Key Points

• In 2014, the majority of abortion patients (60%) were in their 20s, and the second-largest age-group was in their 30s (25%). 

• The proportion of abortion patients who were adolescents declined 32% between 2008 and 2014.

• No racial or ethnic group made up the majority of abortion patients: Thirty-nine percent were white, 28% were black, 25% were Hispanic, 6% were Asian or Pacific Islander, and 3% were of some other race or ethnicity.

• Fifty-nine percent of abortion patients in 2014 had had at least one previous birth.

• In 2014, three-fourths of abortion patients were low income—49% living at less than the federal poverty level, and 26% living at 100–199% of the poverty level.

• The vast majority of abortion patients (94%) identified as straight or heterosexual. Four percent identified as bisexual; fewer than 1% as lesbian, gay or homosexual; and 1% as something other than straight, gay or bisexual. 

• Many abortion patients reported a religious affiliation—24% were Catholic, 17% were mainline Protestant, 13% were evangelical Protestant and 8% identified with some other religion. Thirty-eight percent of patients had no religious affiliation. 

• Abortion patients were less likely to have no health insurance coverage in 2014 than in 2008 (28% vs. 34%), likely because of the Affordable Care Act. Thirty-five percent of patients had Medicaid coverage, 31% had private insurance and 3% each had either insurance through HealthCare.gov or a different type of insurance.

• The majority of patients (53%) paid for their abortion out of pocket; Medicaid was the second-most-common method of payment, used by 24% of patients. 

Background

Abortion is common in the United States and is a critical component of comprehensive reproductive health care.1 However, information about individuals who have abortions is limited. For example, population-based surveys, which are used to obtain information about many aspects of reproductive and sexual health, do not adequately measure the prevalence of abortion, and only about half of abortions provided in the United States are captured by these types of surveys.2,3 While the Centers for Disease Control and Prevention (CDC) publishes annual abortion statistics, including selected demographic characteristics of abortion patients, this information is limited and incomplete,4 as it is collected from individual state health departments with variable abortion-reporting requirements.5 For example, the CDC does not report abortion data from California, New Hampshire or Rhode Island, and the accuracy of abortion information can vary substantially by state over time. To address these limitations, the Guttmacher Institute periodically collects information from U.S. abortion patients, and the results of the most recent survey are summarized in this report. 

The previous Abortion Patient Survey was conducted in 2008, and the landscape of reproductive health in the United States has changed in several important ways since that time. In January 2013, the Affordable Care Act was fully implemented, and it has reduced the number of women of reproductive age who are uninsured, mainly by increases in Medicaid coverage.6 Thus, the Act may have improved access to general health care for all women, and for low-income individuals in particular.7 However, access to abortion care since 2008 may have decreased. Between 2009 and 2014, states enacted 288 new abortion restrictions.8 The abortion rate declined 13% between 2008 and 2011, and while there is little evidence to suggest that the state abortion regulations passed during that time period were responsible for the drop in abortions,9 restrictions passed in some states in more recent years have been particularly onerous. These include waiting periods that may require patients to visit the clinic twice, requirements that abortion clinics meet the standards of ambulatory surgical centers or acquire hospital admitting privileges for their clinicians, and bans on the use of private insurance and plans purchased through state exchanges to pay for abortion services.10 

Finally, while the most recent recession officially ended in June 2009, the recovery has been particularly slow; for example, median family incomes and earning have not increased since that time.11 The increase in abortion restrictions, coupled with a sluggish economy, may have reduced access to abortion services, as well as altered the population of patients who are able to obtain them. These trends may have also motivated more individuals to attempt to self-induce an abortion outside of a clinical setting.12,13 Information in this report can help identify those groups that are most likely to be affected by restrictions, and can help policymakers and health care providers determine which groups of women, and at which point in their lives, may need greater access to assistance in preventing unintended pregnancies.

Data Collection and Analytic Strategy

Data in this report come from the Guttmacher Institute’s fifth national survey of abortion patients, which used a sampling design, questionnaire and fieldwork protocol similar to those used in previous iterations conducted in 1987, 1994–1995, 2000–2001 and 2008.14–17 A detailed description of the data collection procedures and copies of the survey instruments can be found in the appendices of this report; we provide a brief summary below.

Between April 2014 and June 2015, we collected information from 8,380 respondents obtaining abortions at 87 facilities. We used a four-page, self-administered questionnaire available in English and Spanish. Two versions were developed, and respondents were randomly provided with Module A or Module B; the modules were identical for all questions, with the exception of three items on current school enrollment, prior abortions and pregnancy timing. Participating facilities provided a total of 11,024 abortions during the survey period, yielding a response rate of 76%. Facilities eligible for participation included clinics and physicians’ offices that provided at least 30 abortions in 2011 (according to the Guttmacher Institute’s 2011 Abortion Provider Census9); hospitals were excluded from the survey because of past recruitment and logistical challenges. In 2011, hospitals accounted for 4% of all abortions,9  and it is unlikely that their exclusion biased the sample. 

Key demographic characteristics of abortion patients include age, relationship status, race and ethnicity, nativity, educational attainment, number of prior births, family income level, religious affiliation, prior attempts to self-induce an abortion, health insurance coverage and method of payment for abortion services. One new measure—sexual orientation—was included in the 2014 survey. 

In this report we present descriptive statistics on abortion patients’ demographic characteristics, and compare these characteristics with those of patients in 2008. Percentage distributions for patients obtaining abortions in 2008 have been recalculated from the previously published report to exclude hospital respondents; these patients made up 4% of the 2008 sample, and their demographic profile is very similar to that of all patients.17 We used bivariate logistic regression analysis to assess whether patient characteristics were significantly different between the two surveys. 

Weights were constructed to account for patient nonresponse and variation from the original facility sampling plan. Missing information for key demographic variables was imputed using the answers of respondents with similar characteristics. All analyses were based on weighted data and were conducted using the svy command in Stata version 13.1 to account for the complex sampling design.

Because 2011 is the most recent year for which the total number of abortions in the United States is available, we were unable to estimate the abortion rate (the number of abortions per 1,000 women) by subgroup for this report. Instead, we constructed an abortion index (or relative abortion rate) as a proxy measure of rates to assess the relative levels of abortion across subgroups. Each abortion index is the proportion of abortion patients in a given subgroup (e.g., a particular age-group) relative to the proportion of all U.S. women aged 15–44 who are in that same subgroup. If these proportions are the same—indicated by an index of 1.0—the subgroup’s relative abortion rate is the same as the overall national rate. If the subgroup is overrepresented among abortion patients (index greater than 1.0), its relative abortion rate is above average; if it is underrepresented (index less than 1.0), its relative rate is below average. Notably, an increase in the abortion index for a subgroup over time does not necessarily indicate an increase in the subgroup’s abortion rate. This may be the case if the subgroup’s abortion rate decreased at a slower rate than that of the total population. 

