Medicaid Coverage and Medical Interventions During Pregnancy
Leo Turcotte
Department of Economics and Finance
West Chester University
West Chester, PA 19383
E-mail: [email protected]
John Robst*
Centers for Medicare and Medicaid Services
7500 Security Blvd.
Mail stop C3-19-26
Baltimore, MD 21244
E-mail: [email protected]
Solomon Polachek
Department of Economics
SUNY – Binghamton
Binghamton, NY 13902
E-mail: [email protected]
*
Corresponding author. This research was completed while I was a performing post-doctoral
work at SUNY-Binghamton. The opinions in this paper are the authors’ and do not represent
those of the Centers for Medicare & Medicaid Services.
Medicaid Coverage and Medical Interventions During Pregnancy
This paper extends prior research on the effect of Medicaid coverage on medical interventions
during pregnancy (ultrasound) and birth (cesarean delivery, inducement, and fetal monitor). The
data are from two sources: the New York State Vital Statistics (VS) matched infant birth-death
file and the Statewide Planning and Research Cooperative System (SPARCS) file for 1993-1996.
Medicaid coverage increases the likelihood of teens and adults receiving prenatal care relative to
being uninsured. Overall, the effect of insurance type varies depending on whether the
procedure is part of standard care (ultrasound and fetal monitor) or more likely to be elective
(inducement and cesarean delivery). Insurance type has a greater effective for elective
procedures than for procedures that are part of standard care.
Keywords: Health insurance; Medicaid; medical interventions; procedure utilization
1
1. Introduction
The late 1980s marked a time of expanding Medicaid coverage. By the early 1990s,
Medicaid covered all pregnant women and newborns with family incomes below 133 percent of
the poverty line, leading to a substantial increase in the number of Medicaid covered births
(Singh, Gold, and Frost, 1994). In New York State, pregnant women and infants were covered
up to 185 percent of the poverty line. The age limit for dependent children also increased in
New York, from 1 year old in 1989 to 14 years old by 1996 (American Academy of Pediatrics,
1995). Such expansions in coverage for pregnant women were expected to improve infant
outcomes, particularly by reducing infant mortality.1
Infant outcomes may improve for two reasons. First, Medicaid lowers (or eliminates)
out-of-pocket costs encouraging pregnant women to seek prenatal care. Adequate prenatal care
is expected to reduce the incidence of premature deliveries, low birth weight babies, and infant
mortality.2 Second, current thinking suggests that one’s health insurance quality affects the care
received. Profit-maximizing behavior implies that providers supply greater care for patients with
more generous insurance reimbursements implying that the uninsured, Medicaid covered, and
privately insured receive different procedures.3 For example, Haas et al. (1993) showed that
Medicaid patients in Massachusetts have a higher likelihood of cesarean delivery than the
uninsured. Epstein and Newhouse (1998) found that Medicaid expansions reduced differences in
cesarean deliveries between the privately insured and Medicaid/uninsured in California, but not
South Carolina.
While cesarean sections are an important medical intervention, there are many other
obstetrical procedures so that limiting the analysis to a single procedure may provide an
incomplete picture. For example, Currie and Gruber (1997) examined four interventions that
2
may occur at or near birth (cesarean delivery, ultrasound, fetal monitor, and induction). In their
study, the use of interventions increased among teens eligible for Medicaid, but fell among
Medicaid eligible adults. Currie and Gruber (1997) hypothesized that the Medicaid expansions
lead to more teens moving to Medicaid coverage from the ranks of the uninsured, and as a result
utilization increased. Adults were being crowded out from private insurance leading to lower
utilization.
While Currie and Gruber (1997) examined a number of interventions their focus is on
Medicaid eligibility. Studies that focus on Medicaid coverage limit their analysis to a single
intervention (e g., Haas et al., 1993; Epstein and Newhouse, 1998). While government policy
directly influences the number of Medicaid eligible, it is important to also examine procedure
utilization and outcomes for those actually covered by Medicaid. The purpose of this paper is to
examine how actual Medicaid coverage affects medical interventions for pregnant women.
