Payment Models to Coordinate Healthcare Providers with Partial Attribution of Outcome Costs

Problem definition : In many healthcare systems, general practitioners refer patients to specialists, who make treatment decisions under limited capacity. We evaluate the effectiveness of different payment schemes, both traditional ones, where the payer contracts separately with the providers, and bundled schemes, where the providers share a single bundled payment. A key feature considered is a performance-based payment component to coordinate the decisions of the general practitioner and specialist by a single payer. The providers are partially responsible for the patient outcomes and costs stemming from their treatment decisions. Academic/practical relevance : We propose and analyze a model to address how referral and specialist treatment decisions in healthcare impact each other and how payment schemes impact the coordination of care between the general practitioner and specialist. Our work is valuable to policymakers in understanding the trade-offs between using bundled and unbundled payment schemes in a referral-based healthcare system. Methodology : The underlying research method is the analysis of optimization models (with congestion effects embedded) to explore the effects of different payment schemes on different entities and various operational performance measures in a healthcare system. Results : A bundled scheme has higher referral rates, lower time spent with the patient by the specialist, higher specialist utilization relative to an unbundled system in most scenarios, and higher system costs. Our conclusions are robust to various model changes. Managerial implications : Our work provides specific managerial insights into the relative performance of bundled and unbundled payment systems on operational metrics in a referral-based healthcare system. It also sheds light on how the level of attributability of service outcomes to providers interacts with payment schemes to influence referral and specialist treatment decisions and impact various quality and cost metrics.