The default effect in end-of-life medical treatment preferences

Living wills are intended to preserve patient autonomy, but recent studies suggest that they do not always have their desired effect. One possible explanation is that living wills do not capture the authentic preferences of the patients who write them but instead reflect transient contextual effects on preferences.

Two experiments examined whether end-of-life treatment preferences expressed in a living will were influenced by the presence of default options.

College students participated in 2 Web-based questionnaire experiments (Ns = 182 and 51). Participants were randomly assigned to 1 of 2 or 3 default conditions.

In experiment 1, participants expressed significantly different treatment preferences in 3 normatively equivalent, check box-formatted living wills that were either positively worded ("indicate medical treatments you would want administered"), negatively worded ("indicate treatments you would want withheld"), or of forced-choice format (P = 0.01). Participants expressed a stronger preference to receive treatment in the negatively worded document than in the positively worded document as a consequence of preferring the default option in both cases. Participants in experiment 2 were also influenced by the presence of a default option, but this time, while writing narrative living wills after viewing 1 of 2 sample living wills. In this experiment, the sample living will represented the default preference. The participants' own living wills tended to express preferences similar to those in the sample (P = 0.0005).

The default manipulations in both experiments had potent but transient effects and influenced what participants wrote in their living wills but not their responses to later medical scenarios. Expression of end-of-life treatment preferences appears to be temporarily constructed from the decision-making context. These results have implications for surrogate decision making and the use of the living will as a tool to preserve patient autonomy.