Patient preference-based treatment thresholds and recommendations: A comparison of decision-analytic modeling with the probability-tradeoff technique

Background. Decision analysis (DA) and the probability-tradeoff technique ( PTOT) are patient preference-based methods of determining optimal therapy f or individuals. Using aspirin therapy for the primary prevention of stroke and myocardial infarction (MI) in elderly persons as an example, the object ive of this study was to determine whether group-level treatment thresholds and individual-revel treatment recommendations derived using PTOT are iden tical to those of DA incorporating the patients' own values. Methods. Perso ns in a pilot study of the efficacy of aspirin in the prevention of stroke and MI were asked to participate. Participant values and utilities for pert inent health states (e.g., minor and major stroke, MI, major bleeding episo de) were determined. Then, in three hypothetical clinical situations in whi ch the chance of stroke or MI was varied, PTOT was used to directly determi ne treatment thresholds for aspirin therapy (i.e., the smallest reduction i n MI or stroke risk for which participants would be willing to take aspirin ). Using DA modeling, with the same probabilities of events as in the PTOT exercise and incorporating participants' own values, treatment thresholds f or the three clinical situations were determined. The thresholds determined by the two approaches were compared. Finally, based on these treatment thr esholds, using the best estimates of the efficacy of aspirin to prevent fir st-time stroke and MI, PTOT and DA treatment recommendations for individual participants were compared. Results. The 42 participants reported that a m ajor stroke was the least desirable health state, followed by MI, minor str oke, and major bleeding. The minimum risk reduction required to take aspiri n was greater for MI prevention compared with stroke prevention. For the tw o clinical situations in which the hypothetical efficacy of aspirin to prev ent stroke was varied, treatment thresholds for the PTOT Versus DA approach es differed (p < 0.04), but this difference was not significant (p = 0.19) for the MI-based clinical situation. Using the best estimate of the efficac y of aspirin to prevent first-time stroke and MI, PTOT and DA treatment rec ommendations whether or not to take aspirin were discordant for 38% of part icipants (16 of 42) (2 < 0.001). Conclusions. Patient preference-based grou p-level treatment thresholds and individual-level treatment recommendations can differ significantly depending on whether PTOT or DA is used, apparent ly because the two emphasize different aspects of the decision-making proce ss. DA theory assumes that effective therapeutic decision making should max imize both quality and quantity of life; with PTOT, the emphasis for effect ive clinical decision making allows patients to be fully engaged in the pro cess, thus hopefully leading to fully informed decisions that may result in satisfaction and compliance.