An application of Bradley-Terry-Type models to the measurement of pain

Abstract

A case-study concerning the measurement of pain experienced in the administration of routine subcutaneous injections to patients undergoing regular haemodialysis is presented. The four treatments studied had a 2 times 2 factorial structure: one factor comprised two forms of the anaemia correcting substance erythropoietin, and the other whether an anaesthetic or placebo cream was applied to the injection site. On a visit to the dialysis unit, a patient is given two of the treatments, one in each arm, and asked to nominate the less painful injection; on a subsequent visit the patient is asked to compare the remaining treatments in a similar way. These paired comparisons are analysed by using a version of the Bradley-Terry model which accommodates ties and order effects. Extensions to the method to allow for factorial treatments and dependent judgments are developed.

Publication
Journal of the Royal Statistical Society: Series C (Applied Statistics) 1995; 44(2):243-255

Related