Inconsistencies in study eligibility criteria are common between non-Cochrane systematic reviews and their protocols registered in PROSPERO

The author should give careful consideration to the study eligibility criteria of systematic reviews (SRs) and follow it after review protocol development to reduce the possibility of manipulation of inclusion. Our aim was to investigate the prevalence of differences in study eligibility criteria between non-Cochrane SRs and their pre-registered protocols on PROSPERO, and determined what changes were involved as well as whether those changes were explained. We searched the protocols registered on PROSPERO platform in the year of 2018 and then selected these protocols which full-text have been published up to June 9, 2020. A random sample (n = 100) was included. Published full-texts were identified through the protocol's final publication citation. The following five key components of study eligibility criteria were compared: participants, intervention(s)/exposure(s), comparator(s), types of study design, and outcome(s). A total of 90% of included SRs exhibited differences in study eligibility criteria, and 59/90 altered in no less than two key components. Only one SR reported and explained the rationale for changes to the individual key component (the definition of exposure). The Outcome(s) exhibited the greatest variation, with changes in 61% of the SRs. The Comparator(s)/control exhibited the smallest variation, with changes in 20% of the SRs. Differences in study eligibility criteria between the non-Cochrane SRs and their protocols registered on PROSPERO were widespread but were seldom explained. Authors themselves, PROSPERO platform, as well as peer-review journals and their peer-reviewers should play a role in further improving transparency.

Automated summary

The majority of systematic reviews exhibit differences in study eligibility criteria, with many changes occurring in at least two key components, yet these changes are rarely explained. Results show that outcomes have the highest variation, while comparators have the lowest. Authors, the PROSPERO platform, peer-review journals, and reviewers should work towards improving transparency.