Physician payments and infant mortality: evidence from Medicaid fee policy.

While efforts to improve the health of the uninsured have focused on demand side policies such as increasing insurance coverage, supply side changes may be equally important Yet there is little direct evidence on the effect of policies designed to increase the supply of Medicaid services to the poor. We provide such evidence by examining the relationship between infant mortality and the ratio of Medicaid fees to private fees for obstetriciaiilgynecologists. We build a state and year specific index of the fee ratio for 1979-1992, a period of substantial variation in relative Medicaid fees. We find that increases in fee ratios are associated with significant declines in the infant mortality rate. We also find that higher fees raise payments made to physicians and clinics under the Medicaid program, but reduce payments to hospitals. Finally, we compare the cost effectiveness of reducing infant mortality by increasing fee ratios to the efficacy of reducing mortality by expanding the Medicaid eligibility of pregnant women. Although our results are sensitive to the time period used, we conclude that raising fee ratios is at least as cost effective as increasing eligibility. Janet Currie Jonathan Gruber Department of Economics Department of Economics, E52-274c UCLA MIT 405 Hilgard Avenue 50 Memorial Drive Los Angeles, CA 90024 Cambridge, MA 02139 and NBER and NBER Michael Fischer Yale Medical School Yale University New Haven, CT 06520 At 10 infant deaths per 1000 births, the U.S. ranks below 20 other industrialized nations in terms of the infant mortality rate (U.S. House of Representatives, 1992). Congress has responded to this situation by dramatically increasing the number of pregnant women eligible for health insurance coverage under the Medicaid program. As a result, the number of women 15 to 44 years old who were eligible for Medicaid coverage in the event of pregnancy grew by 140% between 1979 and 1990 (Currie and Gruber, 1994). However, even if these changes were effective in increasing the demand for prenatal, obstetrical, and neonatal care among the poor, they do not in themselves guarantee access to medical care. Many observers have alleged that there is a shortfall in the supply of physicians, and especially of obstetrician/gynecologists (ob/gyns), willing to serve Medicaid patients. Mitchell and Schurman (1984) report that in 1977 and 1978, 23.5% of primary care physicians, and 35.6% of obstetricians, saw no Medicaid patients. And in a recent survey by the Physician Payment Review Commission, 43 state Medicaid directors identified low physician participation rates as a problem (PPRC, 1991). These figures suggest that the increased demand for services generated by expansions of the Medicaid program could go unmet.' An alternative approach to improving birth outcomes is to focus on the supply side of the market. One natural supply side tool is Medicaid fee policy. The low fees paid by state Medicaid programs represent a major potential deterrent to physician willingness to see Medicaid patients. Holahan (1991) reports that the ratio of Medicaid fees to private fees was approximately 0.5 for most procedures surveyed, and 0.56 for total obstetrical care with vaginal delivery. And in the PPRC survey cited above, 38 states identified low fees as the major cause of low physician participation 'It is even possible that increases in demand for services among the near-poor could "crowd out" (,potentially) needier poorer women. In this case, increasing eligibility could worsen birth outcomes. rates. A large body of research suggests that increasing the ratio of Medicaid fees relative to private sector fees will increase physician participation in the Medicaid program (Hadley, 1979; Sloan, Mitchell, and Cromwell, 1978; Held and Holahan, 1985; Mitchell, 1991). Furthermore, Mitchell and Schurman (1984) and Adams (1992) find that the participation of ob/gyns is especially responsive to fee increases. But increasing the number of physicians willing to serve Medicaid patients will not necessarily lead to improvements in infant mortality, for two reasons. First, many poor women and children already receive care from clinics and emergency rooms.2 Decker (1993) and Long et aL (1985) find that increased Medicaid fees caused patients to shift from other sources of care to physician's offices but did not increase the number of ambulatory visits they received. This kind of change will have no effect on health outcomes unless the quality of care is significantly higher in physician's offices. Second, Fossett and Peterson (1989) and Fossett et a!. (1992) argue that fee policy is likely to have small effects due to the segregation of the poor into areas that are underserved by physicians. For example, Fossett et a!. (1992) compared Chicago neighborhoods with 50% of the population on welfare to neighborhoods with 10% of the population on welfare and found that there were twice as many physicians practicing in the wealthier areas (on a per child basis). They conclude that fee increases will not increase the supply of physician services available to poor families unless they are large enough to induce physicians to move into underserved markets. Thus, the sensitivity of physician participation to fee policy uncovered by the earlier literature does not prove that fee policy can be effective in improving health outcomes. 