Characteristics of studies falsely excluded during single-reviewer abstract screening: a meta-epidemiological analysis

Abstract screening is a crucial step in systematic reviews, yet it can be prone to errors, leading to the exclusion of potentially relevant studies. Understanding the characteristics associated with falsely excluded studies can inform strategies to improve screening processes and minimize errors. We conducted a meta-epidemiological analysis of a dataset from a methods study, comprising 924 inclusion and exclusion decisions for 73 studies. We examined the study, abstract, journal, and reviewer characteristics associated with the falsely excluded abstracts. The characteristics examined included the study design, publication year, risk of bias, abstract content, country of conduct, journal indexing, SCImago Journal & Country Rank, and reviewer characteristics. Abstract decisions referring to nonrandomized studies of interventions (NRSI) constituted the highest proportion of false exclusions (19.6%; 95% confidence interval [CI], 12.7%–28.8%) compared to nonrandomized controlled trials (NRCTs) (6.7%; 95% CI, 1.2%–29.8%), randomized controlled trials (RCTs) (9.3%; 95% CI, 7.5%–11.7%), and interrupted time series (ITS) (12.7%; 95% CI, 7.0%–21.8%). Abstracts that insufficiently reported study information (18.4%; 95% CI, 14.6%–22.8%) and study design (15.4%; 95% CI, 10.3%–20.6%) were more often falsely excluded by reviewers than abstracts sufficiently reporting study information (6.0%; 95% CI, 4.4%–8.3%) and study design (9.4%; 95% CI, 7.3%–11.5%). To reduce false exclusion decisions in abstracts, study authors should implement clear, structured, and informative abstract reporting, explicitly mentioning the study design and key details. These findings highlight the need for adherence to current reporting standards, reviewer training, and reviewer support (e.g., meetings, discussions), to improve the accuracy of systematic reviews. https://osf.io/7xhkj.