The identity of two meta-analytic likelihoods and the ignorability of double-zero studies

In a recent paper in Biostatistics, a leading journal in biostatistics and biometry, S3RI member Dankmar Böhning, Professor of Medical Statistics, jointly with Patarawan Sangnawakij, PhD, Assistant Professor in the Department of Mathematic and Statistics at Thammasat University of Thailand publish a result which shows that double-zero studies carry no information on the risk ratio which compares a treatment with a control. For some time it has been debated if the exclusion of studies with no events in treatment and control group in the set of studies investigating the effect of interest could lead to a loss of information. The paper shows that there is no loss of information if double-zero studies are excluded. In contrast, studies in which only one group has no events are informative for the effect of interest. The result is of relevance, for example, in the area of post-admission drug monitoring of adverse events which as quite rare.