The impact of a rapid risk of bias assessment compared to a traditional assessment with QUADAS-2

Abstract Rapid evidence synthesis plays a valuable role in healthcare. However, rapid reviews commonly do not undertake risk of bias assessment, even using truncated implementation approaches, despite its importance. In this work, we aimed to evaluate a rapid QUADAS-2 risk of bias method and compare the results with a standard QUADAS-2 assessment performed on 47 diagnostic test accuracy (DTA) studies for mild cognitive impairment (MCI). We also explored use of a rapid GRADE approach. The overall risk of bias agreed for the majority (29, 61.7%) of studies using the two methods, with major disagreements (high versus low) for two studies (4.3%), and minor disagreements (unclear vs high/low) for 16 studies (34.0%). For the 18 studies which had disagreements, using the QUADAS-2 rather than rapid assessment did not affect the GRADE certainty of evidence for five, five decreased by one grade, and two decreased by two grades. None increased due to the nature of the ‘one strike’ method adopted in the rapid approach. The collaborative nature of the rapid method may be beneficial in terms of robustness due to the inherent subjectiveness involved with any risk of bias assessment. In conclusion, this research demonstrated the suitability of using a rapid method to assess the risk of bias of DTAs. This method could be valuable for incorporating into rapid reviews to provide an evidence synthesis in a short timeframe without overly sacrificing robustness.