A Cross-sectional literature survey showed the reporting quality of multicentre randomised controlled trials should be improved.

Abstract

OBJECTIVE\nTo assess the reporting quality of randomised controlled trials (RCTs) with multicentre design, particularly whether necessary information related to multicentre characteristics was adequately reported.\n\n\nSTUDY DESIGN AND SETTING\nThrough a search of 4 international electronic databases, we identified multicentre RCTs published in English from 1975 to 2019. Reporting quality was assessed by the CONSORT (Consolidated Standards of Reporting Trials) checklist (37 items) and by a self-designed multicentre-specific checklist (27 items covering multicentre design, implement and analysis). The scores of trials published in three time periods (1975-1995; 1996-2009; and 2010-2019) were also compared.\n\n\nRESULTS\nA total of 2,844 multicentre RCTs were included. For the CONSORT checklist, the mean (standard deviation) reporting score was 24.1 (5.5), 12 items were assessed as excellent (>90%), 12 items as good (50%-90%), and 13 items as poor (<50%). For the multicentre checklist, the reporting score was 3.9 (2.2), only 3 items were excellent or good, and the remaining 24 items were poor. Time period comparison showed that reporting quality improved over time, especially after the CONSORT 2010 issued.\n\n\nCONCLUSION\nAlthough CONSORT appears to have enhanced the reporting quality of multicentre RCTs, further improvement is needed. A CONSORT extension for multicentre trials should be developed.