We read with interest the recent article by Goodman et al evaluating the relationship between synovial fluid cellularity and synovial lymphocytic inflammation (SLI) in patients with rheumatoid arthritis.1 Although the use of paired synovial tissue and fluid samples is a notable strength, several methodologic and statistical concerns warrant consideration because they may influence the interpretation and clinical applicability of the findings. First, the diagnostic accuracy of the synovial fluid white blood cell (WBC) count appears to be overestimated. The proposed cutoff value of ≥1,400 cells/μL and the associated sensitivity, specificity, and area under the receiver operating characteristic curve were derived and tested within the same cohort. Methodologic literature clearly demonstrates that deriving and evaluating diagnostic thresholds in a single data set introduces optimism bias in the absence of internal validation techniques such as bootstrapping or cross-validation or confirmation in an independent cohort, thereby limiting reproducibility and generalizability of the reported diagnostic performance.2, 3 Second, the multivariable logistic regression analysis raises concern for overfitting. Only 29 patients were classified as having high SLI, yet multiple correlated predictors were included simultaneously in the regression models. Established statistical guidance indicates that insufficient outcome events per variable can result in unstable coefficient estimates, inflated odds ratios, and misleading confidence intervals, particularly in small samples.4, 5 Third, the interpretation of the synovial fluid WBC count as a predictive marker of SLI represents closely related biologic processes assessed within the same joint and at the same time point. Diagnostic accuracy research has shown that such nonindependence between the index test and reference standard introduces incorporation bias, which may artificially exaggerate apparent predictive or diagnostic performance.6, 7 Fourth, the dichotomization of the ordinal SLI score into low (0–2) and high (3) categories may have contributed to loss of information and inflation of effect estimates. Methodologic studies have consistently demonstrated that arbitrary dichotomization of ordinal or continuous variables reduces statistical power and may overstate associations, particularly in multivariable analyses.8, 9 In summary, although the study by Goodman et al provides biologically plausible and hypothesis-generating observations, the diagnostic and predictive implications of the synovial fluid WBC count for identifying high SLI may be overstated. Further research incorporating appropriate validation strategies, adequately powered multivariable models, and independent cohorts is required before these findings can be translated into clinical decision-making. Disclosure form. Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.
Deb et al. (Wed,) studied this question.