Modeling therapy sequence for advanced cancer: A microsimulation approach leveraging Electronic Health Record data

Abstract

Many patients with advanced cancers undergo multiple lines of treatment. We develop methods for estimating quality-adjusted outcomes and cost-effectiveness of therapy sequences, informed by patient-level longitudinal data from Electronic Health Records (EHRs). We develop microsimulation models with a discrete-time health-state transition framework and propose two methods: one using multi-state models to estimate transition probabilities, and one using observed patient trajectories through the health states. We use bootstrap resampling to estimate standard errors. We create synthetic EHR-like datasets to evaluate these methods where within-patient transition times depend on covariates and a copula generator, and compare with Markov cohort models. We demonstrate these methods in two treatment sequences for advanced bladder cancer (cisplatin or carboplatin-based therapy followed by immunotherapy), incorporating external information on costs, utilities, and expected adverse event. Both methods produced well-calibrated overall survivals, although the trajectory approach was often superior. The multi-state model approach generated lower standard errors but was biased when compared to known results from the synthetic datasets. The observed trajectory approach mostly produced confidence intervals that covered known values. In the bladder cancer example, both methods result in a Net Monetary Benefit (NMB)>0 for the cisplatin-based treatment sequence with a willingness to pay of $100,000 per quality-adjusted life year. Both microsimulation methods produce well-calibrated results and offer superior performance to a homogeneous Markov cohort approach when studying therapy sequence. Where available, patient level EHR-based data should be considered to inform cost-effectiveness models.