Capture-mark-recapture as a tool for estimating the number of articles available for systematic reviews in critical care medicine☆

Abstract Introduction Systematic reviews are an important knowledge synthesis tool for critical care medicine clinicians and researchers. With new literature available each day, reviewers must balance identifying all relevant literature against timely synthesis. We therefore sought to apply capture-mark-recapture, a novel methodology, to estimate the population of articles available for a systematic review of effective patient rounding practices in critical care medicine. Methods Capture-mark-recapture was applied retrospectively to estimate the population of articles available for a systematic review of 4 bibliographic databases. All research studies (no methodology restrictions) of patient rounding practices in critical care medicine were included. Estimates of article population size were calculated for search of the bibliographic databases, selection of articles for full-text review, and selection of articles for inclusion in the systematic review. Results Capture-mark-recapture estimated a population of 28 839 articles (95% confidence interval [CI], 12 393-70 990) for search of the bibliographic databases, 169 articles (95% CI, 152-202) for full-text review, and 48 articles (95% CI, 39-131) for inclusion in the systematic review. These estimates suggest that our search identified 15% (4462/28 839) of the population of potentially available articles for the search of the bibliographic databases, 79% (133/169) of articles for full-text review, and 79% (38/48) of articles for inclusion in the systematic review. Conclusions The capture-mark-recapture technique can be applied to systematic reviews in critical care medicine with heterogeneous study methodologies to estimate the population of articles available. Capture-mark-recapture may help clinicians who use systematic reviews to estimate search completeness and researchers who perform systematic reviews to develop more efficient literature search strategies.