Results

Characteristics of Women Obtaining Abortions

Age-Group

In 2014, the largest proportion of abortion patients were in their 20s (60%), followed by those in their 30s (25%—Table 1). Adolescents—those younger than 20—accounted for 12% of abortion patients, and fewer than 4% were younger than 18.

The proportion of abortions accounted for by adolescents declined significantly between 2008 and 2014—by 32%. In particular, the proportion accounted for by 15–17-year-olds declined 44% over this period, and that among 18–19-year-olds dropped by 25%. The 2014 abortion index of 0.4 for the former group indicates that they were substantially underrepresented among abortion patients relative to their representation in the larger population of women. 

In contrast, women aged 20–24 were overrepresented by a factor of almost two, having the highest relative abortion rate of the age-groups examined (1.9). Abortion indices declined with increasing age thereafter. Both younger and older adolescents had slightly lower abortion indices in 2014 than in 2008; there was little change in the indices for women aged 20 or older. 

Relationship Status

Relationship status can be a proxy for exposure to sexual activity, and can also influence individuals’ and couples’ childbearing goals. The distribution of abortion patients and abortion indices varied by relationship status. About 14% of abortion patients were married, and an additional 31% were cohabiting. A slight majority were not living with a partner in the month they became pregnant (46% had never married and 9% had been previously married).  

The 2014 abortion index of 0.4 for married patients indicates that they were substantially underrepresented among abortion patients relative to all women of reproductive age. Cohabiting women were overrepresented by a factor of 2.1, meaning they had an abortion rate twice the national average. The abortion index for never-married, noncohabiting patients was slightly higher than average (1.2). The abortion indices for most relationship statuses remained unchanged from 2008, with the exception of that for cohabiting women, which declined from 2.6 to 2.1. 

Race and Ethnicity

Disparities in reproductive health outcomes by race and ethnicity are well documented,18–20 and may be an important indicator of systemic barriers to preventive services. No racial or ethnic group made up the majority of abortion patients in 2014. Overall, 39% were white, 28% black, 25% Hispanic, 6% Asian or Pacific Islander, and 3% of other background. The racial and ethnic composition of patients was quite similar in 2008. 

White women were slightly underrepresented among abortion patients in 2014, having an abortion index of 0.7, while black women were substantially overrepresented, with a relative abortion rate of 1.9. Hispanic women were slightly overrepresented among abortion patients in 2014 (1.2).* The abortion index for Hispanics declined slightly, from 1.5 in 2008, and there was little change in the indices for blacks and whites.

Nativity

The overwhelming majority of abortion patients in 2014 were born in the United States (84%), while the remaining 16% were born elsewhere; these proportions had remained stable since 2008. Of those patients born outside the United States, about half were Hispanic, 20% Asian, 16% black and 12% white (not shown). 

The relative abortion rates for these two groups were proportional to the overall population of women of reproductive age in the United States and were comparable in both years. 

Educational Attainment

Educational goals are often cited as a reason to delay childbearing, as many individuals wish to complete their schooling and better position themselves economically before having children.21 In 2014, some 9% of abortion patients aged 20 or older had less than a high school degree, and the overwhelming majority—91%—had graduated from high school; more than one in five had a college degree. The proportion of patients aged 20 or older who had not graduated from high school declined significantly over the six-year period (from 12% to 9%). 

In 2014, some 24% of all abortion patients were currently attending school, including 72% of minors and 53% of 18–19-year-olds (not shown). Only 14% of those currently in school had not graduated from high school; 66% had some college or a college degree, suggesting that most abortion patients who were students were pursuing postsecondary degrees. 

Abortion patients aged 20 or older with only a high school education were slightly overrepresented relative to the population of all women aged 15–44 (abortion index of 1.3), while college graduates were underrepresented (0.7). Between 2008 and 2014, the relative abortion rate for women 20 or older declined slightly among those who had not graduated from high school (from 1.1 to 0.9); rates for women in the other educational categories changed little or not at all. 

Prior Births

How individuals achieve their desired family size—including the timing and spacing of any births—is often part of a complicated calculus, and decisions regarding pregnancy outcomes are made in the context of existing and planned children.22,23 In 2014, it continued to be the case that the majority of abortion patients (59%) had had at least one previous birth, including one-third who had had two or more; 41% of abortion patients had had no prior births. These proportions were largely unchanged from 2008.

Individuals with one previous birth were overrepresented among abortion patients (index of 1.5), while those with either no prior births or at least two were slightly underrepresented (0.9 and 0.8, respectively). Abortion indices by number of previous births showed virtually no change between 2008 and 2014.

Income Level

Over the last few decades, abortion and unintended pregnancy have become increasingly concentrated among poor patients.17 This trend continued through 2014, when there was a significant increase in the proportion of abortion patients accounted for by this group: Forty-nine percent of patients had family incomes of less than 100% of the federal poverty level, while 42% were in this group in 2008. An additional 26% of patients in 2014 had incomes that were 100–199% of the poverty threshold. (We refer to patients in the lowest and middle categories as poor and low income, respectively.) The increase in poor abortion patients was countered by a decrease in the proportion of patients in the highest income group (200% or more of the federal poverty level), from 31% to 25% over the six-year period. 

Poor women were substantially overrepresented among abortion patients in 2008 and 2014, and had the highest abortion index of all subgroups examined in the latter year (2.5). Low-income women had an above-average relative abortion rate (1.3), while those in the highest income group were substantially underrepresented compared with the general population (0.4). The increase in poverty among abortion patients somewhat mirrored that of all women of reproductive age over the study period. 

Religious Affiliation    

The majority of abortion patients indicated a religious affiliation: Seventeen percent identified as mainline Protestant, 13% as evangelical Protestant and 24% as Roman Catholic, while 8% identified with some other religion. Thirty-eight percent of patients did not identify with any religion. The proportion of women who identified as mainline Protestant declined by 24% since 2008, whereas the proportion with no affiliation increased by 38%. The proportion identifying as Catholic decreased by 15% from the earlier survey, though this change was only marginally significant. 

The abortion index for Catholic women showed that their relative abortion rate was nearly the same as that for all women (1.1). Mainline Protestants were slightly underrepresented among abortion patients (0.8), while evangelical Protestants had an abortion rate that was half of the national average. Patients with no affiliation were overrepresented among abortion patients, having a relative abortion rate of 1.8. The abortion index had declined slightly for mainline Protestants, and had increased slightly for those with no affiliation. 

