Comparison of a full systematic review versus rapid review approaches to assess a newborn screening test for tyrosinemia type 1.

Abstract

Rapid reviews are increasingly used to replace/complement systematic reviews to support evidence-based decision-making. Little is known about how this expedited process affects results. To assess differences between rapid and systematic review approaches for a case study of test accuracy of succinylacetone for detecting tyrosinemia type 1. Two reviewers conducted an "enhanced" rapid review then a systematic review. The enhanced rapid review involved narrower searches, a single reviewer checking 20% of titles/abstracts and data extraction, and quality assessment using an unadjusted QUADAS-2. Two reviewers performed the systematic review with a tailored QUADAS-2. Post hoc analysis examined rapid reviewing with a single reviewer (basic rapid review). Ten papers were included. Basic rapid reviews would have missed 1 or 4 of these (dependent on which reviewer). Enhanced rapid and systematic reviews identified all 10 papers; one paper was only identified in the rapid review through reference checking. Two thousand one hundred seventy-six fewer title/abstracts and 129 fewer full texts were screened during the enhanced rapid review than the systematic review. The unadjusted QUADAS-2 generated more "unclear" ratings than the adjusted QUADAS-2 [29/70 (41.4%) versus 16/70 (22.9%)], and fewer "high" ratings [22/70 (31.4%) versus 42/70 (60.0%)]. Basic rapid reviews contained important inaccuracies in data extraction, which were detected by a second reviewer in the enhanced rapid and systematic reviews. Enhanced rapid reviews with 20% checking by a second reviewer may be an appropriate tool for policymakers to expeditiously assess evidence. Basic rapid reviews (single reviewer) have higher risks of important inaccuracies and omissions.