A Lung Cancer Policy Model

Principal Investigator: G. Scott Gazelle
Institution: Massachusetts General Hospital
Grant Number: 5R01CA097337-02

Abstract: The goal of the proposed research is to develop a lung cancer model capable of guiding screening policy decisions over the next several years, before results of ongoing clinical trials are available. Through Monte Carlo simulations of large numbers of individuals in selected target populations, the model will provide estimates of the health and economic consequences of proposed screening technologies such as helical CT and chest X-ray imaging. Much of the model's value will lie in translating short-term trial results into lifetime health and economic consequences. Data from a diverse array of sources, including past and ongoing screening trials, will be integrated to derive robust parameter estimates. An underlying natural history component will simulate the generation and growth of cancers and benign lesions, with parameters derived from existing screening trial data, autopsy studies, and information on tumor growth rates. Each individual's probability of developing lesions will be determined by risk factors such as smoking history, age, and sex. Additional model components will simulate the detection and treatment of lung cancer, with outcomes calibrated to existing data from prevalence screens and tumor registries. By modeling tumor growth and detection explicitly, our approach differs from previous decision analyses of lung cancer screening. Principal goals of the research are to determine the cost-effectiveness of proposed and hypothetical screening strategies, to examine the economic and health effects of screening different target populations or of screening with improved technologies, and to make comparisons not included in ongoing trials (e.g., helical CT screening vs. no screening, or screening at different frequencies). The model will also allow identification of threshold performance characteristics, above which new screening programs would be attractive alternatives to no screening or to screening with helical CT. By identifying parameters with both large plausible ranges and important effects on results, the model can serve as a tool for use in the efficient design of future clinical trials. The model will guide imminent screening policy decisions, and, paralleling achievements of the coronary heart disease policy model, could one day evaluate treatment advances or incidence trends.