Optimization

Dramatic changes in the economics of the health care industry were putting the mission of Houston-based Texas Children's Hospital at risk. Private insurers and public health care financing programs such as Medicaid, squeezed the country’s largest pediatric health care facility between rising costs of medical services and equipment, and resistance to absorbing those higher costs. In addition, new reimbursement contract structures, including “diagnosis-related group” (DRG) and fixed-fee-per-patient arrangements, were shifting huge financial risks to Texas Children's, demanding a far more sophisticated approach to contract negotiation.

With a mission of saving lives and training physicians, Texas Children's Hospital (TCH) focus was not on profit maximization unlike an airline, hotel or car rental company. Similar to those struggling companies in those industries, TCH took advantage of PROS revenue management analytical support, tailored to its unique financial circumstances. A large proportion of the hospital's revenue is base on contracts not individual payments, the cost and revenue implications of complex contracts intimate knowledge is required prior to acceptance. Organizational research based optimization models can identify globally optimal solutions that satisfy pre-set goals over multiple forms of contracts, they can provide an advantage to the user of the optimization models that may not be clearly visible to the party on the other side of the negotiation.

TCH and PROS collaborated in the development of a customized solution, known as the Hospital Optimization System (HOS), to monitor contract performance and support contract negotiations. HOS, relying on Bayesian forecasting and nonlinear optimization models, uses data from prior patient encounters, including charges and resource usage, and hospital capacity information, to propose optimized contract terms for use in negotiations. The system also forecasts demand for hospital services and provides decision-support capabilities including “what-if” functionality.

Contracts renegotiated by Texas Children's with the benefit of the HOS were projected, over their first year, to result in a profitable revenue increase up to $17 million. HOS is also expected to enable managers to improve the effectiveness of planning in facilities utilization, staffing and budgeting.

In the United States, health-care providers are paid for patient services almost entirely by private commercial insure and their governmental equivalents, Medicare and Medicaid. Patients meet the requirements of a governmental program, directly or indirectly purchase coverage from an insurer, or remain uninsured. When patients buy insurance coverage, they usually do so through a sponsor, such as an employer or professional organization. Sponsors can purchase access to health-care resources outlined in an evidence of coverage, benefit plan from an insurer, or negotiate flat-rate contracts with insurers who administer all operational aspects of the payor’s health-care coverage plan. Insures in turn negotiate contracted terms of payment with health-care providers for the patient groups they represent. In theses managed вЂ"care contractual relationships, sponsors such as employers, negotiate contracts with insurers to obtain medical coverage fro the individuals they represent. Insurers negotiate contracts with health-care service providers, such as hospitals, to establish terms of payment for services rendered for covered individuals. In addition to insurers, payor’s include government programs, such as Medicare ad Medicaid, that provide coverage for the elderly and disabled, and low-income groups. Payment flows directly or indirectly from individuals to sponsors to insurers to providers.

Risk traditionally was an area of focus for the insurers. Insurers would charge a premium for what they determined as a higher risk. This was done using a many different techniques that usually included underwriting review to estimate the probability of an individual’s use of services or a review of a similar person or population’s historical usage of services. Funded by state and federal government, Medicare and Medicaid function differently as they have a shared budget spread over individuals. In the past, insurers based the primary method of reimbursement to health-care providers on a list of charges generated by hospitals on a very granular level. This is referred to as charge-based reimbursement methods and still exists today. A more common form of reimbursement, discount-off-charges, is where insurers negotiate a percentage reduction of the standard rate. The primary risk is from the calculation of cost in general and the allocation of fixed cost in particular. Throughout the 1990s insurers looked for ways to streamline contract reimbursement rates. Those methods include Per-Diem reimbursement where hospitals receive a unit reimbursement rate for an agreed set of resources; Diagnosis-Related Group Reimbursement (DRG) where ailments are group into diagnosis-related groups for use in clinical studies; and capitated reimbursement structures, payors give hospitals a fixed periodic fee for the people they insure or for a specific population.

Complex, risky reimbursement models and continuing cuts by government and commercial payors caused financial crises for TCH. In 1997, the Texas Medicaid program was moving to a managed-care model similar to that of Medicare that would place a greater burden of risk on the hospital. Medicaid accounted for a significant number of admissions at TCH. This risk was compounded by its struggle with the terms of per-diem, DRG, and capitation contracts with various private insures.

Forecasting is difficult in the health-care industry. This occurs when contract revenue depends on patient encounters and interactions with a patient using hospital resources that generate an itemized list of billable charges. To generate a more acceptable forecast, PROS and TCH needed entities consisting of generic encounters with fairly consistent resource usage. After defining the forecast entities, forecast of the fixed cost were done, variable cost, demand, and charges by day of the week for each entity. In addition, because forecast entities never use resources consistently, simple forecast of critical resources, such as operating room time or room