In addition, this study differentiates between women with and without medical risks and
complications. Most studies control for the presence of medical risks and complications but do
not consider whether utilization differences between insurances depend on such factors. Given
such pregnancies require greater service utilization, policy makers should be interested in
whether Medicaid coverage reduces utilization differences for women with risks and
complications. Examining relatively homogeneous subgroups should also limit selection
problems that result when insurance choices are based on expected utilization.
2. The Literature
Several studies use state-specific data linking birth certificate files with hospital records
that report primary payer. Such studies found that Medicaid coverage increases prenatal care
3
(e.g., Ray et. al., 1997), leads to fewer low birth weight babies (Long and Marquis, 1998), and
reduces infant mortality (Long and Marquis, 1998). As discussed above, the results are mixed
for cesarean delivery (Haas, et. al., 1993; Epstein and Newhouse, 1998), but such studies have
not looked at other medical interventions. A shortcoming of this approach is that it does not
adequately deal with selection issues. Type of insurance may be endogenous with pregnant
women potentially choosing coverage based on their expected need for care or to minimize outof-pocket costs. Women who expect to need more care may enroll in insurance types that
provide better benefits and/or less restrictive treatment protocols (Robinson, Gardner, and Luft,
1993).
Three studies account for selection issues through the use of vital statistics data (birth
and/or death certificates). Dubay et al. (2001) examined changes in prenatal care for various
demographic groups, focusing on demographic groups most likely to be affected by Medicaid
expansions. The authors concluded that Medicaid expansions improved prenatal care. Currie
and Gruber (1996) estimated increases in Medicaid eligibility at the state level using CPS data.
States with larger predicted increases in eligibility had larger reductions in infant mortality.
Currie and Gruber (1997) predicted whether an individual is Medicaid eligible based on
demographic characteristics (age, race, and education). As such these studies examine the
effects of Medicaid expansions without regard to individual choices of actual coverage.
We use a data set that links birth and hospital records in New York State, a state not
examined in the existing literature on Medicaid expansions, to determine the effects of Medicaid
coverage on medical interventions. Similar to other studies using state-level data actual
Medicaid coverage is examined. A model that addresses the potential endogeneity of insurance
coverage would be optimal. However, the data do not have information on premiums, a key
4
variable that determines such choices (Buchmueller and Feldstein, 1997). Whereas prior
research uses Medicaid eligibility as an instrument for Medicaid coverage, Medicaid does not
cover three quarters of eligible women (Cutler and Gruber, 1996). Women who are not covered
by Medicaid may have other insurance or remain uninsured. As such, eligibility is an imprecise
instrument for coverage. To account for selection based on expected health care utilization, we
divide the sample based on the presence of medical risks and complications. The health care
needs of women with medical risks who develop complications during labor or delivery are
expected to be much greater than for women without risks or complications. Examining
subgroups with more similar medical needs should reduce selection biases. In addition, while
women may select insurance based partly on the presence of medical risks, they are locked into
their plan when a complication develops during labor or delivery.
3. The Data and Methodology
This analysis uses data provided by the New York State Department of Health. The data
are from two sources: the NYS Vital Statistics (VS) matched infant birth-death file and the
Statewide Planning and Research Cooperative System (SPARCS) file for 1993-1996 inclusive.4
SPARCS is a comprehensive patient data system that contains patient-level information for
inpatient stays including the actual primary payer. Payers are Blue Cross, self-pay (e.g.,
uninsured, charity or uncompensated care)5, commercial insurers (e.g., fee-for-service and
HMOs), and public insurers such as Medicaid, Medicare, Civilian Health and Medical Plan for
the Uniformed Services (CHAMPUS) and other federal programs. The VS data contain records
of every live birth occurring in NYS. Included in the data are certain obstetrical procedures and
women’s demographic characteristics. The SPARCS data contain inpatient information including
5
a wide range of maternal medical risks, complications during labor or delivery, and non-medical
risks (e.g., smoking, tobacco use).6
Our analysis is conducted for births to women 15-44 years of age that take place in a
hospital. In addition to diving the sample based on the presence of medical risks and
complication, the data are broken into two additional categories, category one is comprised of
teens (14 < age < 19) and high school dropouts, while the second category contains high school
graduates.7 The motivation for this is twofold. First, insurance coverage between these two
groups varies dramatically so that not separating the groups could lead to erroneous conclusions
(See Table 1). Second, we want to account for the higher probability of insurance before
pregnancy among the more educated. While teens are often uninsured before becoming
Medicaid eligible, the expanded eligibility criteria for pregnant women crowds out private
insurance among the more educated (Cutler and Gruber, 1996). Firms may eliminate employersponsored insurance if public insurance is available, or women may choose Medicaid over
private insurance due to the lower out-of-pocket costs.
