MICROSIMULATION METHODS AND THEIR APPLICABILITY IN THE EVALUATION OF PERSONALIZED MEDICINE STRATEGIES

Monday, October 25, 2010
Sheraton Hall E/F (Sheraton Centre Toronto Hotel)
Beate Jahn, PhD1, Ursula Rochau, MD1, Nikolai Mühlberger, DVM, MPH1, Gaby Sroczynski, PhD, MPH1, Annette Conrads-Frank, PhD2 and Uwe Siebert, MD, MPH, MSc, SD1, (1)UMIT - University for Health Sciences, ONCOTYROL - Center for Personalized Cancer Medicine, Hall, Austria, (2)UMIT - University for Health Sciences, Medical Informatics and Technology, Hall i.T., Austria

Purpose: Decision analyses on personalized medicine strategies require modeling multiple constant or time-varying patient characteristics (e.g., genotypes, bloodmarkers, risk factors, disease stages). Our goal was to identify and contrast different microsimulation approaches using well known health policy models as examples (e.g., POHEM, UKPDS) and discuss the applicability of these approaches in the evaluation of personalized medicine.

Method: We performed a review on microsimulation and applications in social sciences, health care and politics. Assessment criteria include the modeling of patient characteristics/patient history/prior events, the way events or transitions between health states are modeled, the inclusion of life years/utilities/costs, open/closed cohort approach, and the way time is modeled.

Result: Identified approaches range from state-transition models, discrete-event simulation models to equation-based models. Individual characteristics relevant for personalized medicine include individual risk factors, clinical properties, patient history, severity of disease, number of repeated events. Different approaches were used to link risk factors and predictors to prognosis and treatment decisions and success. E.g., the UKPDS outcomes model estimates the long-term impact of health interventions of patients with type 2 diabetes based on United Kingdom Prospective Diabetes Study. The occurrence of events (e.g., ischaemic heart disease, myocardial infarction) and the changes in risk factors (e.g., HbA₁c, blood pressure) are based on a set of regression equations (e.g., logistic, Weibull, Gompertz). Time advances in constant 1-year cycles. POHEM is one of the leading comprehensive Canadian microsimulation models for health care policies. Applications range from lung cancer treatment, breast cancer prevention to the evaluation of cardiovascular diseases. POHEM is a continuous time, open cohort model, in which actors and their characteristics are modeled.

Conclusion: Microsimulation modeling techniques are broadly applied but still underrepresented in health sciences. Microsimulation is a powerful tool for evaluating strategies in personalized medicine, because it can be used to incorporate the genetic and clinical heterogeneity of individuals as well as personalized decision algorithms.