Validity of data extraction in acupuncture meta-analysis: a reproducibility study protocol.

Abstract

Introduction: Systematic review and meta-analysis occupy the apex of the evidence pyramid, serving as the most comprehensive and reliable form of evidence-based assessment. Data extraction is a crucial juncture in meta-analysis, establishing the underpinnings for the outcomes and deductions drawn from systematic reviews (SRs). However, the frequency of data extraction errors in meta-analysis is quite significant. Data extraction errors can lead to biased study results, affect the credibility of study results and even mislead clinical practice. The quantity of acupuncture randomised controlled trials and SRs has expanded rapidly recently, yet the validity of data extraction remains unexplored. Hence, our study aims to investigate the validity of data extraction errors in acupuncture SRs, the effect of data extraction errors on results and the relevant guidelines used erroneous results.

Methods and analysis: Four databases including MEDLINE, Web of Science, Cochrane linbrary and EMBASE will be searched from 1 January 2019 to 31 December 2023 for acupuncture SRs. Two researchers will independently extract data from the meta-analysis and the original study into a standardised data extraction table. A senior investigator, who did not participate in the data extraction process, will verify the results. In cases of discrepancies, the senior researcher will conduct further extraction and consult with another senior researcher to determine the final results. We will analyse the frequency and type of data extraction errors and data estimation errors and evaluate the effect of data extraction errors on results. Quantile regression will be used to explore the factors influencing data extraction error frequency at 25th, 50th and 75th percentiles. Finally, we will further search for guidelines used erroneous results.

Ethics and dissemination: Ethical approval is not necessary for this study. This protocol has been registered in Open Science Framework Registries. REGISTRATION DOI: https://doi.org/10.17605/OSF.IO/CHMPA.