Analysis of the effects of health insurance consists of estimating separate regressions for
the four procedures contained in the VS data. Since Medicaid expansions were mostly complete
by the early 1990s and our data cover 1993-1996, we are unable to compare procedure utilization
before and after the Medicaid expansions. Instead we rely on cross sectional variation to
examine how Medicaid coverage affects the likelihood of receiving each procedure. The four
years of data are pooled to provide sufficient sample sizes for each subgroup of women. Logit
models are estimated as follows:
PROCi = Xi⋅β + INSURANCEi⋅δ +Zi ⋅α + ui
6
where PROC is a dichotomous variable indicating whether individual i received the procedure
being analyzed; X is a vector of the individual's demographic characteristics: age is a continuous
variable for all women of child bearing age (15-44), race is separated into three categories
(white, black, other), education denotes the number of years of schooling completed, degree of
urbanization is a vector of categorical variables denoting the Rural-Urban Continuum Codes for
Metropolitan and Non-metropolitan Counties. INSURANCE is a vector of categorical variables
denoting type of insurance. We focus the analysis on Medicaid, commercial insurance, and selfpay (i.e., the uninsured). While there are some individuals in Medicaid HMOs, the vast majority
of Medicaid enrollees in New York during this time frame had fee-for-service coverage. As
such, we differentiate between individuals enrolled in commercial fee-for-service plans and
commercial HMOs, and focus on fee-for-service type plans only. This will minimize estimation
problems that may result from differential treatment provided by fee-for-service plans and
HMOs. The specification also includes a vector (Z) of medical risks, complications during labor
or delivery, and non-medical risks (e.g., smoking, tobacco use). For those without medical risks
or complications, this vector reduces to non-medical risks only. Since the regressions are
estimated using logit analysis, marginal effects are computed as δp(1-p) where δ denotes the
coefficients on the insurance variables and p is the proportion of the sample receiving the
procedure.
One problem is the potential variation in practice patterns across physicians and
hospitals. Such patterns may be associated with accepting Medicaid patients or the proportion of
patients with Medicaid. Whereas this problem may be relevant, all the studies discussed earlier
suffer the same shortcoming. Only Wright, Gardin, and Wright (1984) control for the practice
patterns of physicians in their study of HMOs, but this is accomplished by limiting the sample to
7
a single provider group. Such a restrictive sample would make it difficult to generalize the
results.
4. The Procedures and Expected Relationship to Insurance Type
Multiple actors including physicians, patients, and insurers (both patient and physician
malpractice) determine the utilization of medical procedures. We discuss some of the
motivations that each actor has to affect utilization, and then explore how these motives apply to
each insurance type.
A number of theoretical models predict a relationship between insurance type and
physician-induced demand for services (Evans, 1974; Farley, 1986; De Jaeger and Jegers, 2000).
As long as the price received by physicians exceeds their marginal cost, they may provide care in
excess of what a perfectly informed patient would desire. Accordingly, in the absence of other
utilization controls, payment differences across insurances may influence patient care. While
physicians may induce some demand, patients also demand medical care based on the out-ofpocket cost of care. Insurance lowers the marginal price of medical care since patients are
usually responsible for copayments and coinsurance with the insurer often paying the vast
majority of the cost. As a result, insured patients may demand services in excess of those
purchased by someone without insurance. Insurers can also play a direct role in the provision of
services. The insurer covering the patient can limit utilization by denying coverage for elective
procedures. At the same time, malpractice insurers may encourage or require physicians to
perform procedures that are considered to be the standard of care.
Two comparisons are relevant in this study. As noted earlier, individuals who are
covered by Medicaid during birth are often uninsured or covered by private insurance prior to
8
pregnancy. For example, Adams, Gavin, Handler, Manning, and Raskin-Hood (2003) found that
the majority of women covered by Medicaid during the birth were uninsured before pregnancy.
