The Medical Hospice Benefit: The Effectiveness of Price Incentives in Health Care Policy Written By Vivian Hamilton, McGill University RAND Journal of Economics 24(1993) Presented By Jing Zhou 1. Introduction • 1. This paper mainly examines the effectiveness of per- diem reimbursement within the context of the Medicare Hospice Benefit program. • 2. Specifically, this paper gives answers to the following three questions: • 1. Does the rate of reimbursement influence the individual hospice’s certification decision? • 2. Does the rate of reimbursement influence the access to hospice care? • 3. How to explain the the variations in certification rates across different regions in US? 2. A Description of Hospices and the Medicare Hospice Benefit • 1. Hospices care for terminally ill persons who have exhausted their efforts to cure the disease that afflicts them • 2. The Medicare Hospice Benefit lays out the type of care hospices are expected to provide(HCFA, 1988) • 3. A hospice that becomes under the Medicare Hospice Benefit (certified) must provide the expected care to its patients. • 4. Then the Health Care Financing Administration(HCFA) reimburses hospices for their services on a per-diem basis. • 5. So the reimbursement rate plays a significant role in determining whether or not a hospice seeks medicare certification 3. The Model(1) • 1. The basic model in this paper is a model of the hospice’s certification decision process. • 2. Since 95 percent of hospices are nonprofit organizations, the primary goal of a hospice is not likely to be profit maximization. Instead, utility maximization is its goal. • 3. So the hospice chooses to be certified if doing so leads to an increase in utility. 3. The Model(2) • 1. In this model, We should consider the factors that will affect a hospice’s utility before and after it is certified • 2. The number of patients is an important index of a hospice’s utility. Why? • 3. Providing services on a large scale is assumed to lead to the enhancement of the prestige of the institution and the likelihood of its ability to continue operating within the community. • 4. High quantity may also attract donations. 3. The Model(3) • 1. The other important factor influencing a hospice’s utility is its costs and revenues from operation. Why? • 2. Excess revenues can be used to satisfy other possible utility-increasing objectives, such as improving facilities or increasing salaries. • 3. So the model of the hospice’s certification decision process should take into consideration the quantity and the revenues and costs changes before and after a hospice is certified. 3. The Model(4) • 1. The basic model in Hamilton’s paper is: I * Z (log P log P ) i i ic in i • 2. I i*is a continuous, unobservable index. When it is greater I than 0, this shows the hospice is certified.Zi is a vector of explanatory variables that affect the hospice’s expected costs and revenue before and after the certification • 3. log Pic and log Pinare the expected value of the logarithm of the number of patients a hospice will serve if it is certified and noncertified. 3. The Model(5) • 1.We also have two patient equations for each hospice: • log Pic X i If Certified (1) 1 i1 • log Pin X i If Noncertified (2) 2 i2 • Where X i is a vector of market-specific variables that characterizes the demand for the hospice’s services. • 2. Taking expectation of equations (1) and (2), we have: E(log Pic X i ) X i 1 E (log Pin X i ) X i 2 • 3.Taking these two Conditions into our original model, we have I * Z ( X X ) • i i i 1 i 2 i (3) • 4. Estimation of Equation (1) through (3) will determine which factors are significant in explaining a hospice’s certification decision • 3. The Model(6) • 1. The explanatory variables in the vector include four kinds of variable: • 1. Reimbursement rate: Hospice Rate and Home Health Rate • 2. Case Mix variables: % Female, % Noncancer and Length of Stay • 3. Organizational structure variables: Hospital Dummy, Home Health Dummy, % Volunteers, Voluntary Nonprofit and Years of Operation • 4. Labor Cost Variables: Hospital Cost Index and Home Health Cost Index • 2. The explanatory variables used to estimate the patient equations include region-specific demographic variables, County-specific health care market variables and some organizational structure variables. 4. Estimation • 1. This system of Equations is a switching regression model with endogeneous switching(Maddala, 1983) • 2. Using OLS to get the parameters in (1) and (2), then using the derived parameters to run a probit estimation of equation(3) will yield inconsistent estimates • 3. Using the maximum likelihood method to estimate this system of equations is a more appropriate approach. • 4. Most data are from National Hospice Organization(NHO)’s annual hospice census(1987). The county-level data are from the Department of Health and Human Services’ 1989 Area Resource File(ARF) 5. Estimation Results • Results from the certification-decision equation (3) • 1. Holding all other factors constant, a $1.00 increase in the hospice reimbursement rate increases the probability of certification by 1.7%. This shows hospices do indeed respond to price incentives. • 2. A percentage increase in the number of patients a hospice can expect has a positive impact on its probability of certification. • 3. Some other variables, such as a higher share of female patients and years of operation also have positive influence on a hospice’s probability of certification Estimation Results(2) • Results from the patients equation(1) and (2) • 1.A higher percentage of specialists has a positive influence on certified hospice size. • 2. The total number of hospices in the county has a negative impact on the number of patients a certified hospice serves. • 3. These market variables influence’s size differently when it is certified as opposed to noncertified. Generally, hospice can expect a significant increase in size with certification. Estimation Results(3) • 1.Using a probit equation, this paper also shows the variation in the Medicare reimbursement rates across regions play a significant role in explaining the observed differences in certification across geographic divisions and between urban and rural areas. • 2. The underlying reason for this variation is that the wage indices used to adjust the Hospice Benefit rates don’t correctly account for variations in cost across regions. • 3. Even using two other alternative wage indicies, the Hospital Cost Index and the Home Health Cost Index, we can barely alter the observed differential. 6. Conclusion • 1. The hospices are indeed responsive to change in reimbursement rate. • 2. The fixed-price reimbursement mechanism can be an effective instrument in encouraging the provision of more high-quality hospices care for Medicare enrollees. • 3. However, this mechanism increased access comes at a price: the higher reimbursement rate may reduce the cost-effectiveness to the Medicare program of hospice care relative to conventional care. • 4. In addition, reimbursement rare must be set carefully in order to narrow the variations in certification rates across different regions. Comments • 1. This paper uses cross-sectional data analysis method. This mainly is due to the limits of data in 1993. In today, maybe we can also use time series analysis to do the test work. • 2. This paper does not consider the influence of private insurance policies because at that time, such insurance did not pay much for hospice care. However, now many private insurers also have added coverage for hospice care. So the model here may needed modification.