Unintended pregnancy has long been acknowledged as an important health, social and economic problem in the United States, one that creates hardships for women and threatens the health and well-being of their infants.1–6 Those consequences, in turn, have a broad societal impact, such as on the national economy and the extent of government expenditures. Rates of unintended pregnancy are far higher among women living in poverty and low-income women (those with an income at 100–199% of the federal poverty level) than among higher income women (those with an income at or above 200% of poverty)—a disparity that grew substantially between 1994 and 2001.7 Most of these poor and low-income women are eligible for public coverage of pregnancy-related care through Medicaid and the Children’s Health Insurance Program (CHIP).
Care related to unintended pregnancy presents substantial costs to the federal and state governments in the form of reimbursements through these programs, although little information has existed to gauge the overall magnitude of these costs. One obstacle to an in-depth look at the public costs of unintended pregnancy is that states vary widely in every component of the necessary calculations: from rates of unintended pregnancy and resulting births, to the proportion of those births that are publicly funded, to the cost to public programs of covering such births. State-level estimates require state-level data on these indicators, which have never before been available for all 50 states. This analysis is the first to use state-level data to estimate public expenditure on births resulting from unintended pregnancy, as well as the contribution of public insurance programs in providing essential care to pregnant women and infants.
Our analysis focuses on the cost of publicly funded births resulting from unintended pregnancies: those births paid for by Medicaid or CHIP, including Medicaid and CHIP managed care plans, and Medicaid and CHIP programs operating under Section 1115 waivers (which permit states to receive federal funding for programs that do not meet federal Medicaid and CHIP requirements). We include costs for prenatal care, labor and delivery, postpartum care and one year of care for the infant. This is the same convention used in a number of studies of cost savings associated with publicly funded contraceptive services and supplies.8–12 It is also the model that the federal government has used, and has required individual states to use, to evaluate the impact of demonstration programs that expand Medicaid eligibility specifically for family planning services.13,14
As in prior studies and evaluations of public costs and savings, our analysis includes all unintended births and makes no distinction between births resulting from mistimed pregnancies (i.e., pregnancies among women who had wanted to get pregnant, but at a later time) and those resulting from unwanted pregnancies (i.e., pregnancies among women who had not wanted to become pregnant at any time). Other studies, including one by Monea and Thomas,15 have argued for discounting births resulting from mistimed pregnancies in the calculations of public costs or potential public savings, because these births might “replace” later, planned births that would have been paid for with government funds. For this analysis, such an approach would have underestimated public costs and potential public savings. In some cases, a woman who has a mistimed birth achieves her preferred family size, only earlier; in other cases, if a woman has additional unintended pregnancies, a mistimed birth is an “extra” birth beyond what she preferred. Indeed, 44% of women aged 15–44 who have had an unintended pregnancy have had two or more unintended pregnancies, and 36% of women who have had a birth resulting from an unintended pregnancy have had two or more such births.16
Moreover, even if a mistimed birth replaces a later, planned birth, preventing it may do more than merely delay the public cost of that birth. For example, a woman who today is eligible for pregnancy-related care under Medicaid may have a higher income later in life that precludes her from eligibility. (Although the opposite is also possible, it is less likely, because income typically increases with age.17) Furthermore, for teenagers and young adults, a birth following an unintended pregnancy may curtail educational achievement and lifetime earnings potential,5,18,19 and the large majority of unintended pregnancies among these women are mistimed rather than unwanted.20 Accounting for these possibilities would be exceedingly difficult, if not impossible, with available U.S. data. Finally, immediate costs (e.g., one-year), rather than long-term costs, are typically paramount for policymakers and advocates.
To estimate the costs of publicly funded births, we needed to obtain three underlying state-level estimates: the number of births resulting from unintended pregnancies in a given year, the proportion of such births paid for by public programs and the cost to programs for each birth.