In addition, approximately 10 percent moved from private insurance prior to the pregnancy to
Medicaid coverage. As such, to examine the effect of Medicaid coverage on utilization requires
comparison of Medicaid enrollees to the uninsured and to those who are privately insured.
Relative to private fee-for-service insurance, providers treating Medicaid patients face
different financial considerations and thus may provide different services. Medicaid programs
typically have lower provider payments for services than commercial plans. For example, Currie
and Gruber (1997) found that Medicaid pays substantially less than private insurers for cesarean
deliveries. Such a price differential limits physician-induced demand for Medicaid enrollees
relative to the privately insured. Alternatively, being uninsured increases the out-of-pocket cost
of medical care for patients, which is likely to reduce the quantity demanded of procedures.
Treating the uninsured may also reduce provider payments if the patient is unable to pay, again
leading to a lower physician-induced demand. At the same time, we expect malpractice insurers
limit differences in procedure utilization. Physicians work in a litigious environment and most
physicians and malpractice insurers work to minimize exposure to lawsuits. As such, physicians
are unlikely to limit services that are considered to be standard care and insurance type is more
likely to affect utilization of elective procedures. Below we discuss whether providers are
expected to alter utilization of each procedure based on patient insurance.
Ultrasound
The first procedure, ultrasound, is typically performed 18-20 weeks into the pregnancy in
order to detect problems early in the pregnancy and to confirm the due date.8 Still there is some
9
debate over whether women without risks or complications need to have an ultrasound
performed (Seeds, 1996; Long and Sprigg, 1998). Given such debate, we expect that some
women do not receive ultrasounds during pregnancy and that women without risks or
complications are less likely to receive the procedure than women with risks and complications.
Despite the debate about the necessity for low-risk patients, Medicaid coverage has been
found to improve prenatal care relative to being uninsured and we expect that Medicaid coverage
will increase the likelihood of receiving an ultrasound. Medicaid coverage reduces the out-ofpocket cost to patients leading to greater utilization. We expect little difference in ultrasounds
between Medicaid and privately insured patients given that ultrasounds have become a standard
part of prenatal care.
Fetal Monitor
The remaining three procedures, inducement/stimulation, fetal monitor, and cesarean
section are provided immediately prior to or during delivery. A fetal monitor records the
women’s contractions and the baby’s heartbeat. Fetal monitors can be either external, such as a
microphone placed near the women’s abdomen, or internal where electrodes are placed in the
baby’s scalp. Fetal monitors are used during the vast majority of deliveries. In 1996, 83 percent
of women were monitored electronically (Haggerty, 1999). Electronic fetal monitoring is useful
for detecting early fetal distress and monitoring high-risk women during delivery (Sweha,
Hacker, and Nuovo, 1998). EFM has risks however, including a tendency to produce falsepositive results that result in unnecessary surgical procedures. As such, fetal monitor usage is
more likely when the woman has maternal risks and complications, but we have no reason to
expect Medicaid enrollees to use fetal monitors more or less often than similar enrollees with
10
private insurance. Once again, fetal monitors have become a part of standard care that is
unlikely to vary based on insurance type.
Inducement/Stimulation
Inducement involves the starting or speeding up of labor contractions by the use of drugs
or other methods. There are several reasons why labor may be induced. Some women have
small pelvises, and birth is induced before the baby becomes too large to be delivered vaginally.
Some pregnancies are induced because the baby is post-term, because of an illness associated
with pregnancy such as toxemia, or due to a long labor. In other cases, labor is induced for
convenience to deliver the baby on a specific date. Stimulation involves the augmentation of
established labor typically through the use of oxytocin (Mathews, 1997). The procedure is
usually used when contractions occur in an irregular pattern. Induction and stimulation may be
used together or individually.
