Assessing patients' preferences for treatments for angina using a modified repertory grid method.

A current popular theme in medicine concerns whether and how patients should be involved in treatment choice. Assuming patient involvement is desirable, how should one go about eliciting preferences? A variety of quantitative and qualitative methods exist that may be used for this purpose, one of which is the repertory grid method. This method involves eliciting constructs (reasons) for preferences through comparing sets of three options. This method allows the structured elicitation of the reasons behind individual preferences, but also, when used with generalised procrustes analysis (GPA), allows aggregation of individual data to reveal general preference patterns. In this study the repertory grid method was used to examine patient preferences for angina treatments with the goal of, first, gaining some understanding of general patterns of patient preference, and second, examining the likely utility of the technique in this setting. A sample of 21 patients with mild and stable angina from two general practices in Norfolk, UK was interviewed using the repertory grid method to elicit the constructs underlying their preferences amongst seven angina treatments (including 'no treatment'). Individualised questionnaires were then produced and sent to the patients for self-completion, which required rating the extent to which each construct was relevant for each treatment (scored on visual analogue rating scales). Analysis of the ratings, using GPA, showed that the constructs clustered around two dimensions: 'some treatment' versus 'no treatment', and drug treatment versus surgical treatment. While some treatment was generally preferred to no treatment, individuals varied in preference for drug treatments or surgical treatments. Although the latter were generally perceived as 'effective' they were also perceived, for example, as 'invasive', 'frightening', related to 'negative experiences', and being more appropriate for when symptoms are severe ('proportionate'). We consider the implications of these results for involving patients in choosing amongst treatments.