Sexual Orientation

Unintended pregnancy is not limited to heterosexual women. Sexual minority women may have an elevated risk for unintended pregnancy because of differences in sexual health knowledge or behaviors, or because of a higher prevalence of risk factors such as previous exposure to abuse.24,25 The vast majority of abortion patients identified as heterosexual or straight (94%). Four percent of patients said they were bisexual, while only 1% identified as “something else” and 0.3% as homosexual, gay or lesbian. Respondents who indicated “something else” could write in a more specific response; 12 of the 81 who answered affirmatively indicated “pansexual,” which was the only response provided by more than one respondent. 

Prior Attempts to Self-Induce Abortion 

Although abortion is legal in the United States, some individuals still obtain, or attempt to obtain, abortions outside of a clinical setting. For example, in 2008, some 1.2% of patients who accessed clinical abortion services reported that they had ever used misoprostol to try to end a pregnancy on their own, and an additional 1.4% had attempted to do so using some other substance (e.g., herbs).26 A study of 1,425 women accessing health care services found that a slightly higher proportion—4.6%—had ever attempted to self-induce an abortion using misoprostol or other substances.27 While these figures suggest that self-induced abortion was relatively uncommon, media reports indicate that this practice may have increased in recent years, particularly in restrictive states.12 For example, one study estimated that more than 100,000 women living in Texas had ever attempted to self-induce an abortion,28 and some may have been motivated to do so because of difficulties in accessing clinical abortion services.13  

In the 2014 survey, 1.3% of abortion patients reported that they had ever taken misoprostol to try to bring back their period or end a pregnancy (Table 2). This figure is comparable to the 1.2% who reported such an attempt in 2008. However, while the proportion of abortion patients reporting this behavior changed little, the practice may have become more dispersed. In the 2008 survey, the 99 patients who reported ever having taken misoprostol to self-induce an abortion were obtaining clinical abortion services at 47 facilities and resided in 23 states and Mexico. In the 2014 survey, the 108 patients who reported such attempts were accessing abortion care at 54 facilities and resided in 32 states and Mexico. 

The proportion of patients who reported using substances other than misoprostol to attempt to self-induce an abortion declined from 1.4% in 2008 to 0.9% in 2014, representing a small but significant decrease. In both years, the most common write-in responses for types of substances used were vitamin C, herbs and herbal teas. 

Health Insurance Coverage and Payment for Abortion Services

The implementation of the Affordable Care Act resulted in fewer uninsured patients in 2014, mostly because of increases in Medicaid coverage.6 However, insurance coverage for abortion, unlike many other types of health care services, is subject to stricter regulation and scrutiny. The Hyde Amendment bans abortion coverage through federal Medicaid except in cases of rape, incest or life endangerment. Moreover, in 2014–2015, some 25 states had laws essentially banning abortion coverage in plans offered through the health insurance marketplaces, including 10 that banned such coverage more broadly in all private insurance plans regulated by the state. Finally, a number of private plans, as well as plans covering all federal and many state employees, exclude abortion coverage. 

While federal Medicaid dollars can be used to pay for abortions only under very limited circumstances, 15 states allow state funds to cover all or most medically necessary procedures for patients with Medicaid coverage, including states with large populations, such as California and New York. All but two of these states (Alaska and Montana) expanded Medicaid eligibility under the Affordable Care Act, and it is possible that Medicaid now plays a more prominent role in abortion care in these states than in 2008. Thus, we also examined patterns in insurance coverage and payment for abortion services according to whether patients lived in a state where state funds are used to pay for abortion services; such states are hereafter referred to as Medicaid-coverage states, while their counterparts are non–Medicaid-coverage states.

In 2014, some 35% of abortion patients reported that they had Medicaid insurance coverage, and 31% had private insurance (Table 3). Starting in January 2013, women had the option of obtaining health insurance through HealthCare.gov or their state’s health insurance exchange, and 3% of abortion patients indicated that they had done so.§ A similar proportion said they had some other type of insurance. None of the changes in type of insurance coverage between 2008 and 2014 were sizeable, but the decline in the proportion of abortion patients who were uninsured—from 34% to 28%—was statistically significant. 

While fewer abortion patients were uninsured in 2014 than in 2008, there were no significant changes in how patients paid for their abortions. Regardless of insurance coverage, 53% of patients reported that they paid for the abortion themselves. Medicaid was the second-most-common method of payment, reported by 24% of patients; the overwhelming majority of these patients (96%) lived in the 15 states that allow state funds to be used to pay for abortions (not shown).** Fifteen percent of patients reported that they used their private insurance to pay for the procedure, and 14% relied on some type of financial assistance. Notably, most patients with private health insurance (61%) paid out of pocket for their abortion (not shown). Eight percent relied on more than one payment method, most commonly paying themselves and getting financial assistance. While there were shifts in type of payment between 2008 and 2014—in particular, a decrease in the proportion who were self-paying and an increase in reliance on Medicaid—the changes were not statistically significant. 

Patterns of change in type of insurance and how patients paid for the procedure differed according to whether the patient lived in a Medicaid-coverage state. In both types of states, the proportion of patients who were uninsured declined significantly between 2008 and 2014. For patients in Medicaid-coverage states, this was due to a significant increase from 43% to 51% in the proportion with Medicaid coverage. In non–Medicaid-coverage states, 4% of abortion patients obtained coverage through HealthCare.gov or a state exchange in 2014, so it is unclear if these individuals were covered by private insurance or Medicaid.

In Medicaid-coverage states, there was a significant decrease in the proportion of patients paying out of pocket (from 36% to 24%), likely because of the significant increase in the proportion using Medicaid to pay for the procedure (from 44% to 52%). It is worth noting that in states that used their own funds to pay for abortion care in 2014, some 89% of patients with Medicaid coverage made use of this method of payment (not shown). The proportion of individuals receiving financial assistance to pay for care saw a small but significant increase (from 2% to 4%). Among patients in non–Medicaid-coverage states, the only significant change in payment type between surveys was for those using private insurance (from 9% to 11%). 

Finally, other differences were observed in how patients paid for their abortions. Those in noncoverage states were much more likely to pay for abortions out of pocket than were patients in Medicaid-coverage states (75% vs. 24%, p<.001), and were more likely to rely on financial assistance (22% vs. 4%, p<.001). Patterns in use of private insurance also differed. Patients in non–Medicaid-coverage states were more likely than those in other states to have private insurance (36% vs. 26%, p<.001), but they were still less likely to use it to pay for the procedure (11% vs. 19%, p<.001).

Discussion

In many ways, abortion patients in 2014 were quite similar to those in 2008. As in the earlier survey, the majority of patients were in their 20s, unmarried and nonwhite, and had graduated high school, had at least one previous birth and had a religious affiliation. However, smaller proportions of patients in 2014 were adolescents and were uninsured, and a larger proportion were poor. 

