Abstract
Aim
A substantial portion of periprosthetic joint infections (PJI) literature is comprised of observational studies. The Strengthening the Reporting of Observational Studies in Epidemiology guidelines emphasize the importance of clearly defining variables and providing diagnostic criteria. Well-defined variables in these studies play a crucial role in ensuring data consistency, fostering comparability among studies, and laying a robust foundation for evidence-based decision-making. We aimed to determine the definition of these variables and determine the objectivity of the definitions.
Method
We reviewed observational studies on hip or knee PJI that focused on variables and their association with treatment outcomes. The inclusion criteria comprised studies from Jan 2017 to Jan 2023. We focused on 13 variables that were possibly subjective. These were smoking, alcohol use, diabetes mellitus, hypertension, lung disease, rheumatoid arthritis, liver diseases, kidney diseases, cardiovascular diseases, malignancy, immunosuppression, use of antibiotics, and type of infection. The reviewers examined the text of the articles, along with any available online supplements or protocols, for definitions of the selected variables. We classified a definition as objective if there was the presence of time element, severity, staging, frequencies, laboratory cut-off, medication dependence, among others. Chart review was deemed subjective.
Results
We included 75 studies in the analysis. The most common factors studied were diabetes mellitus (79%), cardiovascular disease, smoking history, and rheumatoid arthritis (47% each). The variables that were objectively defined most often were antibiotic use (100%) and type of infection (95%). Smoking history (16%) and alcohol use (20%) were the least frequently objectively defined variables. Further analysis revealed that a considerable number of studies incorporated variables into their primary analyses without clear definitions. For instance, out of the 59 studies where diabetes was considered a variable, 41 studies (70%) included diabetes in their main analyses as a factor for PJI treatment outcomes, despite only 34 studies having defined this variable. Moreover, of the 34 provided definitions of diabetes mellitus, only 12 provided objective criteria for diagnosis. The rest of the provided definitions relied on “chart review” without further specification. Table 1 outlines the proportion of studies with variables defined and included in their analysis.
Conclusions
Study variables were not clearly defined in most of the observational studies raising concerns about the reproducibility and reliability of findings. Our study underscores the vital need for standardized variable definitions in PJI research. Professional societies may play a crucial role in setting standards for the definition of variables.
For any tables or figures, please contact the authors directly.