The use of induction and stimulation has been increasing rapidly. The rate of induction
rose from 9.4 percent of births in 1990 to 19.4 percent in 1998 (Rayburn and Zhang, 2002),
while the use of stimulation rose from 10.9 to 16.1 percent between 1989 and 1995 (Mathews,
1997). Inductions are performed for medical reasons such as a post-term pregnancy, medical
condition, and fetal compromise. However, much of the growth in utilization was due to elective
reasons such as patient convenience. This has lead some to raise concerns over the use of a
medical intervention that is not medically necessary (Rayburn and Zhang, 2002). Despite the
growth in elective inducement/stimulation, many inductions and stimulations are still performed
due to the presence of medical conditions, and we expect greater utilization among women with
medical risks and complications.
11
It is difficult to predict whether Medicaid enrollees are more or less likely to be
induced/stimulated during labor. Given that some inductions are for patient convenience, one
might expect that Medicaid programs would discourage the use of a procedure that is medically
unnecessary.9 Post-term pregnancies however increase the odds of expensive adverse outcomes,
which may lead insurers including Medicaid to allow for inducement/stimulation for post-term
pregnancies. One might expect lower Medicaid reimbursements to lead to fewer elective
inducements/stimulations and as a result, any differential between Medicaid enrollees and the
privately insured will be greater when the woman does not have medical risks or complications.
Similarly, relatively few elective inducements are anticipated among the uninsured without risks
or complications. Once again, assuming that inducement/stimulation is less likely to be elective
when the woman has medical risks and complications, we expect Medicaid coverage to have
little effect on utilization for women with risks and complications.
Cesarean Delivery
Cesarean deliveries involve a surgical procedure to deliver the baby. Cesareans may be
performed for medical reasons such as if the woman is ill, the pelvis is too small for natural
childbirth, the baby is in a breach position, or active labor has been ongoing for a long time.
Other times, cesareans are performed in order to schedule the delivery of the baby. The majority
of cesarean sections are due to four complications: breech, dystocia (slow to progress labor),
fetal distress, and previous cesarean section. The rapid rise in cesarean deliveries has been
accompanied by an increase in the dystocia and fetal distress diagnoses. Given that these
diagnoses are somewhat subjective, several have suggested that physicians are using these
diagnoses somewhat liberally (Tussing and Wojtowycz, 1994).
12
In the absence of constraints or other incentives, physicians may provide more cesarean
deliveries than optimal. The fee for a cesarean section is typically greater than for a vaginal
delivery, and the time required to perform a c-section is typically less than for a normal delivery.
Indeed, many argue such financial and time incentives have contributed to the tremendous
growth in cesarean deliveries over the past few decades. Given that Medicaid reimbursements
are lower than private insurance, we expect Medicaid enrollees receive fewer cesarean deliveries
than the privately insured but more than the uninsured. Once again, we expect a greater
difference by insurance type when the procedure is elective, and that this occurs when the
woman has no medical risks or complications
5.
Empirical Results
Women without medical risks or complications
Descriptive statistics are provided in Table 1 for women without medical risks or
complications. There are important differences between teens and high school graduates. Aside
from the age and educational differences, the teens/dropouts are more likely to be on Medicaid
(61 versus 14 percent) or uninsured (13 versus 8 percent) and less likely to have commercial
insurance (8 versus 29 percent). Similarly, procedure utilization differs across the insurance
groups. For example, among the higher educated group, cesareans are least likely for participants
in Medicaid. Nonparametric ANOVA tests (Kruskal-Wallis) indicate statistically significant
differences in utilization across types of insurance for all four procedures for high school
graduates and three of the four for teens/high school dropouts.
<Table 1>
13
Table 2 analyzes each procedure’s utilization for teens/dropouts and high school
graduates. These represent the logit results for each procedure with coefficients in the first
column and the marginal probabilities in the second column. Relative to being uninsured, teens
on Medicaid are 5.1 [2.0 – (-3.1)] percent more likely to receive an ultrasound, 3.2 percent less
likely to be induced, and 1.2 percent less likely to receive a cesarean delivery. Similarly, high
school graduates are more likely to receive an ultrasound and fetal monitor, but less likely to be
induced or have a cesarean delivery. The second comparison is between those with Medicaid
coverage and those with commercial insurance. For those crowded out of private insurance or
who choose to leave private insurance for Medicaid coverage, the rate of ultrasounds is not
significantly lowered. The rate of cesarean deliveries is lower as is the likelihood of being
induced.
