Uncertainty Analysis in Population-Based Disease Simulation Models: A Practical Framework

• Activity Number: 458
• Type: Invited
• Date/Time: Wednesday, August 7, 2013 : 8:30 AM to 10:20 AM
• Sponsor: Health Policy Statistics Section
• Author(s): Behnam Sharif*+
• Companies: University of British Columbia
• Keywords: Microsimulation ; Computer Simulation ; Disease Modelling ; Uncertainty Analysis ; Parameter Uncertainty

Abstract

Even though Uncertainty Analysis (UA) is essential as a part of the validation of population-based disease simulation models, literature is imprecise on how it should be performed. In this study, we provided a practical framework for UA for these models and performed an UA in POHEM-OA, a microsimulation model of Osteoarthritis (OA) for Canadian population. METHODS: We examined guidelines of UA methodologies in simulation models of decision analysis and environmental models. We have also developed a mathematical model for calculating the components involved in UA. Using Latin-hypercube sampling, we have sampled from a multivariate lognormal distribution the values for sex-specific hazard ratios of OA by BMI categories. RESULTS: We included two sources for UA: the Monte Carlo error and the parameters uncertainty . For one day run of a 12G memory, CPU=i7-980 Intel, 3.3 GHz, we have calculated m*=1,000,000 and n*=500 based on the developed algorithm. The sex-specific prevalence of OA in Canada predicted by POHEM including the 95% confidence intervals is produced; in particular, in POHEM-OA, the uncertainty of sex-specific prevalence of OA has been shown to be increasing over time.