Rating the methodological quality of single-subject designs and n-of-1 trials: Introducing the Single-Case Experimental Design (SCED) Scale

Abstract

Rating scales that assess methodological quality of clinical trials provide a means to critically appraise the literature. Scales are currently available to rate randomised and non-randomised controlled trials, but there are none that assess single-subject designs. The Single-Case Experimental Design (SCED) Scale was developed for this purpose and evaluated for reliability. Six clinical researchers who were trained and experienced in rating methodological quality of clinical trials developed the scale and participated in reliability studies. The SCED Scale is an 11-item rating scale for single-subject designs, of which 10 items are used to assess methodological quality and use of statistical analysis. The scale was developed and refined over a 3-year period. Content validity was addressed by identifying items to reduce the main sources of bias in single-case methodology as stipulated by authorities in the field, which were empirically tested against 85 published reports. Inter-rater reliability was assessed using a random sample of 20/312 single-subject reports archived in the Psychological Database of Brain Impairment Treatment Efficacy (PsycBITETM). Inter-rater reliability for the total score was excellent, both for individual raters (overall ICC = 0.84; 95% confidence interval 0.73–0.92) and for consensus ratings between pairs of raters (overall ICC = 0.88; 95% confidence interval 0.78–0.95). Item reliability was fair to excellent for consensus ratings between pairs of raters (range k = 0.48 to 1.00). The results were replicated with two independent novice raters who were trained in the use of the scale (ICC = 0.88, 95% confidence interval 0.73–0.95). The SCED Scale thus provides a brief and valid evaluation of methodological quality of single-subject designs, with the total score demonstrating excellent inter-rater reliability using both individual and consensus ratings. Items from the scale can also be used as a checklist in the design, reporting and critical appraisal of single-subject designs, thereby assisting to improve standards of single-case methodology.