sought before the
schools they were in charge of were randomized. As pointed out recently
(4), such an approach is essential to prevent post-randomization
withdrawal of clusters, which inevitably leads to discarding these
clusters for the statistical analysis and thus induces selection bias.
1. Muckelbauer R, Libuda L, Clausen K, Toschke AM, Reinehr T,
Kersting M. Promotion and provision of drinking water in schools for
overweight prevention: randomized, controlled cluster trial. Pediatrics
2009;123(4):e661-7.
2. Green S, Higgins J, editors (2005) Glossary. Cochrane handbook for
systematic reviews of interventions 4.2.5. Available:
http://www.cochrane.org/resources/glossary.htm. Accessed 22 April 2009.
3. James J, Thomas P, Cavan D, Kerr D. Preventing childhood obesity by
reducing consumption of carbonated drinks: cluster randomised controlled
trial. BMJ 2004;328(7450):1237.
4. Giraudeau B, Ravaud P. Preventing bias in cluster randomised trials.
PLoS Med 2009;6(5):e1000065.
Bruno Giraudeau
INSERM, CIC 202, Tours, France
Université François Rabelais, Tours, France
CHRU de Tours, Tours, France
INSERM, U738, Paris, France
and
Philippe Ravaud
Assistance Publique–Hôpitaux de Paris, Hôpital Bichat, Département
d’Epidémiologie, Biostatistique et Recherche Clinique, Paris, France,
Université Paris 7–Denis Diderot, Paris, France
INSERM, U738, Paris, France
Conflict of Interest:
None declared
eLetters: ![[Read eLetters]](/icons/misc/sectionDOWN.gif){kind=icon}
