PRM130 - OPEN-SOURCE SOFTWARE FOR DEVELOPING COMPUTATIONALLY EFFICIENT COST-EFFECTIVENESS MODELS TO EVALUATE SEQUENTIAL TREATMENT STRATEGIES IN ONCOLOGY

Abstract

hesim is an R package for cost-effectiveness modeling that provides a general framework for integrating statistical analysis for parameter estimation with simulation techniques for economic evaluation. Cost-effectiveness models are built by specifying a model structure, which consists of a set of statistical models for disease progression, utility, and costs as a function of estimated parameters and input data describing the patient population and treatment strategies. The package currently supports N-state partitioned survival models (PSMs), but PSMs are limited because they do not account for structural relationships between survival endpoints. Our aim was to extend hesim for the economic evaluation of sequential treatment strategies in oncology. We developed software for simulating semi-Markov and inhomogeneous continuous time state transition models. Hazard functions for semi-Markov and inhomogeneous models are estimated using “clock-reset” and “clock-forward” multi-state models, respectively. State membership in semi-Markov models is simulated using a patient-level simulation and transition probabilities in inhomogeneous models are estimated using the Aalen-Johansen estimator. In patient-level simulations, time on each treatment can be simulated from time-to-event distributions. Survival functions for state transitions and treatment duration can be fit using parametric, spline, or fractional polynomial models. Separate survival models can be joined together (e.g., pre/post follow-up) to assess assumptions during the extrapolation period. The core code is written in C++ to facilitate probabilistic sensitivity analysis (PSA) and patient-level simulation. The software is illustrated with a cost-effectiveness analysis of sequential treatment strategies in oncology along with a new method for multi-state network meta-analysis with summary-level data. A new toolkit for developing state-transition models for cost-effectiveness modeling in oncology is available. State transition models with hesim are computationally efficient and allow users to model sequential treatment strategies, conduct scenario analyses for the extrapolation period, and conduct PSAs.