The PRECIS-2 tool seems not to be useful to discriminate the degree of pragmatism of medicine masked trials from that of open-label trials.

Abstract

To assess, with all available trial information, whether the assessment of the PRECIS-2 nine domains could provide a clear distinction between medicine masked pragmatic randomized controlled trials (pRCTs) and open-label pRCTs. A search was conducted of participant-level pRCTs on medicines published on 25 influential medical journals in July 2018-December 2019. All pre-licensing (phases 1-3) and cluster pRCTs were excluded. All trials' available reports were searched through the published article information, Google Scholar, and trial websites. Instead of providing a score to each PRECIS-2 domain, these were classified as E (explanatory), N (neutral), or P (pragmatic). Of 128 pRCTs, 18 (14%) were participant-level pRCTs on medicines. The full trial protocol was available for 14 trials; 12 had published the protocol and nine had additional reports published. All trials were prospectively registered, and none was funded by industry. Ten and eight were masked and open-label trials, respectively. Masked pRCTS had 34% of pragmatic and 60% of explanatory domains; open-label pRCTS had 45% pragmatic and 45% explanatory domains. Among the 10 masked trials, only one had a majority of five pragmatic domains; among the eight open-label trials, four had a majority of six or five pragmatic domains. "Follow-up" was considered explanatory in the 18 pRCTs; "primary analysis" was pragmatic in 17 pRCTs. The PRECIS-2 tool seems not to be sensitive enough to clearly discriminate between medicine masked pRCTs and open-label pRCTs. When conducting systematic reviews, it is suggested that the PRECIS-2 tool should not be used to support placing masked trials in the pragmatic side of the explanatory/pragmaticcontinuum.