SPINNING THE NARRATIVE: “SPIN” IS HIGHLY PREVALENT IN THE ABSTRACTS OF SYSTEMATIC REVIEWS FROM LEADING ORTHOPAEDIC SURGERY JOURNALS

Systematic reviews (SRs) are critical to advancing high-level understanding of clinical topics in orthopaedics, however as demonstrated in other fields of medicine are often misrepresentative of the data their conclusions are drawn from. “Spin” is a term used to describe such misrepresentation of data with regards to language used in narrative reporting of abstracts within high-level clinical studies. We aimed to quantify the prevalence of spin in systematic reviews from three top impact factor English general orthopaedic surgery journals (Journal of Bone and Joint Surgery, The Bone & Joint Journal, and Clinical Orthopaedics and Related Research). Furthermore, we aimed to test for association between prevalence of spin and bibliometric factors from included studies. MEDLINE, Embase, and PubMED were searched from April 19, 2014 to April 19, 2024 for SRs of comparative nature in the aforementioned journals. Included studies were assessed in duplicate to identify the nine most commonly reported types of spin published by Yavchitz et al. (2016) (Figure 1). Descriptive statistics were calculated for prevalence of spin, and studies with more than one type of spin were classified to have “severe” spin. Chi-Square analysis was completed to investigate the association of spin (both mild and severe) with the journal it was published in. Correlational analyses were conducted for associations between spin prevalence and study year of publication, and number of citations per year. Spin was observed in 26/75 of included studies (34.7%), with 12 presenting with severe spin (16.0%). The most common types of spin found in this sample of reviews were “claiming treatment benefit despite high risk of bias in included studies” (21/75, 28.0%), and “selective reporting on outcomes to overemphasise treatment effect (11/75, 14.7%)”. The presence of severe spin was associated with a statistically significant higher number of yearly citations, however effect size was marginal (mean: 10.6 ±1.9 vs 16.2±5.7, p