METHODS: Patients used the aid between their initial medical consultation and when they made their actual decision: 60 men used the interview version and, subsequently, 69 men used the computer version. With both versions, each patient was provided information, identified attributes important to his decision, and completed a values clarification task to arrive at the treatment option most consistent with his values. The values clarification task changed between the versions. The patient then made his actual decision, and 3 months later had a follow-up interview. Outcomes included: Treatment Value Assessment, (TVA) 5-point Likert scales reflecting the patient's rating of each option, his actual treatment choice, and Decisional Conflict Scale, Satisfaction with Preparation for Decision Making Scale, and Regret Scale (all Likert scales: 1-5 high).

RESULTS: Most men in each group were aged 60-69 (interview 52% vs computer 56%) and married (88% vs 89%). The number of treatment options increased between the versions: in the interview group, 45% were offered 2 treatments and 55% offered 3; in the computer group, 25% were offered 2 treatments, 30% offered 3, and 50% offered 4. Despite those differences, the same percentage of patients in the 2 groups showed TVA changes during the decision aid (both 72%, [c²<1]), and a similar percentage had TVA scores that predicted their actual treatment decisions (65% and 60%, [c²<1]). The two groups showed similar decreases in decisional conflict [means: .34 vs .28, t<1], and similar levels of satisfaction with decision support at the times of the aid (means: 4.0 vs 3.8, [t=1.3, p>.1]) and of the actual decision (3.9 vs 3.6, [t=1.8, p>.05] ). However, 3 months later, the computer group experienced less regret, with a substantial increase in patients reporting no regret at all (interview 16% vs computer 43%, [c²=10.3, p=.001]).

CONCLUSION Despite differences between the two versions, most outcomes are similar, thus, appear to provide similar support to patients overall. The difference in regret between the two versions suggests that (a) the impact of values clarification exercises needs further investigation, and (b) longer term outcomes should be included in the assessment of decision aids.