OBJECTIVES:

The aim of this study was to empirically compare the results obtained from forced and unforced choice questions in a discrete choice experiment using the same sample.

METHODS:

Patients with advanced or metastatic, and localized breast cancer in six European countries (France, Germany, Ireland, Poland, Spain, UK) completed a discrete choice experiment (DCE). The survey included sixteen DCE scenarios with three options: treatment A or B, and “Opt-out” of treatment. The latter had fixed levels for all the scenarios. Respondents were first asked to choose between the three alternatives in a non-forced choice task. If the “opt-out” option was chosen, the patients were further asked to choose between the two treatments (forced-choice task). Attributes were: Overall survival (OS), Hyperglycaemia, Rash, Pain, Functional well-being (FWB) and Out-of-pocket payment (OOP). Multinomial logit analysis was used to estimate attribute weights, level preferences, and marginal rates of substitution (MRS) for OOP versus the rest of attributes.

RESULTS:

A total of 561 patients responded. Preferences were greatest for avoiding severe pain followed by severe impairment in functional wellbeing in both analyses. The forced choice data produced smaller magnitude of coefficients for the severe levels of pain and impairment in FWB, resulting in different MRS between the analyses. The ordering of the attributes was not affected by the type of the question.

CONCLUSIONS:

This empirical analysis confirms that the hierarchy of the preferences do not differ between the forced and non-forced formats of choice experiments. The issue of inclusion of the “opt-out of treatment” option in DCE in health should be examined considering the sensitivity of price attribute and the econometric implications. However, differences in MRS may be important for decision-making.