Single data extraction generated more errors than double data extraction in systematic reviews

Abstract

Background and Objective To conduct a pilot study to compare the frequency of errors that accompany single vs. double data extraction, compare the estimate of treatment effect derived from these methods, and compare the time requirements for these methods. Methods Reviewers were randomized to the role of data extractor or data verifier, and were blind to the study hypothesis. The frequency of errors associated with each method of data extraction was compared using the McNemar test. The data set for each method was used to calculate an efficacy estimate by each method, using standard meta-analytic techniques. The time requirement for each method was compared using a paired t-test. Results Single data extraction resulted in more errors than double data extraction (relative difference: 21.7%, P = .019). There was no substantial difference between methods in effect estimates for most outcomes. The average time spent for single data extraction was less than the average time for double data extraction (relative difference: 36.1%, P = .003). Conclusion In the case that single data extraction is used in systematic reviews, reviewers and readers need to be mindful of the possibility for more errors and the potential impact these errors may have on effect estimates.