بیمه سلامت عمومی و درمان پزشکی: تاثیر یکسان گسترش خدمات درمانی

We investigate the impact of expanding public health insurance on the medical treatment received by women at childbirth, using Vital Statistics data on every birth in the US over the 1987–1992 period. The effects of insurance status on treatment are identified using the tremendous variation in eligibility for public insurance coverage under the Medicaid program over this period. Among low education mothers who were largely uninsured before being made eligible for Medicaid, eligibility for this program was associated with significant increases in the use of a variety of obstetric procedures. Among women with more education, however, there is a countervailing effect on procedure use. Most of these women had private insurance before becoming Medicaid-eligible, and some may have been ‘crowded out’ onto the public program, moving from insurance which reimburses medical care more generously to insurance with much less generous reimbursement. This movement was accompanied by reductions in procedure use. Thus, on net, the Medicaid expansions had an equalizing effect, increasing the treatment intensity of the previously uninsured while lowering it among the previously insured.

The share of the US population without health insurance coverage has grown by 15% over the past 8 years to 17.4% (Employee Benefits Research Institute, 1996). This decline in insurance coverage has motivated calls for expanded public insurance as a means of guaranteeing access of the uninsured to high quality medical care. Past research suggests that expansions in public insurance can indeed increase contacts of the uninsured with the medical system. A large number of studies further suggest that those with private insurance coverage are treated more intensively than the uninsured. These studies, however, are able to control for differences in the underlying health of insured and uninsured patients in only a limited way. In studies of the effects of insurance coverage on the treatment of hospitalized patients, it is also difficult to control for selection into hospital on the basis of health status. Perhaps for these reasons, the available evidence regarding the effects of public insurance coverage on treatment intensity is mixed. Moreover, no previous work has considered a potentially countervailing effect of public insurance expansions on treatment intensity. Expansions in public insurance can be associated with reduced private insurance coverage among the target population. This ‘crowdout’ of private insurance coverage may lead to reductions in treatment intensity, as public coverage generally reimburses providers at a much lower level than does private coverage. Thus, overall, the impacts of public insurance expansions on treatment intensity may be ambiguous. In this paper, we address both of these issues in the context of the treatment of childbirth. The main advantage to our approach is that we are able to exploit the tremendous variation in insurance status that arose from expansions of the Medicaid program, the public insurance program that covers low-income women and children.1 Among pregnant women, eligibility for Medicaid coverage has greatly expanded since 1987, and this expansion has occurred at a differential pace across the states. These eligibility changes can be used to identify the effect of insurance status on treatment, producing estimates that are not contaminated by unobserved individual heterogeneity. Moreover, since virtually every woman in the United States delivers her baby in a hospital, and hospitals are essentially required to treat women in labor, it is possible to obtain a picture of treatment patterns that is not contaminated by the selection of patients into the hospital. We do so using excellent national data on the treatment of childbirth that is available from the National Center for Health Statistic’s (NCHS) uniform birth certificate data. These data cover the full census of births in the US in each year, and provide information on several common interventions used during childbirth. We find that recent expansions of the Medicaid program had significant effects on the medical treatment of child birth. We focus first on mothers who are teens, high school dropouts, or unmarried high school graduates, a group that was largely without insurance before becoming eligible for Medicaid. In this group, eligibility expansions increased the generosity of insurance coverage, and we find that increased eligibility was associated with an increase in the utilization of a variety of obstetric procedures. But we find evidence of a countervailing effect on aggregate procedure utilization among mothers who were married high school graduates or who had some college education. These women were much more likely to have had private insurance coverage before becoming eligible for Medicaid, but a large share became eligible for the program. Some of these women may have been ‘crowded out’ of private insurance onto the public program in response to becoming eligible. To the extent that this movement occurs, these women are moving to insurance (Medicaid) which reimburses physicians much less generously than do private insurance plans. As a result, we find that in this group, increased eligibility is accompanied by reductions in procedure utilization, which in the aggregate largely offset the increases in procedure use among the (smaller) group of teens and dropouts. We confirm these findings by using data on relative Medicaid reimbursement of Cesarean section delivery to show that the effects are consistent with physician responses to reimbursement differentials. And we show that for college graduate women, for whom the expansions are largely irrelevant, there is little association between Medicaid eligibility and procedure utilization. The paper proceeds as follows: Section 2 provides background information about the Medicaid expansions and prior evidence regarding the effects of insurance coverage on the utilization of hospital care. Section 3 describes the data sources and empirical strategy. Section 4 documents the effects of Medicaid eligibility on the treatment of childbirth. Confirmation and Conclusions are presented in 5 and 6, respectively.

