|
Abstract:
|
An objective, quantitative measurement of assessing severity and outcomes plays a vital role in examining the effectiveness of a treatment. In aesthetic procedures, baseline status and post-treatment outcomes are best measured using photonumeric scales that allow raters to reproducibly determine treatment effects. Often, the reliability or agreement of raters on a scale is measured using intra-class correlation coefficient or weighted kappa, which are affected by the base rate issue where inter- and intra-rater agreements are artificially low when subjects are not equally distributed across all grades in the scale. This is inherently realized in multiple-scaled studies where several features of subjects are judged simultaneously using various scales. This makes it operationally difficult to achieve equal number of subjects across all grades for the different scales. In this research, a mechanism is developed that minimizes this imbalance across grades and thus diminishes the impact of the base rate problem on reliability indexes. Simulated and real-life data are used to show how this method reduces the effect of base rate issue by introducing balanace across grades of a scale.
|
