Interpreting interrater reliability coefficients of the Braden scale: a discussion paper

There are many studies investigating psychometric properties of the Braden scale, a scale that predicts the risk for pressure ulcers. The main focus of these studies is validity as opposed to reliability. In order to estimate the degree of interrater reliability a literature review revealed that numerous statistical approaches and coefficients were used (Pearson's product-moment correlation, Cohen's kappa, overall percentage of agreement, intraclass correlation). These coefficients were calculated for the individual items and the overall Braden score and were used inconsistently. The advantages and limitations of every coefficient are discussed and it is concluded that most of them are inappropriate measures. Therefore, estimating the degree of the Braden scale interrater reliability is limited to a certain extent. It is shown that the intraclass correlation coefficient is an appropriate statistical approach for calculating the interrater reliability of the Braden scale. It is recommended to present intraclass correlation coefficients in combination with the overall percentage of agreement.