The percentage of abortion patients accounted for by adolescents has been declining for decades,29 but the 32% drop between 2008 and 2014 was particularly notable. A comparable drop was seen in the teenage birthrate, which declined 40% during this period,30,31 meaning that fewer teenagers were getting pregnant in 2014 than in 2008. There were no significant changes in sexual activity or contraceptive use patterns among adolescents during this time period,32 and economists speculate that increased educational opportunities, the media and the economy may have influenced these trends.33 Understanding the reasons behind these declines could have important policy implications, and more research is needed to better understand the range of factors influencing these patterns.

Poor women continue to account for a disproportionate share of abortion patients, and this representation increased from 42% to 49% over the six-year period, mostly driven by an increase in the population of women of reproductive age who are poor. The abortion index for poor women changed little, and disparities in abortion rates by income did not increase between 2008 and 2014. Still, it is now the case that 75% of abortion patients are low income, having family incomes of less than 200% of the federal poverty level. 

The increased representation of poor women among abortion patients is, perhaps, more surprising when placed in the context of increased abortion restrictions. Between 2009 and 2014, some 288 abortion restrictions were enacted in 31 states.8 Many of these regulations, such as hospital admitting privileges and ambulatory surgery center requirements, have had the effect of closing clinics,10,34–36 while others, such as gestational age limits and medication abortion restrictions, limited the services that patients could access.10,37 Poor and low income individuals are disproportionately affected when these types of restrictions are passed; such restrictions can increase delays to and costs of abortion care, including by necessitating additional travel for patients to access services. This study was unable to assess how many individuals were prevented from obtaining abortions because of economic or other barriers, but if these restrictions had not been enacted, the proportion of poor patients able to access these services might have increased even more. 

While it is still the case that a majority of abortion patients reported a religious affiliation, the proportion who did not identify with any religion increased substantially. This pattern was also seen among the larger population of women of reproductive age, though the change was less pronounced. This trend was largely driven by fewer patients identifying as mainline Protestants. 

No racial or ethnic group made up the majority of abortion patients. White patients accounted for the largest proportion (39%), and black and Hispanic patients accounted for similar proportions (28% and 25%, respectively). Relative abortion rates declined slightly for Hispanic patients between 2008 and 2014, which may be associated with use of long-acting reversible contraceptive (LARC) methods among this group; between 2009 and 2012, Hispanic women saw the most significant increase in LARC use, from 9% to 15%.38 Black women have the lowest rates of LARC use, which may reflect barriers to contraceptive care, or a mistrust in provider-controlled methods.38 However, LARC use alone cannot fully explain the discrepancies in relative abortion rates by race and ethnicity, and future research might explore other potential factors contributing to this persistent pattern. 

The vast majority of abortion patients identified as heterosexual, but a nonnegligible proportion identified as a sexual minority, including gay or lesbian, bisexual or “something else.” To the extent that sexual behavior matches sexual identity, it stands to reason that the majority of nonheterosexual patients identified as bisexual (and not as lesbian, for example), as they are more likely to have had recent sex with a man. The fact that some respondents identified themselves as “something else” rather than gay or lesbian provides evidence that our current understanding and measurement of sexual orientation are imprecise and will continue to evolve. For example, the most common write-in answer provided for this item was “pansexual,” a term that is often defined as a sexual or romantic attraction to people of any sexual or gender identity. Pansexual and other umbrella terms, such as queer, suggest that some of the mainstream terms used to define sexual orientation may be too narrow in their traditional understanding, especially when considering that some people conceptualize their sexual identity as being fluid. Information on sexual orientation among abortion patients may help to inform our understanding of unintended pregnancy risk among sexual minority women as the field continues to refine measurements and documentation of sexual activity, health behaviors and pregnancy among these populations. 

Media reports and at least one research study have suggested that, as states impose more restrictions on clinical abortion services, more individuals are attempting to self-induce abortions using misoprostol and other substances.12,13 We did not find an increase in reliance on misoprostol among those who were able to obtain clinical abortion services, and there was actually a decline in the proportion of patients who reported using other substances. One major shortcoming of our study regarding assessing trends in self-induced abortion is that many individuals who were able to successfully end their pregnancies on their own were not captured in our survey because they would have no need for clinical services. Still, if use of misoprostol (and other substances) to self-induce abortion is actually increasing, we might also expect to see an increase in ever-use of this method among patients who relied on clinical abortion services. For example, in the 2008 study, two-thirds of patients who had ever attempted to self-induce using misoprostol reported that they had done so for the current pregnancy, and were presumably presenting to the clinic because the method had not worked.26 

The Affordable Care Act is likely responsible for the decline between 2008 and 2014 in the proportion of abortion patients who did not have health insurance. While more abortion patients were covered by Medicaid in the more recent survey, this increase was limited to states where state funds are used to pay for abortions. This pattern likely reflects the fact that all but two of the 15 states adopted the Medicaid expansion program offered under the Act.  

While most abortion patients had health insurance coverage, it was still the case that a majority paid for their abortion care out of pocket. The second-most-common method of payment was Medicaid, and nearly all of these procedures were to patients in Medicaid-coverage states. In fact, 52% of patients in these states used Medicaid to pay for their abortions, demonstrating the importance of this program for women’s reproductive health. Nationally, poor and low-income patients and those living in states that did not expand Medicaid are still substantially less likely to have health insurance.39 This is especially notable since abortion is increasingly concentrated among poor women. The inability to use health insurance for abortion services represents a significant impediment to women’s health and well-being, and disproportionately impacts poor women and women of color, who are more likely to rely on Medicaid.40,41

While half of abortion patients in Medicaid-coverage states relied on this program to pay for abortion care, patients in these states were also more likely to rely on private health insurance to pay for their procedures than were patients in noncoverage states—despite the fact that patients in the latter states were more likely to have private insurance coverage. These patterns could be due to several circumstances, including that some of these noncoverage states had laws prohibiting private plans, or at least those purchased through the state exchange, from covering abortion services. Additionally, employers in these states may be more likely to exclude abortion coverage from their plans. Finally, because abortion providers in some noncoverage states face a large number of regulations, they may be unwilling to accept private insurance or lack the resources to do so. 

Limitations

This study has several limitations. We excluded individuals obtaining abortions at hospitals and, therefore, the data are not representative of all abortion patients. Yet these types of facilities accounted for only 4% of all abortions in 2011, and the demographic profile of the 2008 sample that excludes hospital patients is very similar to the one that includes them. Thus, it is unlikely that including these populations would have substantially altered the findings.

Prior studies have documented that self-reported health insurance status is prone to reporting error,42 and this may be even more pronounced as individuals adjust to the Affordable Care Act. Indeed, that 3% of abortion patients indicated they obtained coverage through HealthCare.gov or a state exchange could be seen as evidence of this confusion. Our measure of coverage obtained from these sources did not include a follow-up item to assess what type of plan, and coverage under private plans or Medicaid plans may be understated (e.g., if most individuals who procured coverage from these sources obtained private insurance, coverage levels would be even higher). Nonetheless, patterns in coverage and payment corresponded with state patterns in Medicaid coverage of abortion and participation in the Medicaid expansion program. Thus, while subject to more reporting error than characteristics such as age and race, we expect that patterns of insurance coverage and payment for abortion services, as well as changes related to these characteristics, are real. 

