Pooling of cohort studies and RCTs affects GRADE certainty of evidence in nutrition research

Background: There is a little empirical evidence of the impact of pooling randomized controlled trials (RCTs) and cohort studies (CSs) on the certainty-of-evidence. To evaluate the hypothetical-scenario of pooling bodies-of-evidence from RCTs with matched bodies-ofevidence from CSs on the certainty-of-evidence.Methods: We extracted GRADE ratings of bodies-of-evidence from RCTs in Cochrane reviews, and rated the certainty-of-evidence from matched bodies-of-evidence from CSs. We then evaluated the impact of pooling both bodies-of-evidence on the overall certaintyof-evidence, and on individual GRADE domains.Results: Fourty-two pooled bodies-of-evidence were rated, ranging from very-low (bodies-of-evidenceRCTs: 9.5%; bodies-of-evidenceCSs: 40.5%; pooled-bodies-of-evidence: 0%) to low (bodies-of-evidenceRCTs: 38.1%; bodies-of-evidenceCSs: 45.2%; pooled-bodies-ofevidence: 19.1%), moderate (bodies-of-evidenceRCTs: 33.4%; bodies-of-evidenceCSs: 14.3%; pooled-bodies-of-evidence: 57.1%), and high (bodies-of-evidenceRCTs: 19%; bodies-of-evidenceCSs: 0%; pooled-bodies-of-evidence: 23.8%). Certainty-of-evidence was downgraded mostly for imprecision and risk of bias for bodies-of-evidence from RCTs, and for risk of bias and inconsistency for bodies-ofevidence from CSs. Pooling both bodies-of-evidence mitigates rating down for imprecision compared to bodies-of-evidence from RCTs and inconsistency compared to bodies-of-evidence from CSs.Conclusion: Our hypothetical study suggests that pooling both bodies-of-evidence would reduce the amount of very-low and low certainty-of-evidence ratings, but how to integrate RCTs and CSs and whether or not to pool these bodies-of-evidence requires proper guidance before systematic review authors or guideline developers should consider this approach. (c) 2022 Elsevier Inc. All rights reserved.

Automated summary

The study evaluated the impact of pooling evidence from randomized controlled trials and cohort studies on certainty-of-evidence. Pooling both types of evidence reduced the amount of very-low and low certainty ratings, but proper guidance is needed before considering this approach.