The Assessment of Quality of Life (AQoL) II instrument overview and creation of the utility scoring algorithm

MAU instruments seek to measure the ‘utility’ of health states in a way suitable for use in economic evaluation studies and, in particular, cost utility analysis (CUA). The Assessment of Quality of Life, Mark 2 (AQoL 2) project was undertaken specifically to increase the sensitivity of measurement in the region of full health, where most other instruments, including the earlier AQoL 1 instrument are relatively insensitive. In sum, the AQoL 2 instrument estimates utility using a three stage procedure. Items are (i) weighted and combined using a multiplicative model to obtain dimension scores; (ii) these are similarly weighted and combined to obtain an initial AQoL score; (iii) this is then transformed econometrically to produce the final estimate of a health state utility. As with AQoL 1 the research program also sought to experiment with new methods for achieving this. AQoL 1 was the first instrument to use a multi level descriptive system with five dimensions of health separately modelled and then combined. After experimentation it incorporated a new way of modelling the utility of health states worse than death. AQoL 2 adopted this same multi level structure It was developed in 2 stages. The first used a series of confirmatory factor analysis using Lisrel, to construct dimension models. The second was a confirmatory factor (SEM) analysis of the overall AQoL which combined all of the dimensions. Utility scores were modelled in three stages. Time trade-off (TTO) importance weights were first combined into dimensions and to the dimensions into a single score using multiplicative (non stochastic) models (as with AQoL 1). However these were subsequently adjusted in a third stage econometric ‘correction’ based upon independently collected multi attribute – TTO – scores. Section 2 of the paper below summarises the work published in Working Paper 144 ‘Overview of the Assessment of Quality of Life (AQoL) Mark 2 Project (Richardson et al 2004b). This summarises the AQoL program, outlines the methods used to obtain the AQoL descriptive system and the methods used to obtain the stage 1 and stage 2 utility algorithms based upon this system. Section 3 presents the results from these two stages of the modelling. Section 4 of the paper outlines the methods data and results of the econometric analysis conducted to obtain a ‘stage 3’ adjustment or correction to improve the explanatory power of the model. It presents standard and new tests of the correction. One such task is the use of the corrected models to predict utility in an independently collected data set. Results using MA-TTO data collected for the VisQoL instrument are reported in Section 5. Model selection is discussed in Section 6. Computation of final utilities is complex. Algorithms for the dimension and final utilities are provided in Appendix 3 . Syntax for the STATA statistical software package to carry out these calculations has been placed on the CHE website: http://www.buseco.monash.edu.au/centres/che/