Conclusions

A better understanding of the characteristics of abortion patients can be used to address the structural inequalities that exist within the U.S. health care system. The characteristics of abortion patients presented in this report may reflect which groups of individuals are better able to access reproductive health information and services, and can be used to inform public health policies aimed at decreasing these disparities. For example, the onslaught of increased abortion restrictions between 2009 and 2014 likely disproportionately affects poor and low-income women, black women and young adults, as these populations are overrepresented among abortion patients.

The ability to obtain and use health insurance to cover abortion care represents an important means to reducing systematic inequities that drive disparities in care. Though fewer individuals, including abortion patients, were uninsured in 2014, the only significant change in how patients paid for abortion services between 2008 and 2014 was seen in states that provide state Medicaid funds for abortion. Medicaid is the largest source of funding for medical services for poor and low-income individuals in the United States. To reduce barriers to abortion care, all states need to participate in the Medicaid expansion program, laws restricting the use of insurance coverage of abortion services must be prevented or repealed, and—most critically—the Hyde Amendment must be struck down. Abortion services are an integral part of reproductive health care, and they should be covered by health insurance without exception. 

Appendix 1: Methods

Data Collection

The 2014 Abortion Patient Survey is the fifth in a series by the Guttmacher Institute, and uses a design and survey instrument similar to those used in the 1987, 1994–1995, 2000–2001 and 2008 surveys. The questionnaire and procedures were approved by the Guttmacher Institute’s federally registered institutional review board.

As in previous years, the 2014 survey collected information directly from abortion patients using a four-page, paper-and-pencil, self-administered questionnaire, available in English and Spanish. Questionnaires contained many of the same questions as in previous years, including items to collect demographic information on age, relationship status, race and ethnicity, nativity, educational attainment, prior births and abortions, family income and religious affiliation, as well as indicators such as gestational age, contraceptive use and exposure to disruptive events in the past year. Several new items were introduced in the 2014 survey, among them measures of sexual orientation, reason(s) the patient chose that particular facility for the procedure and how long ago the patient had made the appointment. 

To validate three items on the survey, two versions of the instrument were developed and a split-sample method was employed, wherein respondents were randomly provided with Module A or Module B (see Appendix 2, page 19). The two modules were identical with the exception of three questions, which were worded differently to assess school enrollment, prior abortions and pregnancy timing. These items will be examined in subsequent analyses. 

Participating facilities were sampled from the universe of all known abortion-providing facilities as of 2011, according to information obtained from the Guttmacher Institute’s 2011 Abortion Provider Census,9 and excluding hospitals and those facilities that provided fewer than 30 abortions in 2011. Hospitals were excluded because of the logistical difficulties with recruitment (e.g., in past surveys we often had to obtain approval from several administrative authorities at each hospital), and facilities with small caseloads were excluded because of the high likelihood that they would not provide any abortions during the survey period. It is unlikely that the omission of these facilities introduced bias, because combined they accounted for only 4% of all abortions in 2011.

The universe was stratified by facilities’ 2011 annual caseload of abortions (30–399; 400–1,999; 2,000–4,999; and 5,000 or more), and by whether they were affiliated with national organizations for women’s reproductive health; the latter attribute was used to ensure broad representation of facility types, sizes and organizational affiliations without overrepresenting facilities that operated within national networks. Within each stratum, facilities were organized by census region and state. Next, we systematically sampled facilities from each stratum by selecting them at specified intervals within the list; the interval varied by stratum. Facilities with the largest caseloads were oversampled to ensure a diverse representation of facility types within the sample.

Selected facilities were then recruited and assigned to a survey period that was inversely proportional to the probability of being selected, ranging from two weeks for the largest facilities to 12 weeks for the smallest. During this period, facility contacts were asked to administer the questionnaire to all patients obtaining an abortion on the day of their procedure; in the case of multiday procedures, questionnaires were to be administered on the first day. Our goal was to recruit 113 facilities; the final sample was obtained from patients at 87 facilities (77% of the original goal). An additional 123 facilities were approached, but did not participate. Twenty of these facilities were found to no longer be providing services; 25 agreed to participate, but were unable to adhere to the study protocols or recruit a sufficient number of patients into the study; and 78 declined to participate. In all, 87 of the 190 active facilities (46%) approached agreed to participate.†† Common reasons that facility contacts gave for declining to participate included limited staffing resources, a belief that their facility would not see any abortion patients during the study period, a belief that their patients would not be interested in participating and the research team’s inability to reach the appropriate staff member at the facility to discuss the study. Each of these facilities was ultimately replaced with the next facility within their stratum. The most difficult providers to recruit and retain in the study were those that were not affiliated with a national organization and that provided 30–399 abortions each year (the smallest caseload); they often had small offices with limited staff. In some cases where facilities indicated that participation would be difficult given limited staffing and other constraints, small stipends—in the form of gift cards or as contributions to the facilities’ patient funds—were offered as a token of appreciation. 

The questionnaire was distributed to patients during their clinic visit at a time facility staff determined was most appropriate. Communications with facility staff suggest that questionnaires were typically distributed with office or clinic intake forms; patients most often completed the questionnaire along with their other paperwork while they waited for their procedure. With the exception of currently incarcerated individuals, all patients obtaining abortion services at participating facilities were eligible to participate. The questionnaire was clearly identified as separate from the office or clinic forms, and the introductory language on the front of the survey described the purpose of the study, indicated to patients that the questionnaire was voluntary and anonymous, and served as implied consent. All respondents were provided with an envelope in which to place the survey before they sealed and returned it to facility staff. At the end of each week, staff compiled all collected, sealed questionnaires and mailed them back to the Institute. Returned survey packets also included information about the total number of abortion patients seen that week, so that we could calculate response rates. 

To be included in the final sample, participating facilities needed to obtain usable questionnaires from at least 50% of abortion patients seen during their survey period to ensure the representativeness of patients within that facility. In 14 instances, facilities that were close to achieving a 50% response rate from their patients were allowed to field for additional time, and were then weighted accordingly to account for their longer fielding period. 

During fielding periods, participating facilities reported providing a total of 11,024 abortions. Usable surveys were returned from 8,380 patients, for a response rate of 76%. For patients who refused or were unable to participate, an employee of the facility was asked to complete and return a small portion of the questionnaire covering the patient’s basic demographic characteristics (age, race and ethnicity, and insurance coverage). This enabled us to assess whether patients who were missed differed from patients who completed the survey. Facility staff returned 1,066 surveys with basic demographic information; no information was available for the remaining 1,578 patients who declined to participate.

