Drivers of citation impact in leading pharmaceutical sciences journals: a hybrid manual and API-driven bibliometric analysis

Abstract Background Citations may be used as a metric of scholarly impact, yet traditional bibliometric approaches often rely on manual data extraction from citation databases that presents scalability constraints. Hybrid manual and Application Programming Interfaces API-driven approaches may be used to systematically identify predictors of citations in scientific journals. Objective To identify whether journal, thematic, methodological, geographic, open access, or promotion factors drive citations in three leading pharmaceutics journals and to evaluate a hybrid manual/API-driven approach for bibliometric data extraction and analysis. Methods A cross-sectional analysis of original research articles published between 2016 and 2018 in the AAPS Journal , the European Journal of Pharmaceutical Sciences , and the Journal of Pharmaceutical Sciences was conducted. Article metadata were extracted manually and using API pipeline (NCBI E-utilities, Google Scholar, Altmetric APIs). The dependent variable was citation count. Predictor variables included publication year, journal, study topic, design, geographic region, open-access status, route of administration, formulation type, therapeutic area, social-media mentions, number of disciplines, institutions, and references. The dependent variable was dichotomized using median number of citations, and multivariable logistic regression (α = 0.05) was used to identify independent predictors of citations above median. Results Of the 3,510 records identified, 3,252 original research articles met inclusion criteria. The median citation count was 23 (range 0–976). In the multivariable model, publications from 2017 (OR 0.784; 95% CI: 0.664–0.926) and 2018 (OR 0.581; 95% CI: 0.488–0.691) had lower odds of being cited above the median compared to 2016 papers. Among study designs, review articles had over threefold higher odds of being cited above the median compared to basic science studies (OR 3.385; 95% CI: 2.515–4.606). Modeling and simulation studies had lower odds of citations above median (OR 0.754; 95% CI: 0.612–0.927) compared to basic science studies. Animal-based research was less likely to be cited above the median compared to in-vitro studies (OR 0.611; 95% CI: 0.480–0.775). Papers with corresponding authors from Africa (OR 2.165; 95% CI: 1.341–3.590) and the Middle East (OR 1.936; 95% CI: 1.226–3.121) were more likely to be cited above the median compared to those from Europe. Papers with no specified route of administration (OR 0.803; 95% CI: 0.666–0.968) and those on parenteral drugs (OR 0.689; 95% CI: 0.549–0.863) were less likely to be cited above the median compared to articles on orally administered drugs. Open-access papers (OR 1.240; 95% CI: 1.064–1.454) and those with over two X mentions (OR 1.572; 95% CI: 1.135–2.191) were more likely to be cited above the median. Papers describing colloidal systems (microspheres/hydrogels) were more likely to be cited above the median (OR 1.598; 95% CI: 1.187–2.158) compared to articles on solid formulations. As for the therapeutic focus, infectious-disease studies were less likely to be cited above the median (OR 0.622; 95% CI: 0.500–0.774) compared to no specific disease articles. Each additional discipline increased the odds of the article being cited above the median by 20% (OR 1.205; 95% CI: 1.088–1.337), and each extra reference by 2% (OR 1.020; 95% CI: 1.017–1.024). Conversely, each additional collaborating institution reduced the odds of the paper being cited above the median by 10% (OR 0.896; 95% CI: 0.804–0.997). Conclusions Citation counts in pharmaceutics articles are influenced by various factors, including article type, methodology, geographic origin, open access status, and promotion via social media. The presented hybrid citation analyses may streamline future bibliometric research across pharmaceutics and other scientific fields. Graphical Abstract