Number of Births
In a related analysis, Finer and Kost estimated 2006 unintended pregnancy rates for all 50 states and the District of Columbia.21 They obtained each state’s total number of births for 2006 from the U.S. vital statistics system.22 The proportion of those births that resulted from unintended pregnancies came, for most (39) states, from the Pregnancy Risk Assessment Monitoring System (PRAMS), a population-based surveillance project of the Centers for Disease Control and Prevention (CDC).23 For an additional five states, data came from similar state-conducted surveys.21 And for the remaining six states and the District of Columbia, multivariate linear regression was used to predict the unintended pregnancy rate.21 We obtained the estimated number of unintended births for each state from unpublished tabulations of data by Finer and Kost.24
Proportion of Births Paid for by Public Programs
PRAMS was the primary source for the proportion of births resulting from unintended pregnancies paid for by Medicaid and CHIP. The core PRAMS questionnaire for 2006 asked how the respondent’s delivery was paid for. Possible responses included Medicaid, personal income, private health insurance and up to two additional categories defined by individual states; respondents could also answer “other” and provide additional information.*
For some states, information about deliveries paid for by CHIP, Medicaid managed care plans or waiver programs were captured by the additional, state-defined categories and by the write-in responses (see appendix). For 28 states, estimates of the proportion of births resulting from unintended pregnancies paid for by Medicaid or CHIP were drawn from individual-level PRAMS data, obtained from the CDC, for either 2006 or the closest available year. For these states, complete data were available for the related programs, including any state-specific categories or eligible responses written in by at least 10 respondents. For eight other states, we requested tabulations that included these additional responses directly from state health departments. For seven additional states, data were limited to the Medicaid category included on the PRAMS questionnaire or that of a similar survey, and were obtained in aggregate from CDC’s Web site or from state health department tabulations or publications.
For the remaining seven states and the District of Columbia, the proportion of births resulting from unintended pregnancies paid for by Medicaid or CHIP was predicted using a multivariate linear regression model similar to that used by Finer and Kost to estimate state unintended pregnancy rates.21 In the model, each of the 43 states with data represented an observation. The dependent variable was the proportion of births following unintended pregnancies for which the delivery was covered by Medicaid or CHIP. Independent variables, measured at the state level, were demographic characteristics of women aged 15–44,† overall birthrate, birthrate associated with unintended pregnancies and income-eligibility threshold for pregnancy-related care under Medicaid and CHIP.22,25–28 The R-squared of the final model indicated that 77% of the variation in the dependent variable could be accounted for by the independent variables. The same procedures were used to obtain data on the proportion of births resulting from intended pregnancies and of all births that were paid for by Medicaid or CHIP; the R-squared coefficients for the two models indicated that 80% and 85% of the variation, respectively, could be accounted for by the independent variables. (See appendix for additional details.)
Cost per Publicly Funded Birth
State-level data on the average cost of a Medicaid-funded birth were drawn from earlier studies.9,11 (Data on the cost of a CHIP-funded birth were not available, but are assumed for the current analysis to be the same as for a Medicaid-funded birth.) Briefly, data on these costs are not consistently collected for all states, but were available in applications or evaluations completed by 24 states that have sought a federal waiver to expand Medicaid eligibility specifically for family planning services. For the remaining states, the authors obtained estimates by averaging the available data and adjusting for states’ Medicaid payment rates.9,11
For the current analysis, we adjusted the published data for inflation to 2006 dollars, using the medical care component of the Consumer Price Index.29 We then separated the average cost of a Medicaid-funded birth for each state into state and federal costs, on the basis of the state’s federal medical assistance percentage for FY 2006.30 We multiplied the number of births resulting from unintended pregnancies in each state by the proportion of such births paid for by public programs to arrive at each state’s number of publicly funded births from unintended pregnancies. That figure was then multiplied by the average cost of a Medicaid-funded birth in the state to arrive at a total cost for the state. A similar process was used for the cost of all publicly funded births in each state (including those from intended pregnancies, which we subsequently calculated by subtraction).
According to the National Survey of Family Growth (NSFG), an estimated 1.6 million births resulted from unintended pregnancies in the United States in 2006;31 by comparison, the estimates we use in this analysis24 sum to 1.8 million births from unintended pregnancies that year. To account for this difference, we present both unadjusted U.S. totals (summed from the state-level data) and adjusted U.S. totals (calculated as 89%—1.6 million divided by 1.8 million—of the unadjusted totals). The estimates may differ in part because of the timing of the survey interview in relation to the birth (the gap between a given delivery and the survey date could be up to five years for the NSFG, as compared with six months for PRAMS) or because of differences in the questions measuring intention status. Nevertheless, because the NSFG is designed to provide national estimates and because its intention status measure may be superior to the PRAMS measure, we expect the adjusted national totals to be more accurate. Coincidentally, they are also more conservative. We refer in this article exclusively to adjusted totals.
