Duke 2008 | ASHE | Equity and Efficiency in Health and Healthcare

A large literature documents disparities in hospital care between privately-insured patients and underinsured (i.e., uninsured and Medicaid) patients. However, there remains a largely unanswered question; how much of the disparities are due to differences in patient characteristics and how much are due to hospital responses to differences in profitability? In addition, previous studies have concentrated on the effects of hospital payment rates on treatment intensity and outcomes, and few studies have examined the possibility that hospitals selectively admit patients based on profitability. This study exploits a plausibly exogenous policy change to examine whether hospitals change the admission practice depending on level of reimbursement. In the 1980s, the state of New Jersey introduced the rate setting system under which hospitals receive the same amount of payment for cases in the same DRG regardless of what type of health insurance the patient has or whether the patient is insured at all. In the early 1990s, however, the system was abolished and the state subsidy for charity care was drastically reduced. This policy change created an extremely large reduction in hospital profit margins for uninsured and Medicaid patients, without causing noticeable changes in uninsurance rate or in Medicaid coverage rate. Using hospital discharge records from the two states, I compare hospital admission practice in New Jersey and New York before and after the reform in New Jersey based on a fixed-effects model. I find that hospitals responded to price reduction by selectively reducing inpatient admission of underinsured patients in relatively non-emergency conditions. This result contrasts with previous studies finding that Medicaid physician fees have little effect on physician caseload of Medicaid beneficiaries. At the same time, the volume of underinsured patients in relatively emergency conditions did not change, probably because federal law prohibits hospitals from turning away emergency patients in unstable conditions. Then, I test the possibility that the reduction in hospitalization of underinsured patients in stable or non-emergency conditions could lead to an increase in the hospital cost of treating underinsured patients in unstable emergency conditions. If poorly-treated, stable or non-emergency conditions could develop into unstable emergency conditions. Obliging hospitals to treat underinsured emergency patients in unstable conditions could induce hospitals to treat these patients before they become unstable, provided that early treatment reduces the cost of hospital care. Nevertheless, a free-rider problem reduces such an incentive when underinsured patients have choice of hospital. I find a greater increase in the (severity-unadjusted) hospital costs of treating a Medicaid patient suffering from pneumonia in post-reform New Jersey than in New York, although there are no significant changes in (severity-unadjusted) in-hospital mortality from pneumonia. New Jersey hospitals might have responded to the large decline in reimbursement rates for uninsured and Medicaid patients by reducing treatment intensity, which would decrease the hospital cost per discharge. My results indicate that this effect is dominated by the cost increase arising from the increase in patient severity.