Evaluation of Indicators of Reproducibility and Transparency in Published Cardiology Literature

Abstract

Abstract Background The extent of reproducibility in cardiology research remains unclear. Therefore, our main objective was to determine the quality of research published in cardiology journals using eight indicators of reproducibility. Methods Using a cross-sectional study design, we conducted an advanced search of the National Library of Medicine (NLM) catalog for publications from 2014-2018 in journals pertaining to cardiology. Journals must have been published in the English language and must have been indexed in MEDLINE. Once the initial list of publications from all cardiology journals was obtained, we searched for full-text PDF versions using Open Access, Google Scholar, and PubMed. Studies were analyzed using a pilot-tested Google Form to evaluate the presence of information that was deemed necessary to reproduce the study in its entirety. Results After exclusions, we included 132 studies containing empirical data. Of these studies, the majority (126/132, 95.5%) did not provide the raw data collected while conducting the study, 0/132 (0%) provided step-by-step analysis scripts, and 117/132 (88.6%) failed to provide sufficient materials needed to reproduce the study. Conclusions The presentation of studies published in cardiology journals does not appear to facilitate reproducible research. Considerable improvements to the framework of biomedical science, specifically in the field of cardiology, are necessary. Solutions to increase the reproducibility and transparency of published works in cardiology journals is warranted, including addressing inadequate sharing of materials, raw data, and key methodological details.

Ähnliche Arbeiten