Publicly Funded Births
Nationally, 64% of the 1.6 million births resulting from unintended pregnancies in 2006 were paid for by public insurance programs (Table 1); in comparison, 48% of all births and 35% of births resulting from intended pregnancies were funded by these programs. We estimate that 1.0 million—or 51%—of the 2.0 million total publicly funded births in 2006 resulted from unintended pregnancies. By comparison, 38% of all births nationwide followed unintended pregnancies (1.6 million out of 4.3 million).
States vary considerably in their eligibility levels for pregnancy-related care, as well as in the demographic composition of their populations. Thus, they also vary considerably in the proportion of births that are publicly funded, regardless of pregnancy intention status. In 11 jurisdictions, at least 70% of births resulting from unintended pregnancies were paid for public programs; Louisiana and Mississippi had the highest proportions (81% each). All but one of those 11 jurisdictions are in the South (as categorized by the U.S. Census Bureau), a region with high levels of poverty. In eight states, by contrast, the proportion was below 50%; Hawaii had the lowest proportion (42%). The eight states with the lowest proportions follow no clear geographic pattern.
State-level patterns for public coverage of births following intended pregnancies and of overall births were very similar. Louisiana, Mississippi and the District of Columbia had the highest proportions (52–54% of births resulting from intended pregnancies and 63–69% of all births); other southern states followed closely. New Hampshire had the lowest proportions (15% and 26%, respectively).
Public-Sector Costs for Unintended Births
Government expenditures on births resulting from unintended pregnancies nationwide totaled $11.1 billion in 2006 (Table 2); of that, $6.5 billion were federal expenditures and $4.6 billion were state expenditures. On average, a publicly funded birth cost $11,647. To put these figures in perspective, the federal and state governments together spent an average of $180 on maternity and infant care related to births from unintended pregnancies for every woman aged 15–44 in the country.
Because of the wide variation in the number of births resulting from unintended pregnancies paid for by public programs and in the cost of a publicly funded birth, public expenditures on births following unintended pregnancies varied considerably across states. In seven states, these costs exceeded half a billion dollars. California and Texas spent the most—about $1.3 billion each. Controlling for population size, spending per woman aged 15–44 ranged from $97 in Oregon to $476 in Alaska.
In addition to the costs of births resulting from unintended pregnancies, the federal and state governments spent $10.7 billion for births from intended pregnancies, for a total of $21.8 billion for all publicly funded births (Table 3). Thus, 51% of government expenditures on births in 2006 were spent on births following unintended pregnancies ($11.1 billion of $21.8 billion).
This analysis demonstrates the importance of Medicaid and CHIP for assisting American women and families to afford the expense of pregnancy and childbirth. According to our estimates, 48% of all births in the United States in 2006 were paid for by these programs. (This estimate is somewhat higher than the 41% found by the National Governors Association for 2003;32 however, that estimate was based on an unweighted average of state rates.) The role of Medicaid in funding U.S. births has increased dramatically since the mid-1980s, when Congress first allowed and then required states to expand Medicaid eligibility to pregnant women at income levels well above those most states set for Medicaid more generally. In 1985, Medicaid paid for 15% of U.S. births; by 1991, that figure had more than doubled, to 32%.33
Our findings also reflect and help illustrate the increasing concentration of unintended pregnancy and resulting births among poor and low-income women. Sixty-four percent of births resulting from unintended pregnancies in 2006—one million of them—were publicly funded. By contrast, only 33% of women 15–44 that year had a family income below 200% of the federal poverty level34—which is roughly the income-eligibility ceiling for pregnancy-related care in most states’ Medicaid and CHIP programs.
The health, social and economic consequences of unintended pregnancies are undoubtedly substantial for women and families. In addition, these pregnancies create immense budgetary costs for federal and state governments—$11.1 billion in a single year. Indeed, births resulting from unintended pregnancies account for half of publicly funded births and their resulting costs. This is a disproportionate burden on programs, given that only 38% of all U.S. births result from unintended pregnancies.
Staggering as these numbers are, they would be even higher if not for continued federal and state investments in family planning services. In 2008, an estimated $2 billion in expenditures for services at publicly supported family planning centers resulted in $7 billion in gross savings from helping women avoid unintended pregnancies and the births that follow.11 In other words, in the absence of the services provided by these centers with government support, the annual public costs of births from unintended pregnancy would increase 60%, to $18 billion.
