Estimating quality adjusted life years in the absence of standard utility values – a dynamic joint modeling approach

Abstract

Estimation of Quality Adjusted Life Years (QALYs) is pivotal toward cost-effectiveness analysis (CEA) of medical interventions. The popular multi-state decision analytic modeling approach to CEA uses standard utility values assigned to each disease state to estimate QALY. In this paper, we have formulated a new approach to estimate QALY by defining utility as a function of a longitudinal covariate significantly associated with disease progression. Association parameter between the longitudinal covariate and survival times has been estimated through joint modeling of the longitudinal and the Weibull accelerated failure time survival model. MCMC techniques have been used to predict expected survival times of each censored case using the fitted model. Time-dependent utility values, calculated using projected values of the longitudinal covariate, have been used to evaluate QALYs for each patient. Proposed methodology has been demonstrated on a retrospective survival data of HIV/AIDS patients. A simulation exercise has also been carried out to gauge the predictive capability of the joint model in projecting the values of the longitudinal covariate. Results show that the proposed dynamic approach to estimate QALY can be a promising alternative to the popular multi-state decision analytic modeling approach, especially when the standard utility values are not available.