Diagnosing endocarditis in patients with Staphylococcus aureus bacteremia.


should have been discarded in favor of a model that allowed for study-to-study variability. Reporting only pooled results when the I2 statistic is large is a recipe for misleading results. With modern-day sample sizes being relatively small, a large observed I2 statistic ismore likely to reflectmodel failure than the failings of the statistic itself. Third… (More)
DOI: 10.1001/jama.2014.16951


  • Presentations referencing similar topics