As in prior surveys, we employed a three-stage weighting process to correct for any bias produced by deviation from the original sampling plan and patient nonresponse. First, individual weights were developed to adjust for the demographic characteristics of the 1,066 nonrespondents for whom facility staff provided information. Second, facility-level weights adjusted for the other 1,578 nonrespondents for whom no demographic data were available. Third, stratum weights were constructed to correct for departures from the number of facilities to be sampled in each grouping by caseload and provider type. With the final weight adjusted to a mean of 1.0, the standard deviation is 0.24 and the range is 0.5–2.0.

Some questionnaires were returned partially completed. Nonresponse on specific items was 1–2% for most nonsensitive questions, but ranged from 5% (for any prior abortions) to 13% (for family income) on more sensitive items. Missing information on core demographic items was imputed using “hot-deck” single imputation. This method identifies variables most strongly associated with each item requiring imputation, and sorts the data file accordingly to replace the missing value with that from a similar, adjacent case. 

Data Quality and Comparability

For purposes of comparability, the majority of survey items were maintained from previous abortion patient surveys conducted by the Guttmacher Institute. However, some items were updated or added, and are discussed below.

Race and Ethnicity

To measure race and ethnicity, the 2014 Abortion Patient Survey adopted items from the 2013 Current Population Survey. Respondents were first asked “Are you Spanish, Hispanic or Latina?” and could answer yes or no (question 2). This was followed by “Please choose one or more races that you consider yourself to be.” Six response categories were available: “American Indian or Alaska Native,” “Asian,” “Native Hawaiian or Pacific Islander,” “black or African American,” “white” and “other.” As in our 2008 Abortion Patient Survey (but unlike in the Current Population Survey), the last response category provided a space to write in a specific race. The 2014 survey differed from the 2008 one in that it allowed patients to choose more than one racial group; the prior survey asked patients to choose the racial group that best described them. To make the surveys comparable on this key characteristic, the current analysis used a measure of race and ethnicity that was comparable to that of the 2008 survey; even though the earlier survey had encouraged respondents to provide only one race, 118 respondents indicated that they identified with multiple groups (e.g., checked more than one race). In keeping with prior surveys, we constructed a measure of race in which patients who indicated multiple races were typically classified as belonging to the least common of the racial groups checked off. Hence, respondents were classified as a specific racial group besides “other” when possible (e.g., a written response of “Chinese” was coded to “Asian”). Also in line with prior surveys, patients who checked off both “black” and one or more other racial groups were classified as black. In the combined measure used in this analysis, Hispanic ethnicity was given priority over any racial category. (Most commonly, 45% of Hispanic respondents indicated they were white, and 37% indicated “other” race.) 

A comparison of the two measures of race and ethnicity—the traditional version and the one allowing for multiple racial identities—is provided in Appendix Table 1. Slightly fewer than 5% of respondents identified with more than one race. When compared to the traditional measure, allowing for this option reduced the proportion of abortion patients who identified as black from 28% to 25%, and the proportions who were Asian and “other” were each reduced by about one percentage point. The more nuanced measure of race, which allows for more than one racial designation, will be used in subsequent analyses that do not require comparisons over time.

Income Level

To construct our measure of family income, we asked abortion patients to report their total family income before taxes in the previous year, and the number of family members in their household at the time of the abortion (questions 21 and 22). This information was used to calculate three income categories of less than 100%, 100–199% and 200% or more of the federal poverty level.43–46 We used these categories to refer to the patients who fall within them as poor, low-income and highest-income.

Both individual and family income levels are difficult to measure on surveys because these items often suffer from lower response rates than other types of questions. A higher level of nonresponse for this item (13%) may be the result of resistance to disclosing income, or to the fact that some patients (e.g., those living with parents and other adult family members) do not know their annual family income. However, as in 2008, the 2014 survey provided 12 annual income categories listed in $5,000 increments (ranging from less than $9,999 per year to $75,000 or more per year), with weekly incomes given parenthetically to serve as a more tangible guide to assist with estimates. We do not believe that the accuracy of this measure changed over time.  

Religious Affiliation 

Our measure of religious affiliation was adopted from the National Survey of Family Growth (NSFG). Respondents were asked “What religion are you now, if any?” (question 17), and a follow-up item determined if they were fundamentalist (question 18). Following the NSFG, we asked about four categories of evangelism, but for purposes of this survey, we collapsed them into one category. Patients who selected “other” religion were asked to specify which religion, and 991 of the 1,239 eligible did so. In line with the NSFG, we coded patients who wrote in that they were Christian (no denomination given) as Protestant. Our measure of religious affiliation in the 2008 and 2014 surveys distinguishes between mainline Protestants, evangelical Protestants, Catholics, those affiliated with some other religion and those with no religious affiliation. (Individuals who indicated that they were evangelical but affiliated with Catholicism or some other religion were not included in our measure of evangelicals.)

Sexual Orientation

In recognizing that not all abortion patients have sex only with men or identify as straight or heterosexual, sexual orientation was measured for the first time in 2014. We adapted an item from the NSFG audio computer-assisted self-interview module that asked “Do you think of yourself as…” and provided three response categories: “heterosexual or straight,” “homosexual, gay or lesbian” or “bisexual” (question 32). In our modification of this question, we added a “something else” category, and allowed patients to write in a response. While 81 respondents, or 1%, chose this option, none of the write-in answers achieved enough responses to justify analysis as its own category. Nine percent of respondents did not answer this item, higher than for standard demographic items, likely because of its sensitive nature. Gender identity was not measured.

Health Insurance Coverage

The Affordable Care Act went into effect in January 2013, and introduced state-based marketplaces for individuals to shop for private health insurance. As such, we modified the health insurance item to include this as an answer category for respondents; specifically, patients were provided with the option of indicating that they obtained coverage through HealthCare.gov or a state-run health exchange (question 4). Of the 527 respondents who indicated they obtained coverage through HealthCare.gov or a state exchange, 275 also said they had coverage under a private plan or under Medicaid; these cases were coded to the more specific plan.

An additional 149 respondents (1.7% of the sample) indicated coverage under multiple plans. Our measure of insurance allows for one type of coverage and priority was given to private insurance, followed by other, Medicaid and, finally, HealthCare.gov or the state health exchange. That is, if an individual indicated they had private insurance and some other type of coverage, they were coded to have private coverage. 

More so than for characteristics such as race and age, insurance coverage is subject to reporting error; for example, some individuals may not know which type of health insurance coverage they have, especially if it was obtained through a parent, spouse, domestic partner or other family member. In the context of health care reform, these types of errors may be more pronounced. 