Reductions in the public costs from the current level of $11.1 billion would translate to gross savings for the federal and state governments. Realizing those potential savings would require substantial public investments beyond those in place today. These should include continuing to increase access to family planning services and comprehensive sex education. Indeed, the Patient Protection and Affordable Care Act—the sweeping health care reform legislation that President Obama signed in March 2010—includes several major provisions to achieve those ends, such as broad expansions of public and private insurance coverage that will address reproductive health needs; new authority to states to expand Medicaid eligibility for family planning services specifically; the possibility (pending federal regulation) of required private insurance coverage of contraceptive services and supplies, free of any out-of-pocket costs; and new grants to states and community-based groups for programs that educate adolescents about both abstinence and contraception for the prevention of pregnancy and STDs.35–37
Our estimates are subject to a number of limitations, many of which are inherent to the array of sources we have drawn upon and have been discussed at length.9,21 Several others are important to highlight here.
For about half of the states, estimates for the average cost per Medicaid-funded birth were based on indices of Medicaid payment rates.9,11 These indices reflect relative costs across states for a broader set of services than maternity and infant care, and could therefore underestimate or overestimate costs in some states for those specific services.
Our method of attributing costs to state and federal governments has shortcomings. It does not reflect that states receive an enhanced federal reimbursement for pregnant women enrolled in CHIP, rather than Medicaid. Similarly, it does not reflect that the federal reimbursement for women covered by Medicaid only for labor and delivery, on an emergency basis (e.g., for undocumented immigrants), is at 50%, a rate that is for most states lower than their standard reimbursement rate. Both groups of women, however, are relatively small compared with the group for whom states receive reimbursement at their federal medical assistance percentage.
This analysis was limited to public costs for births resulting from unintended pregnancies. An estimate of the overall public costs of unintended pregnancies should also include some costs related to abortion and fetal loss, although such costs should be relatively small. The average cost of an abortion at 10 weeks’ gestation, for example, was $451 in 2009,38 which is much less than the $11,647 we estimate as the average cost of a Medicaid-funded birth in 2006. And according to one report, $89 million in public funds were spent for abortion nationally in 2006,39 substantially less than the $11.1 billion in public expenditures that year for births following unintended pregnancies.
The public costs related to births resulting from unintended pregnancies also, in theory, go far beyond maternity and infant care costs. Uncounted are costs from the increased likelihood of preterm birth, low birth weight and other negative perinatal outcomes;4 children’s medical care beyond their first year; pregnancy-related care paid for by other government-related health programs, including the Indian Health Service and indigent care programs that subsidize hospitals’ uncompensated care; and other government benefits, such as food stamps and welfare payments. In addition, because the income-eligibility thresholds for health and welfare benefits increase with family size, a birth from an unintended pregnancy may make family members eligible for additional benefits at additional public cost. A 2010 evaluation of a Medicaid family planning program in California, for example, found that the savings to the state from an averted publicly funded birth were nearly five times as high when analyses included both health and social services costs over a five-year period as when they considered only health care costs over a two-year period.40 The data that would be necessary to conduct such a broader analysis for all 50 states, however, are not available. Given the scope of our analysis and what was left out, it is safe to say that our estimates are conservative ones.
Clearly, the public costs of births following unintended pregnancies are substantial and place a burden on federal and state governments. For that reason, investments in programs and policies to reduce unintended pregnancies not only would enable women and families to meet their childbearing needs, but would produce public savings that would strengthen government finances and the sustainability of the nation’s health care safety-net programs.
Data Sources for the Proportion of Births Paid for by Public Programs
•Individual-level PRAMS data (28 states). We tabulated weighted estimates for 2006 from the CDC data set for 21 states: Alaska, Arkansas, Colorado, Georgia, Hawaii, Illinois, Maine, Maryland, Michigan, Minnesota, Mississippi, Nebraska, New Jersey, New York, Ohio, Oklahoma, Oregon, Rhode Island, Utah, Washington and West Virginia. PRAMS was conducted separately for New York City and for the rest of New York State; data from both surveys were combined to arrive at figures for the entire state. We tabulated CDC data from other years for seven states: Delaware (2007), Missouri (2007), Montana (2002), North Carolina (2007), North Dakota (2002), Wisconsin (2007) and Wyoming (2007).
