PNS17 Adjusting General Population Utility: Preliminary Results from an Update of the Published UK Regression MODEL

Abstract

A published regression model by Ara and Brazier (2010) is widely adapted to estimate the age- and gender-adjusted utility in the UK (general population utility = 0.9508566 + 0.0212126*male - 0.0002587*age - 0.0000332*ageˆ2). However, the published model did not provide the information to examine the coefficient uncertainty. Our study updated the existing regression model with additional data from Health Survey for England and reported the summary statistics and covariance matrix to test the coefficient uncertainty. As an update, this study followed a similar approach to the original research. The Health Survey for England is an annual survey conducted in a sample of the general population in England. The original research included the 2003 and 2006 surveys, while we included all HSE surveys collected EQ-5D (HSE 2003-2006, 2008, 2010-2012, 2014 and 2017) Data were accessed via UK Data Service. The analysis was conducted in STATA version 16. The intercept from the new model is 0.9545565 (±SE 0.0048217). Coefficients for male, age and ageˆ2 are 0.025311 (±SE 0.0014438), -0.0010594 (±SE 0.0002036) and -0.0000232 (±SE 0.00000199), respectively. Compared to the original regression model by Ara and Brazier (2010), the utility derived from our model is similar. For example, for a 40-year old female, the estimated baseline utility is 0.8874 and 0.8751 based on the original and the new regression models, respectively. Our study provides an updated regression model (general population utility = 0.9545565 + 0.025311 *male - 0.0010594 *age - 0.0000232 *ageˆ2) using all available EQ-5D data collected in the HSE surveys. Summary statistics are reported to enable testing the uncertainty of coefficients in a probabilistic sensitivity analysis. The new regression model could potentially allow much wider adaptation of the age and gender adjusted utility in economic evaluations.