Designing microsimulation models for policy impact

Microsimulation models are flexible and powerful tools with the capacity to provide important evidence to policymakers and improve the lives of families. In this presentation we will discuss the application of microsimulation to genomic medicine and related policies. In particular, we will address how we lay the groundwork for a microsimulation model of genomic medicine when there is almost no data, but the conditions have such significant impacts on quality of life and mortality that it is critical that evidence is robust. Until recently, patients with genetic disorders rarely received a definitive ‘molecular’ diagnosis. Further, what we consider to be a single condition such as genetic blindness, may be caused by many different genetic variants (each with very different symptoms, age of onset and disease trajectory), and for some conditions such as intellectual disability, the condition may be caused by thousands of different genetic variants. As a result of their unique genetic causal mechanisms, these genetic conditions each require a ‘targeted therapy’, such as corrective diets, enzyme replacement, and emerging new gene therapies, meaning that there may be no two patients in a study with the same condition or treatment. Such extreme heterogeneity makes the field of genetics and genomic medicine an ideal application for microsimulation modelling because simulations are run at the individual level. However, the rarity of many genetic conditions, combined with the low likelihood of receiving a molecular diagnosis, has meant that nationally representative survey data generally has not identified these conditions despite their high health and cost impacts for patients, carers and government, leaving a significant data gap for model development. In this presentation, we will describe how we filled substantial data gaps related to genomic medicine by designing new and comprehensive primary data collections to use as microsimulation model base populations, how we worked with our clinical colleagues to gain ethical approval to obtain primary patient data, how we collected the data to ensure a high response rate, and how the data was managed to ensure clean and robust information. We describe the development and structure of our static microsimulation models applied to genomic medicine and present some examples of analysis of model outputs. Finally, we will provide examples of how our models have influenced policy and some current policy applications in development as well as describing the impacts such policy changes have on families.