Paul S. Mueller, MD, MPH, FACP

Nearly two thirds of meta-analyses were duplicated within the next 3 years.

In recent decades, the number of published meta-analyses of randomized trials of medical interventions has grown markedly. However, the extent of overlap of meta-analyses is unknown.

Investigators identified all meta-analyses published in 2010. From these, 73 (defined as index meta-analyses) were selected randomly. The most common subject areas were cardiology (19%), gastroenterology (12%), and neurology (11%); drugs or biological agents were the most common intervention evaluated. For each index meta-analysis, the investigators searched for overlapping meta-analyses published through February 2013. Overall, at least 1 overlapping meta-analysis (median, 2; maximum, 13) was discovered for 49 index meta-analyses. Among pairs of overlapping meta-analyses for 20 randomly selected topics, 17 more-recent meta-analyses did not include additional primary outcomes, and 13 did not include any additional outcomes. Notably, 11 overlapping meta-analyses of statins for preventing atrial fibrillation after cardiac surgery (with similar patient-eligibility criteria) were published during 57 months. All meta-analyses but the first showed similar positive benefits of statins.

Citation(s):

Siontis KC et al. Overlapping meta-analyses on the same topic: Survey of published studies. BMJ 2013 Jul 19; 347:f4501.