A multi-attribute model of prostate cancer patients' preferences for health states

Multi-attribute utility theory (MAUT) provides a way to model decisions inv olving trade-offs among different aspects or goals of a problem. We used MA UT to model prostate cancer patients' preferences for their own health stat e and we compared this model to patients' global judgments of health state utility. 57 patients with prostate cancer (mean age = 70) at two Chicago Ve terans Administration health clinics were asked to evaluate health states d escribed in terms of five health attributes affected by prostate cancer: pa in, mood, sexual function, bladder and bowel function, and fatigue and ener gy. Each attribute had three levels that were used to form three clinically realistic health state descriptions (A = high, B = moderate, C = low). A f ourth personalized health description (P) matched the patient's current hea lth. We first measured patients' preferences using time trade-off (TTO) jud gments for the three health states (A, B, and C) and for their own current health state (P). The TTO for the patient's own health state (P) was standa rdized by comparing it to TTO judgments for states A and C. We next constru cted a multi-attribute model using the relative importance of the five attr ibutes. The MAU scores were moderately correlated with the TTO preference j udgments for the personalized state (Pearson r = 0.38, N = 57, p < 0.01). T hus, patients' preference judgments are moderately consistent and systemati c. MAUT appears to be a potentially feasible method for evaluating preferen ces of prostate cancer patients and may prove helpful in assisting with pat ient decision making.