Levels of evidence ratings in the urological literature: an assessment of interobserver agreement

Abstract

OBJECTIVE: To determine to what extent urologists with no specific training agree upon level of evidence (LoE) ratings of studies published in the urological literature, as LoE are commonly referenced as a measure of evidence quality. MATERIALS AND METHODS: In all, 86 clinical research studies published in four major urology journals were reviewed. Each article was independently reviewed by eight reviewers using a standardized data abstraction form. Articles were assessed for type of study (therapy, prognosis, diagnosis or economic) and LoE (I, II, III or IV). Reviewers received only written instructions and no formal training in the application of this classification system. RESULTS: Of the 86 articles, 69% related to therapy, 16% to prognosis, and 15% to diagnosis. Eight studies (9%) provided Level I evidence, 18 studies (21%) Level II, 14 studies (16%) Level III and 46 studies (54%) Level IV evidence. The intraclass correlation coefficient (95% confidence interval) based on all reviewers (eight reviewers) was 0.67 (0.59-0.74; P= 0.001) for the type of study and 0.55 (0.48-0.64; P= 0.001) for the LoE. In an analysis limited to a subset of studies in which all reviewers agreed upon the type of study question (n= 40) the intraclass correlation coefficient was 0.79 (0.70-0.86; P= 0.001). CONCLUSION: In the present study there was a low interobserver agreement for LoE ratings by urologists with no specific training. These findings suggest caution in the interpretation of LoE ratings and emphasize the importance of specific training for individuals that are charged with quality of evidence determinations.

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 15 citations exist upstream.

Papers that cite the same references — the relational fallback while embeddings are still backfilling.