Introduction: Algorithms have been developed that estimate preferences from the Short Form 36 (SF-36). The objective of this study was to compare SF-36-derived preferences to directly elicited preferences.
Methods: The authors used data from a clinical trial of adult asthmatics to derive preferences from the SF36 and compare those to visual analog scale (VAS) values and the Asthma Quality-of-Life Questionnaire (AQLQ).
Results: The differences between VAS and derived preferences ranged from -0.066 to 0.024 at baseline and 0.006 to 0.107 at the end of follow-up. All measures improved from baseline (P < 0.001); however, derived preferences underestimated change (0.066 to 0.131) compared to the VAS (0.173) (P < 0.001), which could affect economic evaluations. Correlations between preferences and the AQLQ ranged from 0.56 to 0.70 at baseline (P < 0.001) and 0.53 to 0.70 for changes from baseline (P < 0.001).
Conclusions: The derivation methods produce valid and responsive measures of patient preference. However, derived preferences differ from one another and directly elicited VAS preferences.