OBJECTIVES: To conceptualize, implement, and internally and externally validate the Evaluation Platform in COPD (EPIC): a ‘Whole Disease Model’ of chronic obstructive pulmonary disease (COPD) that addresses policy and clinical decisions at different levels of care across the entire clinical pathway of the disease.

METHODS: EPIC is a dynamic (open) population, Discrete Event Simulation model for COPD. COPD care pathways were conceptualized by a panel of clinical experts. Main features representing disease course are 1) prevalence of COPD based on an analysis of a large representative Canadian COPD cohort (CanCOLD), 2) individualized longitudinal trajectories of forced expiratory volume at one second (FEV1) from a dedicated analysis of the Lung Health Study (LHS), and 3) heterogeneous background exacerbation rate and severity from an analysis of a recent major COPD clinical trial (MACRO). COPD incidence equations were stochastically calibrated from the prevalence equations. To externally validate the model, we compared lung function trajectories, exacerbations, and mortality against four external cohorts (EUROSCOP, PanCan, TORCH, and UPLIFT).

RESULTS: FEV1 trajectories showed robust internal and external validity: with 95% prediction intervals having actual coverage probabilities of 96%, 91%, and 90% in LHS, EUROSCOP, and PanCan, respectively. Simulated values for total exacerbation (1.34/PY, 1.30/PY ), severe exacerbation (0.27/PY, 0.28/PY) and mortality rates ( 10%, 12.4%) were consistent with reported values in TORCH and UPLIFT respectively, considering uncertainty intervals around the mean.

CONCLUSIONS: EPIC is a validated microsimulation model of COPD informed from multiple large clinical data. As a Whole Disease Model, it is capable of modeling the health and economic outcomes of many decisions in their interaction. By using an open-population, it can model realistic scenarios such as gradual market penetration and sub-optimal adherence. By considering disease heterogeneity, EPIC can be used to answer questions on efficiency and clinical utility of “personalized medicine” interventions such as biomarker implementation.