This study offers evidence that insurance affects the way patients are treated, in both expected and unexpected ways. Among teen mothers, high school dropouts, and unmarried high school graduates who would be largely uninsured in the absence of Medicaid, we find that expansions of Medicaid eligibility were associated with the increased use of a variety of procedures, suggesting that increasing the generosity of insurance coverage causes an increase in treatment intensity. We also show that physician financial incentives played an important role in this move to increased treatment intensity, as the effect on cesarean-section delivery was largest where differential Medicaid reimbursement of cesarean delivery was most generous. Thus, we extend the results of previous studies that suggested that insurance increases treatment intensity relative to no insurance, but in a more statistically convincing framework. As we have highlighted, however, there is a countervailing effect on procedure use among higher education mothers, some of whom may be ‘crowded out’ of their private insurance by increased Medicaid eligibility. These women can be thought of as moving from more generous to less generous coverage of their pregnancies. In this group we find reductions in procedure use. These reductions are mitigated as Medicaid reimbursement for more intensive treatment rises, confirming that our findings are driven by the reimbursement change induced by moving from private to public coverage. These results have important implications for measuring the social costs of expanding eligibility for public health insurance. The key issue in assessing these implications is the value of the marginal treatments that are consumed or foregone as a result of changes in eligibility for public insurance. At one extreme, it is possible that marginal changes in treatment are irrelevant for both populations, since the truly medically necessary treatment is provided by responsible providers in any case. In terms of the medical effectiveness curve which plots health production against spending, in this case we would be on the proverbial ‘flat of the curve’, and these results would suggest that there is little aggregate impact of the change in treatment intensity associated with the expansions, since overall treatment intensity is not much changing. Another possibility is that the changes in marginal treatment are important for health outcomes for both populations. In this case, there is a tradeoff associated with the public health insurance expansions; more valuable health care for the uninsured, but less valuable health care for the previously insured. Such a tradeoff could nevertheless be welfare improving, under either of two conditions. First, if there is some concavity to the relationship between medical spending and health production, then there are larger marginal gains to increases in treatment for the previously uninsured than for decreases in treatment for the previously insured. Second, if the social welfare function is concave, this policy would amount to redistributing resources from higher income to lower income groups. Finally, the ‘best case’ scenario would be that the marginal increases in treatment for the previously uninsured are valuable in terms of health, but that the reductions in treatment for the previously uninsured are not; that is, that the previously uninsured are on an upward sloping portion of the medical effectiveness curve, and the previously insured are on a flat portion.21 In this case, there are two types of welfare gains from the policy: improvements in the health of the uninsured, and reductions in wasteful treatment of the insured.22 Of course, there is a corresponding ‘worst case’ scenario that is the opposite, but this seems unlikely to be relevant; that is, it seems unlikely that the medical effectiveness curve is downward sloping over the relevant range. Ultimately, which of these scenarios is most relevant is an empirical question. We investigate this question in Currie and Gruber (1997), with mixed results. We examine the impacts of the Medicaid expansions on the infant mortality outcomes of these two groups of mothers. We find some suggestive evidence for the ‘best case’ of mortality reductions for the teen/dropout mothers and no effect on the higher education mothers, but the results are statistically imprecise. We do find, however, that those teen/dropout mothers who live near a Neonatal Intensive Care Unit see more significant reductions in mortality associated with the expansions. This suggests that the expansions may have increased access to this ‘higher tech’ source of treatment of childbirth, as well as to the ‘lower tech’ treatments that we study. At the same time, we continue to find little impact of the expansions on the mortality of children born to mothers with more education, regardless of proximity to a NICU. These results suggest that the social costs of expanding eligibility for health insurance to the needy could be offset to some extent by reductions in the number of procedures obtained by the more affluent, without causing any harm. Indeed, although the Medicaid expansions increased public expenditures, they may have had little effect on the net social costs of paying for child birth and neonatal care, while equalizing the treatment of more advantaged and less advantaged groups of mothers. Our findings raise an important priority for future work: assessing the process by which hospital resources are differentially applied to women with insurance coverage of differing levels of generosity. Does differential procedure use arise largely through hospital choice, or through changes in treatment intensity within hospitals? How do hospital and physician financial incentives interact to determine the treatment of differentially insured patients? Answers to these questions will help provide a richer understanding of the health production process, as well as providing insights into efficient reimbursement and insurance eligibility strategies for the public sector.