Estimates for diagnostic accuracy of rapid antigen tests for SARS-CoV-2 in systematic reviews are consistently similar despite poor methodological rigor: a methodological overview

Abstract

Objectives: To evaluate the design, conduct, and analysis of systematic reviews on the diagnostic accuracy of rapid antigen tests for SARS-CoV-2 during the COVID-19 pandemic.

Study design and setting: We did a methodological overview of systematic reviews on diagnostic test accuracy, exploring methodological rigor, risk of bias and potential factors of between-review variability.

Results: Of the 31 included reviews, 30 provided summary statistics for sensitivity and 29 for specificity. Summary sensitivities ranged from 56.2% to 91.1%, with a median of 71.5% (IQR 68.3%-76.6%) and a mean of 72.7% with a 7.2 SD. Reported summary specificity estimates were consistently high: median 99.5% (IQR 99%-99.9%) and a mean of 99.3% with a 0.9 SD. We found methodological shortcomings in the systematic reviews, with a majority showing critically low confidence in quality and a high risk of bias.

Conclusion: Despite significant methodological flaws in the reviews, the diagnostic accuracy estimates for rapid antigen tests were characterized by a strong central tendency, highlighting the importance of large sample sizes and broad participant representation. This study suggests the need for further research in diagnostic test accuracy assessments of rigor and risk of bias in systematic reviews.

Keywords: COVID-19; SARS-CoV-2; diagnostic accuracy; methods; overview; rapid antigen test; systematic review.