For these 28 states, having access to the individual-level data allowed us to include separately identified CHIP programs, Medicaid and CHIP managed care plan names, and Medicaid and CHIP waiver programs. This can be complicated, because many states operate two or more such programs and contract with multiple managed care plans, and the list of programs and plans may change from year to year. These payment options were either listed on the PRAMS questionnaire within the Medicaid payment category, listed as a separate category or included in the “other” category as a write-in. In the following states, program names either were included in a state-specific category or were written in by at least 10 respondents and were therefore included in this analysis: Alaska (Denali KidCare), Arkansas (ARKids First), Colorado (Child Health Plan Plus), Michigan (Medical Outpatient Maternity Services), Nebraska (Medicaid managed care, including Wellness Option, Share Advantage and Primary Care Plus), New Jersey (New Jersey FamilyCare), New York (Prenatal Care Assistance Program), Ohio (CareSource), Oklahoma (SoonerCare), Rhode Island (RIte Care and Neighborhood Health Plan) and Wisconsin (BadgerCare).
•Other data that include state-specific programs (eight states). Weighted estimates of PRAMS data were obtained from the state health departments in Florida (2005), Louisiana (2007), South Carolina (2006), Tennessee (2007), Vermont (2006) and Virginia (2007). Additional program names identified as a separate category were included in the total estimates of publicly funded births in Florida (Medipass), Tennessee (TennCare) and Vermont (Dr. Dynasaur). State health department analyses confirmed that for all six of these states, fewer than 10 relevant write-in responses were received for each survey. Aggregate data from Massachusetts (2007) were obtained from CDC’s PRAMS On-line Data for Epidemiologic Research (CPONDER) system,41 after we confirmed with the state health department that no state-specific programs were excluded from the Medicaid category and that fewer than 10 observations had write-in answers that included public funding. We produced weighted tabulations of data from Kentucky’s 2007 PRAMS Pilot Project survey.
•Aggregate data that may not include state-specific programs (seven states). For Alabama (2003) and Pennsylvania (2007), we were limited to data available in CPONDER. Alabama’s questionnaire included no state-specific categories. In Pennsylvania, two state-specific categories were likely not included with Medicaid-funded births in the CPONDER tabulations: adult basic and CHIP. For New Mexico (2006) and Texas (2006), estimates were obtained from state health department tabulations. We obtained tabulations from PRAMS-like surveys in Idaho (2006 Pregnancy Risk Assessment Tracking System) and Iowa (2006 Barriers to Prenatal Care survey). The surveys in New Mexico, Texas, Idaho and Iowa had no state-specific programs, and although the numbers of write-in responses are unknown, the Medicaid category likely captured almost all publicly funded deliveries in those states. Finally, for California, we calculated the proportion of births that were publicly funded using published data from that state’s 2006 Maternal and Infant Health Assessment.42
•Predicted data from multivariate regression (eight jurisdictions). We used regression analysis to obtain estimates for Arizona, Connecticut, the District of Columbia, Indiana, Kansas, Nevada, New Hampshire and South Dakota. The proportion of all births paid for by public programs was estimated also for Iowa, because data were not available at the time of this analysis. Standard errors for the nine predicted values of the proportion of all births that were publicly funded and for the eight predicted values of the proportion of intended births that were publicly funded ranged from 0.01 to 0.04, except for the District of Columbia (0.10), which is somewhat unlikely to conform to a model in which all the other observations are states, as opposed to cities. Standard errors for the eight predicted values of the proportion of unintended births that were publicly funded ranged from 0.02 to 0.05 (0.11 for the District of Columbia).
To gauge the accuracy of the model, we used the regression coefficients to calculate predicted proportions for the 43 states for which we had data, and then compared the model’s predictions to the actual data. For each state, we ran a regression using the data from the other 42 states to predict the proportion of publicly funded births. We repeated that procedure for all three measures. In 22 states, the estimates were within five points or less of the actual proportion. The 21 with larger differences were about equally divided into those estimates below and above the actual proportion; both groups had an average discrepancy of roughly nine points. These results provide confidence that the regression procedure was unbiased—that is, it did not lead to consistently higher or lower estimates than the actual.