Predictors of Citations in Neurosurgical Research: A 5-Year Follow-Up

Abstract

Citation rates are an important measure for the impact of articles. This study is the most comprehensive analysis of predictors for scientific neurosurgical research articles. Scientific articles published in 13 neurosurgical journals in 2015 were selected. Data collected included article subject, level of evidence (LOE), journal impact factor (IF), authorship, contributing centers, and study design. Citation counts were collected for each article in Web of Science (WoS) and Google Scholar (GS) 2.5 and 5 years after publication and Scopus 5 years after publication. A generalized linear mixed-effects model using the predictors of search engine, LOE, number of centers, number of authors, and IF was constructed to predict total citation count at 5 years. A total of 2867 articles generated 39,190 citations in WoS, 61,682 in GS, and 43,481 in Scopus. The median number of citations per article was 10 (interquartile range [IQR], 14) in WoS, 15 (IQR, 20) in GS, and 11 (IQR, 15) in Scopus. On average, for every 1 citation in WoS, Scopus and GS identified 1.11 and 1.58 citations, respectively. Significant predictors of citation count in all databases 5 years after publication included search engine, LOE, number of centers, number of authors, number of countries, journal IF, and the month of publication (P < 0.05). The article subject (e.g., tumor or spine) did not significantly predict citation counts. In the most thorough analysis of citation predictors in the neurosurgical literature, search engine, LOE, number of centers, number of authors, number of countries, journal impact factor, and month of publication influenced citations 5 years after publication.