Cost and resource allocation in healthcare delivery systems

Thesis: Ph. D., Massachusetts Institute of Technology, Sloan School of Management, 2015.Cataloged from PDF version of thesis.Includes bibliographical references (pages 145-152).This thesis studies contemporary challenges arising at the market, system, and organization levels in the healthcare industry, and develops novel frameworks that allow us to better understand cost and resource allocation for strategic decision making in healthcare settings. The U.S. healthcare industry is going through a massive transformation process due to the increasing industry consolidation, and the implementation of the recently enacted healthcare reform. These changes have completely transformed the incentives in the industry, and traditional practices have become outdated, or are, in general, inadequate to address the new challenges. Our frameworks combine real data and statistical analysis with novel optimization-driven approaches (e.g., linear programming, game theory) that capture the first order aspects of the dynamics of the corresponding markets, systems, and organizations. Overall, this work has relied on collaboration with industry partners in order to identify trade-offs, validate models, and pursue practical innovation and implementation of the proposed frameworks. In the first part, and motivated by real applications from the healthcare industry, we consider a setting, where one firm provides a service to a second firm that is facing stochastic demand for the service. The changes in the reimbursement system have created new opportunities for business-to-business interactions between healthcare systems and providers. Typical contracts in the healthcare industry are based on a transaction fee per unit of service that is negotiated between the two parties. Unlike traditional product-based 2-echelon supply chains, the two firms have opposing risks with respect to the demand volume. We leverage this insight to design a conceptually simple two-price volume based contract, and analyze it within a game theoretic setting. We show that a two-price contract can optimally ensure risk sharing. Moreover, although the resulting problem is non-convex, we are able to characterize the unique equilibrium contract in closed form for a family of utility functions that captures firms' different risk behaviors, and general demand distributions. Moreover, at equilibrium the new contract has two desirable properties: (1) it allows for better risk reduction (measured by CVaR) for the two firms, and (2) it reduces the uncertainty of the payment transaction. In the second part, we study the strategic cost and resource allocation in large healthcare delivery networks and how these networks can, efficiently, integrate their operations in order to attain network's welfare objectives. Strategic problems, such as resource allocation, capacity placement, and portfolio of services in multi-site networks, require the correct modeling of network costs, network's welfare objectives and trade-offs, and operational constraints. Traditional practices related to cost accounting, specifically, the allocation of overhead and labor cost to individual activities, as a way to account for the consumption of resources, are not suitable for addressing these challenges. These practices often confound resource allocation and network building capacity decisions. In this part, we develop a general methodological optimization-driven framework inspired by network revenue management models, specifically linear programming optimization, that allows us to better understand network costs and provide strategic solutions to the aforementioned problems. We report the application of this framework on a real case study to demonstrate its applicability and important insights derived from it. Finally, in the third part of this thesis, we study the nature and sources of variability in surgical activities in a large pediatric hospital. We use machine learning techniques to quantitatively show that surgery time variability is high among pediatric cases and, against common belief, this is poorly explained by surgeon effect or other commonly considered characteristics. Our studies suggest that pediatric surgery time has higher inherent variability making pediatric ORs necessarily more costly and harder to schedule than adult ORs. They must therefore be sourced accordingly. These findings are novel and will be useful in the management of busy pediatric operating theaters. For administrators and policymakers, it provides a basis for understanding some of the added costs inherent in caring for children.by Maria Fernanda Bravo Plaza.Ph. D