Materials and Methods

The protocol for this retrospective study was registered a priori in an open access database (https://osf. io/puzt9/? view_only= 2d18d6a235b6446dacc267b7fd27877b). Original research studies, including systematic reviews and meta-analyses, published in select radiology journals (Radiology, European Radiology, Journal of the American College of Radiology, Investigative Radiology, and American Journal of Roentgenology) from January 1 to December 31, 2015, were included. Study characteristics, Twitter activity (defined as linking the published article in a tweet), and Altmetric score metrics were collected. Citation counts were obtained through Web of Science. The association between social media metrics (Twitter activity and Altmetric score) and citation rates was assessed using multivariate negative binomial regression models and expressed in terms of odds ratios (ORs) with P values and 95% CIs. The following covariates were included: study design, sample size, and journal of publication. P<. 05 was considered indicative of a statistically significant difference.

Results

Of the 462 included studies, 272 had associated Twitter activity (Table 1). As Table 2 demonstrates, the presence of any Twitter activity was associated with a higher citation rate (OR, 1.29; 95% CI: 1.01, 1.66; P=. 04), with the strongest association seen in studies that were tweeted about more than five times (OR, 1.66; 95% CI: 1.14, 2.40; P=. 01). The association between the presence of any Altmetric activity and citation rate did not reach the statistical significance threshold (OR, 1.26; 95% CI: 0.99, 1.61; P=. 06). Studies with an Altmetric score greater than 5 demonstrated a significantly higher number of citations (OR, 1.72; 95% CI: 1.24, 2.40; P<. 01).