Extract

In an attempt to keep pace with the increasing number of trials being conducted each year, journals make articles available as E-publications ahead of print. E-publications are not available to search through the conventional databases (MEDLINE, EMBASE, CENTRAL) used in systematic reviews, but are searchable using PubMed. We used a search syntax designed to exclusively identify E-publications in PubMed to assess the importance of searching for E-publications in systematic reviews. Two case studies were conducted: updating de novo systematic reviews in particularly active areas of current research, type 2 diabetes mellitus and advanced melanoma. A search for E-publications was conducted concurrently to the conventional systematic reviews. Network diagrams were constructed with and without the results of the E-publications search to demonstrate the potential impact E-publications could have on any evidence synthesis. The advanced melanoma systematic review conducted in conventional databases identified nine studies. The E-publication search identified three additional studies reporting information for three new interventions and additional information for five interventions. Critically, if an evidence synthesis were to be conducted the identification of one of the pivotal nivolumab trials, CheckMate 067, (ipilimumab, nivolumab and ipilimumab+nivolumab) allows the connection of an otherwise disconnected evidence network. The diabetes systematic review conducted in conventional databases identified 28 studies. The E-publication search identified one additional study including an extra intervention; if evidence synthesis were feasible, the E-publication would add a loop to the evidence network which could influence analysis results. Failure to search for E-publications ahead of print may mean that evidence syntheses do not take into account all the data publicly available at the time of review.