[How to evaluate the quality of information transfer from physician to patient? Choice of psychometric tests for a decision tree in a Regional Cancer Centre].
BACKGROUND: Decision boards are used to transfer information from physicians to patients to help them participate in the clinical decision-making process. We present the tests and results of the psychometric properties of a decision board in a sample of healthy volunteers. METHODS: In the Regional Cancer Centre located in Lyon, we developed a decision board for post-menopausal women with breast cancer after lumpectomy without any poor prognostic factors. Two treatment options were proposed, one involving chemotherapy and the other not. We tested for the following psychometric properties: comprehension, construct validity and reliability. Comprehension was evaluated using a questionnaire, in order to test whether the rates of correct answers were due to chance alone. The construct validity was assessed by changing the information provided (relapse and survival rates, characteristics of chemotherapy) and testing whether the proportion of healthy volunteers choosing an option changed in a predictable and significant way. The reliability was evaluated using the test-retest method. Two reliability statistics were computed: the Pearson correlation and the Intraclass Correlation Coefficient. RESULTS: In the sample of 40 healthy volunteers, 23 chose the option with chemotherapy and 17 the option without chemotherapy. Results show that the decision board was comprehensive, valid (the women changed their choices in a predictable way) and reliable (Pearson correlation and Intraclass Correlation Coefficient close to 1). CONCLUSION: The choice of the psychometrics properties tested and the statistical tests used are discussed. The psychometric properties of our tool are found to be satisfactory.