<Table 2>
Women with medical risks and complications
Insurance coverage differs considerably between women based on the presence of
medical risks and complications. Compared to those without risks or complications, a slightly
higher proportion of teens are covered by Medicaid (67 versus 61 percent). A correspondingly
lower proportion is covered by commercial insurance. The proportion uninsured falls by a
relatively small 1.6 percent. Similar results exist for the higher educated group. In addition,
procedure utilization differs between those with and without medical risks and complications.
The biggest differences are seen in inducement/stimulation and cesarean delivery. Individuals
with risks and complications are much more likely to receive the procedures. As discussed
earlier, ultrasounds and fetal monitors are used in the vast majority of pregnancies. While used
14
more often when there are maternal risks and complications, the procedures are also used in the
majority of pregnancies without risks or complications.
<Table 3>
Compared to being uninsured, Medicaid coverage increases utilization of ultrasound, but
decreases utilization of cesarean deliveries. For example, teens with Medicaid coverage are 3.2
percent more likely to have an ultrasound than the uninsured, while high school graduates are 2.4
percent more likely. Medicaid covered teens receive ultrasounds more often than the
commercially insured and none of the procedures significantly less often. High school graduates
with Medicaid are less likely to receive inducements/stimulation or a cesarean delivery compared
to the privately insured. Having Medicaid increases the likelihood of having an ultrasound
relative to being uninsured but has no significant effect on the other procedures.
<Table 4>
6. Discussion
Does Medicaid coverage affect the care received by pregnant women? We utilize vital
statistics data merged with hospital records to concentrate on the use of four procedures. The
selection problem is dealt with by dividing the sample based on the presence of maternal risks
and complications. Utilization of the four obstetrical procedures is much greater for women with
risks and complications. As such, estimation biases introduced by women choosing insurance
type based on expected utilization should be reduced.
Two of the procedures considered, ultrasound and fetal monitor are standard procedures
used in the vast majority of pregnancies and births. As such, physicians were not expected to
provide fewer procedures to Medicaid recipients compared to the privately insured. The results
15
show that teens with risks and complications were more likely to receive ultrasounds while high
school graduates without risks or complications were more likely to receive fetal monitors.
Thus, despite the potential for physicians to provide more services to the privately insured, no
such difference is evident for services that are part of standard obstetrical care.
Utilization of standard procedures was expected to be greater for Medicaid recipients
compared to the uninsured. Such differences may result from the high out-of-pocket costs for
the uninsured. Indeed, compared to the uninsured, these procedures were received more often by
both teens and high school graduates regardless of whether they have medical risks or
complications. Becoming eligible and enrolling in Medicaid due to pregnancy improves prenatal
care, one of the primary goals of the Medicaid expansions.
Differences based on insurance type are more prevalent for procedures that may be
performed electively. The privately insured are more likely to be induced and receive cesarean
deliveries than women covered by Medicaid. There is some support for the hypothesis that
differences by insurance type are greater for women without risks or complications. The
differences between those covered by Medicaid and the privately insured are insignificant for
teens with medical risks and complications, but are statistically significant for teens without risks
and complications. Utilization differences by insurance type are significant for high school
graduates regardless of the presence of medical risks or complications, but the marginal effect on
the likelihood of being induced/stimulated is twice as large for women without risks or
complications.
While many cesarean deliveries are medically necessary, Gruber and Owings (1996) find
cesarean delivery increased 240 percent from 1970 to 1982. They claim that physicians
overused cesarean delivery relative to what would be chosen by a financially disinterested
16
provider. In addition, Keeler and Brodie (1993) argue that women also have economic
incentives to demand too many cesarean deliveries. A similar argument has been made for the
use of inducement and stimulation (Rayburn and Zhang, 2002). This could explain why women
covered by insurance with greater provider reimbursements have greater incidence of such
procedures.
These findings make four contributions to the literature. First, we illustrate the
importance of looking at multiple procedures, instead of prenatal care or cesarean deliveries only
to examine the effect of Medicaid coverage on treatment. Second, we extend prior research on
the effect of Medicaid eligibility to examine the effect of actual Medicaid coverage on such
procedures. Third, we compare procedure utilization for those with Medicaid to two other forms
of insurance. Fourth, we examine whether the effect of insurance type varies based on the
presence of medical risks and complications.
