Stephen J.LurieMD, PhD, Senior EditorIndividualAuthorJody W.ZylkeMD, Contributing EditorIndividualAuthor
In Reply: Dr Liberati and colleagues and Dr
Sánchez García point out an error in the "Results" section of
our article. The data in Table 2 are based on our initial analyses, which
we revised during the review process. We apologize for the confusion. The
correct data are presented in Table 1.
The chosen cutoff for low- and-high-quality studies is, of course, arbitrary,
but the data in Table 1 clearly
show that there was no reason for choosing a cutoff point on a post hoc basis.
In contrast to the opinion of Liberati et al, we think that SDD studies are
very appropriate to explore the relationship between study quality and observed
effects. As opposed to mortality, the end point of pneumonia is rather weak
since establishing this diagnosis is extremely difficult. Therefore, this
end point is prone to investigator bias, which is consistent with our findings.
At no point did we claim that only RCTs were included.
van Nieuwenhoven CA, van Tiel FH, Buskens E, Bonten MJM. Influence of Methodological Quality on Study Conclusions—Reply. JAMA. 2001;286(20):2544-2547. doi:10.1001/jama.286.20.2542