Analytic Strategy

To compare the demographic characteristics of abortion patients in 2014 with those in 2008, we first retabulated the percentage distributions of abortion patients in 2008 after excluding the 399 individuals who had obtained abortions in a hospital setting (4% of the sample). We then used bivariate logistic regression analysis to test for significant differences in the proportions of abortion patient subgroups between the two surveys. All analyses were based on weighted data and were conducted using the svy command in Stata version 13.1 to account for the complex sampling design of the survey. As discussed in the Methods section of the main report, abortion indices were constructed as proxies for abortion rates across subgroups.

The population information for most characteristics reported in Table 1 come from our own tabulations of the 2008 and 2014 American Community Survey (ACS). These include age, race and ethnicity, nativity, educational attainment (for patients aged 20 or older) and family income. We relied on the 2006–2010 and 2011–2013 NSFG for population estimates of relationship status, prior births and religious affiliation. (There were no comparable population data available to allow us to examine sexual orientation.) Reliance on these data sources is a change from the previous report, for which we relied on the 2008 Current Population Survey for all demographic characteristics except religion; for the latter, we used the 2006 and 2008 General Social Survey and analysis was limited to abortion patients aged 18 or older, because this survey is restricted to adults. 

We transitioned to the new data sources for several reasons. The ACS is the largest household survey in the United States (apart from the census) and provides more accurate estimates of population characteristics.47 To make the population data comparable over time, we revised the 2008 population figures (for age, race and ethnicity, nativity, educational attainment and family income) using the 2008 ACS. 

In the past we relied on the Current Population Survey Fertility Supplement for the number of prior births, but this information is not collected in the ACS. Thus, we used the 2006–2010 and 2011–2013 NSFG instruments to generate these estimates for comparison to the 2008 and 2014 Abortion Patient Surveys, respectively. 

Because we assessed marital and cohabiting status at the time of conception—anywhere from three weeks to several months in the past, depending on how many weeks’ pregnant the respondent was—there were no population-based surveys with a precisely comparable measure. The ACS captures only unmarried or cohabiting partners of individuals filling out the survey, and so, for example, cohabiting couples living with a parent would not be captured. The NSFG captures the marital and cohabiting status of all respondents, and hence it represents the most comparable survey. The 2006–2008 NSFG was not available in time to be used for earlier analyses using data from the 2008 Abortion Patient Survey, but we did use it to revise relationship status estimates for 2008; we used the 2011–2013 NSFG to estimate cohabitation for comparison with abortion patients in 2014. 

While we relied on the General Social Survey to measure religious affiliation among all women aged 18 or older in the prior survey, item wording on both the 2008 and 2014 Abortion Patient Surveys was adapted from the NSFG surveys and, thus, we revised the 2008 population estimate using the 2006–2010 NSFG, which was not available when the 2008 Abortion Patient Survey was being analyzed. The 2011–2013 NSFG was used to estimate religious affiliation for women who were of reproductive age in 2014. 

Because of these changes, some of the 2008 population figures and, in turn, abortion indices may be slightly different from previously published figures.‡‡

References

1. American College of Obstetricians and Gynecologists, Increasing access to abortion: committee opinion no. 613, Obstetrics & Gynecology, 2014, 124:1060–1065.

2. Jones RK and Kost K, Underreporting of induced and spontaneous abortion in the United States: an analysis of the 2002 National Survey of Family Growth, Studies in Family Planning, 2007, 38(3):187–197.

3. Fu H et al., Measuring the extent of abortion underreporting in the 1995 National Survey of Family Growth, Family Planning Perspectives, 1998, 30(3):128–133 & 138.

4. Dreweke J, Abortion reporting: promoting public health, not politics, Guttmacher Policy Review, 2015, 18(2):40–47, https://www.guttmacher.org/about/gpr/2014/03/surge-state-abortion-restrictions-puts-providers-and-women-they-serve-crosshairs.

5. Centers for Disease Control and Prevention, CDCs abortion surveillance system FAQs, 2015, http://www.cdc.gov/reproductivehealth/Data_Stats/Abortion.htm.

6. Guttmacher Institute, Fewer U.S. women of reproductive age were uninsured in 2014, 2015, https://www.guttmacher.org/media/inthenews/2015/09/22/index.html.

7. Salganicoff A et al., Women and Health Care in the Early Years of the ACA: Key Findings from the 2013 Kaiser Women’s Health Survey, Menlo Park, CA: Kaiser Family Foundation, 2014.

8. Nash E, Guttmacher Institute, New York, personal communication, Nov. 24, 2015.

9. Jones RK and Jerman J, Abortion incidence and service availability in the United States, 2011, Perspectives on Sexual and Reproductive Health, 2014, 46(1):3–14, doi:10.1363/46e0414.

10. Boonstra HD and Nash E, A surge of state abortion restrictions puts providers—and the women they serve—in the crosshairs, Guttmacher Policy Review, 2014, 17(1):9–15, https://www.guttmacher.org/about/gpr/2014/03/surge-state-abortion-restri....

11. DeNavas-Walt C and Proctor BD, Income and Poverty in the United States: 2014, Washington, DC: U.S. Census Bureau, 2015.

12. Hellerstein E, The rise of the DIY abortion in Texas, The Atlantic, June 27, 2014, http://www.theatlantic.com/health/archive/2014/06/the-rise-of-the-diy-ab....

13. Texas Policy Evaluation Project, Texas Women’s Experiences Attempting Self-Induced Abortion in the Face of Dwindling Options, Austin: Texas Policy Evaluation Project, 2015.

14. Henshaw SK and Silverman J, The characteristics and prior contraceptive use of U.S. abortion patients, Family Planning Perspectives, 1988, 20(4):158–168.

15. Henshaw SK and Kost K, Abortion patients in 1994–1995: characteristics and contraceptive use, Family Planning Perspectives, 1996, 28(4):140–158.

16. Jones RK, Darroch JE and Henshaw SK, Patterns in the socioeconomic characteristics of women obtaining abortions in 2000–2001, Perspectives on Sexual and Reproductive Health, 2002, 34(5):226–235.

17. Jones RK, Finer LB and Singh S, Characteristics of U.S. Abortion Patients, 2008, New York: Guttmacher Institute, 2010.

18. Dehlendorf CE and Weitz TA, Access to abortion services: a neglected health disparity, Journal of Health Care for the Poor and Underserved, 2011, 22(2):415–421.

19. Dehlendorf C, Harris LH and Weitz TA, Disparities in abortion rates: a public health approach, American Journal of Public Health, 2013, 103(10):1772–1779.

20. Cohen SA, Abortion and women of color: the bigger picture, Guttmacher Policy Review, 2008, 11(3):2–12.

21. Sonfield A et al., The Social and Economic Benefits of Women’s Ability to Determine Whether and When to Have Children, New York: Guttmacher Institute, 2013.

22. Finer LB et al., Reasons U.S. women have abortions: quantitative and qualitative perspectives, Perspectives on Sexual and Reproductive Health, 2005, 37(3):110–118.

