Diagnostic uncertainty in pleural effusion research: why excluding unresolved effusions matters

Pleural fluid biomarkers are increasingly being investigated as tools for evaluating challenging pleural effusions. Nevertheless, a substantial proportion of patients remain without a definitive diagnosis despite cytology, microbiology, pleural biopsy, and clinical follow-up. Most pleural diagnostic studies exclude these unresolved cases, potentially introducing spectrum, selection, and verification biases, as well as overly optimistic estimates of test performance. In this perspective article, we examine the methodological implications of handling unresolved pleural effusions in pleural research. We discuss the limitations of complete-case analyses and review alternative strategies, including intention-to-diagnose frameworks, composite reference standards, longitudinal follow-up, probabilistic adjudication, multiple imputation, sensitivity analyses, and latent class models. Rather than arbitrarily excluding unresolved pleural effusions, future pleural studies should explicitly acknowledge and model diagnostic uncertainty to generate more clinically relevant and generalizable estimates of test performance.