The following criteria were used to designate a scale as “recommended,” “suggested,” or “listed” for use in RLS (Table 1). If an RLS severity rating scale has been applied to RLS-related populations, has been employed by investigators other than the group that originally developed it, and has satisfactory clinimetric properties, that instrument is “recommended;” if it has been applied to RLS, but meets only one of the other two requirements, it is “suggested;” and if it has only been applied to RLS, but does not meet either of the other two criteria, it is “listed”. It is cautioned that the tested clinimetric attributes of the scales can only be applied in the circumstances and to the populations in which they were validated.
For each scale, the determination as to whether the scale (1) has been used in RLS populations, (2) has been employed by investigators other than those that developed it, and (3) has adequate clinimetric properties is summarized under the “key criteria” section for each scale.
In general, the elements considered for proper evaluation of the scales included sensitivity, specificity, positive predictive value, negative predictive value, inter-rater reliability, Cronbach's alpha, factor analysis, criterion validity, convergent validity, divergent validity, known groups validity, and responsiveness to change. The group did not require a minimum number of these variables. Rather, the group made an overall decision about each validation study based upon an overview of the number of the variables and the quality of the data. Determination as to the strength of the statistical results obtained from the studies, for example, whether correlation coefficients were weak, moderate, or high and whether effect sizes were small, moderate, or large, were done in consultation with one of our coauthors (P.M.M.) and by standard methodology.