There are several limitations to this analysis. For example, insurance coverage is
dynamic, but the data only contain the actual primary payer for the inpatient stay. Prenatal care
such as the receipt of an ultrasound will depend on insurance coverage during the pregnancy,
which may not be the same as coverage during the delivery.
17
References
Adams, E.K., Gavin, N.I., Handler, A., Manning, W., and Raskin-Hood, C. (2003). Transitions
in insurance coverage from before pregnancy through delivery in nine states, 1996-1999. Health
Affairs, 22, 219-229.
American Academy of Pediatrics. 1995. Medicaid State Report. Downloaded October 21, 2002
from: http://www.aap.org/member/medrpts/95rpts/ny.pdf
Buchmueller, T.C. and Feldstein, P.J. (1997). The effect of price on switching among health
plans. Journal of Health Economics, 16, 231-247.
Currie, J., Gruber, J. 1996. Saving babies: the efficacy and cost of recent expansions of Medicaid
eligibility for pregnant women. Journal of Political Economy 104, 1263-1296.
Currie, J., Gruber, J. 1997. The technology of birth: health insurance, medical interventions, and
infant health. National Bureau of Economic Research Working Paper 5985.
Cutler, D., Gruber, J. 1996. Does public insurance crowd out private insurance? Quarterly
Journal of Economics 111, 391-430.
De Jaegher, K., and Jegers, M. (2000). A model of physician behavior with demand inducement.
Journal of Health Economics, 19, 231-258.
Dubay, L., Joyce, T., Kaestner, R., Kenney, G.M. 2001. Changes in prenatal care timing and low
birthweight by race and socioeconomic status: implications for the Medicaid expansions for
pregnant women. Health Services Research 36, 373-398.
Epstein, A.M., Newhouse, J.P. 1998. Impact of Medicaid expansion on early prenatal care and
health outcomes. Health Care Financing Review 19, 85-99.
18
Evans, R.G. (1974). Supplier-induced demand: Some empirical evidence and implications. In:
Perlman, M. (Ed.), The Economics of Health and Medical Care. Macmillan: London, pp. 162201.
Farley, P.J. (1986). Theories of the price and quantity of physician services. Journal of Health
Economics, 5, 315-333.
Goldenberg, R.L., Rouse, D.J. 1998. Prevention of premature birth. New England Journal of
Medicine 339, 313-320.
Gruber, J., Owings, M. 1996. Physician financial incentives and cesarean section delivery. The
Rand Journal of Economics 27, 99-123.
Haas, J.S., Udvarhelyi, I.S., Morris, C.N., Epstein, A.M. 1993. The effect of providing health
coverage to poor uninsured pregnant women in Massachusetts. Journal of the American Medical
Association 269, 87-91.
Haggerty, L.A. (1999). Continuous electronic fetal monitoring: Contradictions between practice
and research. Journal of Obstetric, Gynecologic, and Neonatal Nursing, 28, 409-416.
Howell, E.M. 2001. The impact of the Medicaid expansions for pregnant women: a synthesis of
the evidence. Medical Care Research and Review 58, 3-30.
Keeler, E.B., and Brodie, M.A. (1993). Economic incentives in the choice between vaginal
delivery and cesarean section. Milbank Quarterly, 71, 365-404.
Long, S.H., Marquis, M.S. 1998. The effects of Florida's Medicaid eligibility expansion for
pregnant women. American Journal of Public Health 88, 371-376.
Long, G., and Sprigg, A. (1998). A comparative study of routine versus selective fetal anomaly
ultrasound screening. Journal of Medical Screening, 5, 6-10.
19
Mathews, T.J. (1997). Trends in stimulation and induction of labor, 1989-1995. Statistical
Bulletin –Metropolitan Life Insurance Company, 78, 20-26.
Pauly, M.V. 1987. Nonprofit firms in medical markets. American Economic Review 77, 257262.
Ray, W.A., Mitchel, E.F., Piper, J.M. 1997. Effect of Medicaid expansions on preterm birth.
American Journal of Preventive Medicine 13, 292-297.