2 For example, Aday and Anderson (1984) report that relative to those with private insurance, Medicaid enrollees were 33% less likely to have a regular physician and 2.4 times more likely to have a source of regular care other than a physician's office. 2 In this paper we move beyond the earlier physician participation literature by directly examining the effect of physician fees on a particular health outcome: infant mortality.3 Using statelevel data we model infant mortality over the 1980-1992 period as a function of the ratio of Medicaid ob/gyn fees to private sector fees. Our estimates exploit the substantial variation in this fee ratio across states and over time. We also examine the relationship between the fee ratio and Medicaid expenditures, in order to determine whether changes in Medicaid fees are a cost effective way to reduce infant mortality. Finally, we use the measure of Medicaid eligibility developed in Currie and Gruber (1994) to contrast the cost efficiency of reducing infant mortality by changing demand-side or supply-side policies. We find that increases in the Medicaid fee ratio are associated with small, but significant, declines in the infant mortality rate. Our central estimates suggest that raising the fee ratio by 10 percentage points lowers infant mortality by 0.5-0.9%. This result is robust to variation in the sample period and to the measure of the fee ratio. We also find that higher fees raise payments to physicians and clinics under the Medicaid program, but that these increases appear to be fully offset by reduced payments toospis, so that lowering infant mortality through fee policy appears to be a 'free lunch'. The finding for hospital costs, however, is fairly sensitive to the specification chosen. As in Currie and Gruber (1994), we fmd that expansions of Medicaid eligibility also reduce infant mortality, but that they increase payments to both physicians and hospitals. Hence, our results suggest that raising Medicaid fees may be more cost effective, and certainly no less cost effective, than expanding eligibility as a means of reducing infant mortality. 3We are aware of no other study which has examined the effect of fee policy on health outcomes. There is a sizeable literature which assess the effects of provider supply, and in particular clinic supply, on birth outcomes; see Grossman and Jacobitz (1981) and Frank (1991) for examples. But this literature does not speak to the effects of changing physician reimbursement, especially if there is important site shifting between clinics and doctors' offices. 3 The rest of the paper is laid out as follows. In Part I we discuss the ways in which access to medical care might be effective in reducing infant mortality. In Part II, we describe our methodology and data sources with special attention to the construction of our measure of the fee ratio. Part III presents our estimates, and Part IV concludes with a discussion of the policy implications of our findings. Part I: The Role o( the Physician in hnproving Infant Outcomes Decreases in the infant mortality rate can occur through one of two channels: improvements in the underlying health of the fetus, and increases in the intensity or efficacy of interventions undertaken to keep fetuses of a given health alive. The latter channel can be quite expensive. Schwartz (1989) reports that although babies weighing less than 2500 grams account for only 9% of neonatal hospital caseloads, they account for 57% of the cost of neonatal hospital care. The average cost of caring for a surviving low birthweight baby was $9,712 compared to $678 for an infant weighing more than 2,500 grams. These costs rise as birthweight fails; in 1984, the cost of saving an infant with birthweight below 1000 grams was $118,000 (U.S. Office of Technology Assessment, 1987). These high costs have led policy makers to promote prenatal care as a means of reducing the infant mortality rate through improved fetal health. Clinical studies indicate that prenatal care is a cost effective way to achieve reduced mortality. The majority of unhealthy infants are pre-term. Inexpensive screenings can identify women at risk of pre-term deliveries due to medical, social, or economic factors. Given early identification of high-risk pregnancies, many pre-term births could be prevented, either through appropriate counseling (about diet and about dangers to the fetus from cigarettes, drugs or alcohol consumption), or drug therapies. Several studies cited in the Institute