QuIP
Item Response Theory (IRT) has driven major improvements in descriptive health-related quality of life measurement. IRT allows many items about a health domain (such as pain or physical functioning) to be calibrated on a single latent construct. Any combination of the calibrated items can then be used to estimate a score on the construct. This means that different items can be used but the final scores are comparable.
The development of IRT based measures provided a unique opportunity to improve preference-based summary scores of health as well. Previously constructed preference-based summary scores have always used limited discrete descriptive spaces which resulted in a variety of problems. Being able to directly link preference scoring to the latent construct could overcome these problems.
We developed and tested four different methods to link preference scores to the latent construct. Our results were presented at ISOQOL’s 23rd Annual meeting. A copy of the presentation is here and a manuscript is forthcoming.