Rayburn, W.F., and Zhang, J. (2002). Rising rates of labor induction: present concerns and future
strategies. Obstetrics and Gynecology, 100, 164-167.
Robinson, J.C., Gardner, L.B., and Luft, H.S. (1993). Health plan switching in anticipation of
increased medical care utilization. Medical Care, 31, 43-51.
Seeds, J.W. (1996). The routine or screening obstetrical ultrasound examination. Clinical
Obstetrics and Gynecology, 39, 814-830.
Singh, S., Gold, R.B., Frost, J.J. 1994. Impact of Medicaid eligibility expansions on coverage of
deliveries. Family Planning Perspectives 26, 31-33.
Sweha, A., Hacker, T.W., and Nuovo, J. (1999). Interpretation of the electronic fetal heart rate
during labor. American Family Physician, 59, 2487-2500.
Tussing, A.D., and Wojtowycz, M.A. (1994). Health maintenance organizations, independent
practice associations, and cesarean section rates. Health Services Research, 29, 75-93.
Wright, C.H., Gardin, T.H., and Wright, C.L. (1984). Obstetric care in a health maintenance
organization and a private fee-for-service practice: A comparative analysis. American Journal of
Obstetrics and Gynecology, 149, 848-856.
20
Endnotes
1
Howell (2001) reviews the literature examining the impact of Medicaid expansions on infant
outcomes.
2
Recent evidence however, has questioned the link between prenatal care and premature
deliveries (e.g., Goldenberg and Rouse, 1998).
3
Even non-profit organizations need to generate a surplus, i.e., be cost minimizers and thus act
like a for-profit firm (Pauly, 1987). However, how they generate this surplus and how much of a
surplus they generate may differ from for-profit firms, leading to differences in treatment.
4
Data from New York City were not available and they were excluded from the analysis.
5
Self-insured includes births for which there is no known payer at the time of hospitalization.
Although the method of payment can be changed at a later date, these changes would not impact
the provider's choice of procedure(s) at the time.
6
Expectant mother’s medical risk factors include anemia, mellitus diabetes, genetic diseases,
genital herpes, heart disease, hemoglobinopathy, hepatitis b, chronic hypertension, previous low
birth-weight infant, chronic lung disease, macrosomia or previous infant > 4000g, previous preterm infant, renal disease, RH sensitized, seizure disorders, previous spontaneous fetal death,
thrombophlebitis, thyroid condition, gestational diabetes, in vitro fertilization, other fertilization
treatment, hydramnios/oligohydramnios, pregnancy related hypertension, preeclampsia,
eclampsia, incompetent cervix, acute lung disease, rubella, syphilis, other sexually transmitted
disease, tuberculosis, uterine bleeding, viral disease, and other medical risk factor. Medical
complications during labor or delivery are abruptio placenta, cephalopelvic disproportion,
chorioamnionitis, coagulation defects, cord conditions, cord prolapse, failure to progress, fetal
distress, fever, postpartum hemorrhage, cervical or vaginal lacerations, marginal sinus rupture,
21
meconium particulate or moderate/heavy, non-vertex presentation, placenta previa, precipitous
labor, prolonged rupture membrane (> 12hr.), premature rupture membrane (> 1hr. & < 12hr.),
prolonged labor, retained placenta, seizures, uterine atony and other complications. Non-medical
risk factors include smoking, alcohol use, drug use, mother’s weight at delivery and weight gain
during pregnancy.
7
Women with medicals risks or complications are excluded in the sample. The point of
dividing the sample was to define two groups of women, one with high expected utilization and
one with lower expected utilization. The implications of having complications but no medical
risks (or risks but no complications) on expected utilization and insurance choice are less clear.
8
Some ultrasounds may also be performed during delivery, especially for certain complications
and prior to some cesarean deliveries. In addition, an ultrasound may be performed for some
post-term pregnancies. Over 96 percent of the ultrasounds reported in our data were performed
prior to labor/delivery.
9
It would be preferable to examine medically indicated and elective induction and stimulation
separately, but the data do not contain the reason for receiving the procedure(s). We assume
inducement/stimulation is more likely elective when the woman had no risks or complications,
and more likely to be medically necessary when risks and complications are present.
22