Revision of a method quality rating scale for single-case experimental designs and n-of-1 trials: The 15-item Risk of Bias in N-of-1 Trials (RoBiNT) Scale

摘要

Recent literature suggests a revival of interest in single-case methodology (e.g., the randomised n-of-1 trial is now considered Level 1 evidence for treatment decision purposes by the Oxford Centre for Evidence-Based Medicine). Consequently, the availability of tools to critically appraise single-case reports is of great importance. We report on a major revision of our method quality instrument, the Single-Case Experimental Design Scale. Three changes resulted in a radically revised instrument, now entitled the Risk of Bias in N-of-1 Trials (RoBiNT) Scale: (i) item content was revised and increased to 15 items, (ii) two subscales were developed for internal validity (IV; 7 items) and external validity and interpretation (EVI; 8 items), and (iii) the scoring system was changed from a 2-point to 3-point scale to accommodate currently accepted standards. Psychometric evaluation indicated that the RoBiNT Scale showed evidence of construct (discriminative) validity. Inter-rater reliability was excellent, for pairs of both experienced and trained novice raters. Intraclass correlation coefficients of summary scores for individual (experienced) raters: ICCTotalScore =.90, ICCIVSubscale =.88, ICCEVISubscale =.87; individual (novice) raters: ICCTotalScore =.88, ICCIVSubscale =.87, ICCEVISubscale =.93; consensus ratings between experienced and novice raters (ICCTotalScore =.95, ICCIVSubscale =.93, ICCEVISubscale =.93. The RoBiNT Scale thus shows sound psychometric properties and provides a comprehensive yet efficient examination of important features of single-case methodology.