OBJECTIVES: Although the progression of non-alcoholic steatohepatitis (NASH) has been described in several health economic models, the relationship with the population burden of disease is not well understood. In seeking to estimate a US population-level burden of disease model for NASH we confronted several methodological challenges: 1) the lack of robust incidence data in the literature; 2) the indolent nature of the early disease stages, and 3) decomposition of the humanistic burden into morbidity and mortality impacts. Our aim is to describe our approach to meeting these challenges. Another aim is to promote the simultaneous use of spreadsheet (Excel™) and script (R) modelling to provide validation and verification while facilitating model sharing.

METHODS: The first step was building a NASH incident model starting from the development of early indolent disease (F0) through advanced symptomatic later stages (F4 and decompensated cirrhosis). We then applied a model for disease incidence based on an underlying rate impacted by risk factors such as obesity and diabetes, calibrating the rate to observed rates of later-stage disease. Recognizing the importance of age for both risk of disease incidence and competing risk of death we developed a presentational tool for displaying age pyramids of morbidity and mortality at the population level.

RESULTS: Our incident model of NASH is similar to the published disease models but was employed to give novel estimates of NASH incidence and population burden. Age/sex pyramids can show distributional impacts of disease burden on the population in terms of both morbidity and mortality. The model was also validated and verified using R modelling compared to Excel.

CONCLUSIONS: We present novel ways of addressing methodological challenges to estimating population level burden of disease estimates, applicable to other chronic diseases beyond NASH.