Funnel plot is not misleading
We feel honoured that Lau and coauthors selected our systematic
review of inhaled sodium cromoglycate (1) to illustrate the potential
pitfalls of the funnel plot in the context of publication bias. The
authors seem to suggest that the finding of heterogeneity should have
prevented us from publishing a funnel plot. We do not see why a funnel
plot should not be made, as it is a helpful tool for visual inspection,
with or without statistical heterogeneity.
Furthermore, they suggest that by subdividing our funnel plot (figure
2 in their paper) by age of the patients or by study design, the asymmetry
disappears. We disagree. The subdivided funnel plots illustrate that the
asymmetry is especially present in studies with older children and in
cross-over studies. It should be noted that there is a large overlap
between these two characteristics, as well as with publication period
(early studies were done in older children and often had a cross-over
design). In our paper we explored and discussed the possibility put
forward by Lau et al of age of study subjects or design (cross-over or
parallel) being an explanation for the dichotomy of studies, but concluded
prudently (from the result of meta-regression) that “publication period”
was probably the variable explaining it (1). As we pointed out in our
discussion, in the phase a new drug is marketed, omission of small,
negative studies is not unlikely. In a more recent Cochrane review (2), we
suggested to perform a randomized trial in older children, adequately
powered to test the hypothesis that the effect size in this age group was
overestimated by previously published trials.
1. Tasche MJ, Uijen JH, Bernsen RM, de Jongste JC, van der Wouden
JC.Inhaled disodium cromoglycate (DSCG) as maintenance therapy in children
with asthma: a systematic review. Thorax 2000;55:913-20.
2. van der Wouden JC, Tasche MJ, Bernsen RM, Uijen JH, de Jongste JC,
Ducharme FM. Inhaled sodium cromoglycate for asthma in children. Cochrane
Database Syst Rev 2003;(3):CD002173.
Competing interests: No competing interests