Methods: 8 preference-sensitive clinical scenarios were devised, involving less-/more-intense options in: the use of specialized medical care (1 scenario); the timing of access to medical care (2 scenarios); and the use of selected interventions in end-of-life care (5 scenarios). A national telephone survey elicited 3,536 Medicare beneficiaries' preferences for the options in each scenario. About 6 months later, for the follow-up study, sub-samples residing in 3 geographic regions with different utilization rates for Medicare services were selected. These individuals were interviewed in person, and again indicated their preferences about the options offered in the same 8 scenarios. (We also obtained more in-depth data about the strength of preference, reasons, and social influences on favored options; these data are reported elsewhere.) The in-person interviewers were blind to the responses collected in the telephone survey.
Results: Of the 585 telephone survey participants in the 3 regions of interest, 366 agreed to be contacted, and 179 completed an in-person interview (participation rate among contacted individuals: 49%). The sociodemographic characteristics and the preferred options reported in the in-person interviews were comparable to the overall distributions in the 3 regions. For all 8 scenarios, the in-person interviewees' favored options were fully concordant with those they had previously selected in the telephone survey
Conclusions: These observations imply that: a) in each region, self-selection into the in-person sub-sample was not biased, in that the people agreeing to the in-person interview were not sociodemographically or preferentially distinct from the telephone-surveyed sample in that region; and b) preferences about more- versus less-intense options in selected clinical scenarios are not systematically affected by the data collection strategy used to reveal those preferences. We conclude that Medicare beneficiaries hold consistent preferences about receiving more-/less-intense health care options, and that these preferences can be efficiently elicited from large sectors of the population using selected scenarios in telephone interviews.