OPTIMIZING SURVIVAL AND COST-UTILITY IN THE TREATMENT OF ACUTE AND CHRONIC GRAFT VERSUS HOST DISEASE: A CALIBRATED SYSTEM DYNAMICS MODELING APPLICATION

• Published in: Value in health 2017-05, Vol.20 (5), p.A213
• Main Authors: Zia, A, Jones, CA, Mesa, OA, Weimersheimer, P
• Format: Article
• Language: English
• Subjects: Clinical research; Clinical trials; Costs; Data processing; Graft-versus-host reaction; Liver; Lungs; Medical treatment; Patients; Plastic surgery; Quality of life; Severity; Skin; Survival; Utility functions
• ISSN: 1098-3015; 1524-4733
• DOI: 10.1016/j.jval.2017.05.005

OBJECTIVES: Outliers in graft vs host disease (GvHD) can sometimes presentdisproportionally extreme costs and quality of life measures, which may be ignored by regression-based estimates of cost-utility. Agent Based Models (ABMs) can capture these outlier effects, which can then be applied to patients whose characteristics match similarly. In this study, our goal was to build a calibrated ABM that could align treatments with heterogeneous patient characteristics, and compare cost-utility in a more granular fashion between extracorporeal photopheresis (ECP) offered to acute vs chronic GvHD patients. METHODS: Using a 20-week clinical trial panel dataset comprising 26 patients, we calibrated an ABM that simulates cost-utility for acute vs chronic GvHD patients before and after ECP. Patient characteristics such as organ involvement (skin, GI, liver, lungs, gut) and severity (II, III, IV for acute; moderate or severe for chronic) were assigned to each individual in the ABM. We scaled up the calibrated ABM to population-based estimates provided by the United Network for Organ Sharing (UNOS). RESULTS: Our preliminary ABM findings suggest that ECP cost-utility estimates are markedly lower in the case of chronic as compared with acute GvHD. We were able to illustrate outlier effects in both conditions. These outlier effects would not be captured in traditional regression and cost-utility analyses. Our findings highlight how patient-specific characteristics such as age, gender, and organ involvement matched to severity must be considered for maximizing survival and quality of life while minimizing cost. Simulations aross the wider UNOS population confirm this effect of outliers on survivability and quality of life. CONCLUSIONS: Using patient-level data, ABMs can illustrate survival and cost-utility in a manner that calibrates the right treatment for the right patient at the right time, such as the application of ECP to chronic vs acute GvHD. Such calibrated ABMs may also be scalable to population-based data.