Purpose: Researchers and clinicians acknowledge the complexity of planning for future medical treatment desired in the event of incapacitation. Unfortunately, many attempts to evaluate the quality of such difficult planning have been stymied by the lack of measures that can be shown to have validity. This study examines the psychometric characteristics of the Decisional Conflict Scale (DCS) when used as a measure of patients' evaluation of their end-of-life decision-making process.

Method: This evaluation used a sample of 59 outpatients with a life-threatening illness and their surrogate decision makers, who were randomly assigned to receive, a decision-aid intervention, the Patient-Centered Advance Care Planning, or usual care only. This intervention was designed to help patients make informed end-of-life decisions that are consistent with their personal values and beliefs and to improve surrogate's understanding of the patient's preferences for end-of-life care. The intervention was a scheduled 30-45 min. interview that was delivered by a trained nurse facilitator. Patients completed the DCS and the Quality of Patient-Clinician Communication about End-of-Life Care shortly after the intervention. Convergent and construct validity, Discriminant validity, and internal consistency were examined using the Spearman product moment correlation, two-sample t-test, Ridit analysis, and Cronbach's coefficient alpha and item-to-total correlations. For the validation of discriminant ability, a known-groups approach was used.

Results: The DCS demonstrated convergent, construct, and discriminant validity based on the total scale scores. The comparisons at the subscale score level between the intervention and control groups showed a lack of discriminating ability for the uncertainty subscale. The internal consistency reliability for the total score of the DCS was reasonably good in this sample (á = 0.81). All three uncertainty subscale items showed the weakest item-to-total correlation (0.22 < r < 0.33). When the uncertainty subscale items were eliminated, Cronbach's alpha coefficient improved to 0.84.

Conclusions: The DCS appears to be a viable research instrument for measuring the quality of end-of-life decision making. The DCS provides good discrimination between groups and has proved reliability in the end-of-life decision-making context, especially with respect to modifiable factors contributing to uncertainty and the effectiveness of the decision-making process and the quality of decisions. However, the uncertainty inherent in such decision making may limit the applicability of the uncertainty subscale in such context.