The Use of Causal Language and Inferences in Observational Shoulder Arthroplasty Database Studies.

BACKGROUND: Over the last few decades, several databases and clinical registries have been used to generate studies for orthopedic surgery research including shoulder arthroplasty. Observational database studies present data that is best used to provide information regarding associations between variables but is unable to prove causation. Yet, there remains a tendency for authors to use causal language and conclusive statements in published shoulder arthroplasty literature utilizing such databases.

METHODS: We systematically identified administrative database and registry studies on shoulder arthroplasty in five orthopedic journals from January 2020 to December 2021. Papers were independently graded by two reviewers for the presence, absence, or inconsistent use of causal language throughout the publication. Chi-squared analyses were conducted to determine the relationship between causality language and article characteristics including the journal of publication.

RESULTS: There were 65 eligible articles. 67.7% of titles and abstracts were considered either consistently causal or inconsistently causal in language and inference. When reviewing the full text, 69.2% contained some degree of causal language (38.5% consistently causal, 30.8% inconsistently causal). There were no statistically significant associations between the journal and title and abstract final grading (p = 0.863) or the journal and full text grading (p = 0.726).

CONCLUSION: The use of causal language remains highly prevalent in observational database studies related to shoulder arthroplasty. Understanding the limitations of observational database studies is paramount for accurate interpretation of these publications.