Letters Section Editor: Stephen J. Lurie,
MD, PhD, Senior Editor.
To the Editor: In their systematic review and
meta-analysis, Mr Bekelman and colleagues1 concluded
that the industry-sponsored research was more likely to have results favorable
to the sponsor, and that it was also associated with restrictions on publication
and data sharing.
We have several concerns about their methodology. First, the review
was restricted to articles published in English. This may have contributed
to a language bias as studies with positive results are more likely to be
published in English.2- 4 Second,
the authors pooled data from 8 systematic reviews and stated that the homogeneity
test showed that the odds ratios (ORs) of numbers of positive studies sponsored
by industry in each systematic review were not significantly different. However,
the pooled results represented many different types of studies including randomized
controlled trials, an economic analysis, and retrospective cohort studies
on topics ranging from oncology to nicotine and cognitive performance. It
is also unclear why the authors pooled the results of only 8 of the 11 systematic
reviews they listed.
Sawka AM, Thabane L. Effect of Industry Sponsorship on the Results of Biomedical Research—Reply. JAMA. 2003;289(19):2502-2503. doi:10.1001/jama.289.19.2502-a