23. Jones RK, Frohwirth LF and Moore AM, “I would want to give my child, like, everything in the world”: How issues of motherhood influence women who have abortions, Journal of Family Issues, 2008, 29(1):79–99.

24. Lindley LL and Walsemann KM, Sexual orientation and risk of pregnancy among New York City high-school students, American Journal of Public Health, 2015, 105(7):1379–1386.

25. Charlton BM et al., Teen pregnancy risk factors among girls and young women of diverse sexual orientations, Journal of Adolescent Health, 2015, 56(2):S61–S62.

26. Jones RK, How commonly do US abortion patients report attempts to self-induce? American Journal of Obstetrics & Gynecology, 2011, 204(1):23–24.

27. Grossman D et al., Self-induction of abortion among women in the United States, Reproductive Health Matters, 2010, 18(36):136–146, https://www.guttmacher.org/about/gpr/2014/03/surge-state-abortion-restrictions-puts-providers-and-women-they-serve-crosshairs.

28. Texas Policy Evaluation Project, Knowledge, Opinion and Experience Related to Abortion Self-Induction in Texas, Austin: Texas Policy Evaluation Project, 2015.

29. Kost K and Henshaw S, U.S. Teenage Pregnancies, Births and Abortions, 2008: National Trends by Age, Race and Ethnicity, New York: Guttmacher Institute, 2012.

30. Martin JA et al., Births: final data for 2013, National Vital Statistics Reports, 2015, Vol. 64, No. 1.

31. Hamilton BE et al., Births: preliminary data for 2014, National Vital Statistics Reports, 2015, Vol. 64, No. 6.

32. Martinez GM and Abma JC, Sexual Activity, Contraceptive Use, and Childbearing of Teenagers Aged 15–19 in the United States, Hyattsville, MD: National Center for Health Statistics, 2015.

33. Kearney MS and Levine PB, Teen Births Are Falling: What’s Going On? Washington, DC: Brookings Institution, 2014.

34. Gold RB and Nash E, TRAP laws gain political traction while abortion clinics—and the women they serve—pay the price, Guttmacher Policy Review, 2013, 16(2):7–12.

35. Grossman D et al., Change in abortion services after implementation of a restrictive law in Texas, Contraception, 2014, 90(5):496–501.

36. Culp-Ressler T, Thanks to Ohio’s new abortion restrictions, more clinics are being forced out of business, Think Progress, Oct. 17, 2013, http://thinkprogress.org/health/2013/10/17/2795351/ohio-abortion-clinics....

37. Boonstra HD, Medication abortion restrictions burden women and providers—and threaten U.S. trend toward very early abortion, Guttmacher Policy Review, 2013, 16(1):18–23.

38. Kavanaugh M, Jerman J and Finer L, Changes in use of long-acting reversible contraceptive methods among United States women, 2009–2012, Obstetrics & Gynecology, 2015, 126(5):917–927.

39. Jones RK and Sonfield A, Health insurance coverage among women of reproductive age before and after implementation of the Affordable Care Act, Contraception, 2016, 93(5):386–391, http://www.contraceptionjournal.org/article/S0010-7824(15)30093-7/abstract.

40. Henshaw SK et al., Restrictions on Medicaid Funding for Abortions: A Literature Review, New York: Guttmacher Institute, 2009.

41. Salganicoff A et al., Coverage for Abortion Services and the ACA, Menlo Park, CA: Kaiser Family Foundation, 2014.

42. Davern M et al., Validating health insurance coverage survey estimates: a comparison of self-reported coverage and administrative data records, Public Opinion Quarterly, 2008, 72(2):241–259.

43. Cook PJ et al., The effects of short-term variation in abortion funding on pregnancy outcomes, Journal of Health Economics, 1999, 18(2):241–257.

44. Morgan SP and Parnell AM, Effects on pregnancy outcomes of changes in the North Carolina State Abortion Fund, Population Research and Policy Review, 2002, 21(4):319–338.

45. U.S. Department of Health and Human Services, 2014 poverty guidelines, 2014, http://aspe.hhs.gov/2014-poverty-guidelines.

46. U.S. Department of Health and Human Services, 2015 poverty guidelines, 2015, http://aspe.hhs.gov/2015-poverty-guidelines.

47. U.S. Census Bureau, Poverty data sources, 2015, https://www.census.gov/hhes/www/poverty/about/datasources/.

Footnotes

*Because Asians, Pacific Islanders and individuals who indicated an “other” race accounted for relatively small proportions of patients, and because there is more variance in estimates of race and ethnicity,16 we did not estimate abortion indices for these groups.

According to federal poverty guidelines, a family of two with an income of $15,730 or less was considered poor in 2014; a family of four meets this threshold with an income of $29,820.

Policy or court decisions in 17 states require the use of state funds to cover all or most medically necessary abortions for low-income women enrolled in Medicaid. These include Alaska, Arizona, California, Connecticut, Hawaii, Illinois, Maryland, Massachusetts, Minnesota, Montana, New Jersey, New Mexico, New York, Oregon, Vermont, Washington and West Virginia. Nonetheless, two states under court order to fund abortion services—Arizona and Illinois—report very few procedures (Sonfield A and Gold RB, Public Funding for Family Planning, Sterilization and Abortion Services, FY 1980–2010, New York: Guttmacher Institute, 2012). As a result, for analyses that distinguish between abortion patients residing in Medicaid-coverage or non–Medicaid-coverage states, we do not include Arizona or Illinois in the former.

§The survey instrument did not include a follow-up question asking if this insurance was private or public, and we were unable to further categorize the responses. We also expect that some women who obtained health insurance through these sources reported it as private or Medicaid.

**We assume the 3% of patients using Medicaid in non–Medicaid-coverage states were terminating pregnancies that were the result of rape or incest or that endangered the life of the woman.

††The rate at which facilities unsuccessfully participated or declined to participate was similar for the 2008 and 2014 surveys. In 2008, a total of 217 active facilities were approached, and 95, or 44%, successfully completed the survey.

‡‡Comparison of the previously published and the revised population figures and abortion indices showed that most stayed the same or changed only slightly, with two exceptions. The 2008 Current Population Survey estimated that 8.4% of women were cohabiting, while the 2006–2008 NSFG figure was 11.2%. In turn, the 2008 abortion index for this group was substantially lower in the current report than in the original one (2.6 vs. 3.5). In addition, religious affiliation for most groups differed by 1–4 percentage points between the 2006–2008 General Social Survey and the 2006–2008 NSFG. Notably, almost twice as many women were affiliated with an “other” religion according to the NSFG (9.4% vs. 5.4%), and the abortion index for this group changed from 1.2 to 0.7. In both cases, we believe the NSFG measures are more accurate, or at least more comparable to the items used to assess relationship status and religious affiliation on the Abortion Patient Survey.