High Prevalence of Causal Language and Inferences in Observational Hip and Knee Arthroplasty Database Studies: A Review of Papers Published Across Four Orthopaedic Journals

Abstract

BackgroundThe use of administrative databases and clinical registries in lower extremity arthroplasty research is growing. Such observational studies are unable to fully control for confounders and cannot establish causality. However, many authors use causal language when describing their aims or findings, potentially misleading readers. We examined the prevalence of causal language and inferences in the lower extremity arthroplasty literature. MethodsWe systematically identified administrative database and registry studies on hip and knee arthroplasty that were published in 4 orthopaedic journals in 2020. Articles were graded independently by two reviewers for the presence of causal language in both the title and abstract and the full text. Chi-squared analyses were conducted to determine the relationship between the causality grading and article characteristics including the journal of publication. ResultsOf 116 eligible articles, we classified 79.3% of titles and abstracts as either consistently causal or inconsistently causal, with only 20.7% as consistently noncausal. A total of 40.5% of full texts were consistently causal, 49.1% were inconsistent, and 10.3% were consistently noncausal. Chi-squared analyses revealed no statistically significant association between the title and abstract’s grading and the journal (P = .720) nor with the use of a database or registry (P = .716). ConclusionCausal language and inferences were present in 79.3% of titles and abstracts of lower extremity arthroplasty observational database studies published in 2020. The high prevalence of causal language and inferences in the arthroplasty literature may mislead readers.