Abortion Incidence and Service Availability in the United States, 2017

Introduction

Abortion surveillance in the United States is an important public health indicator that is needed to estimate pregnancy rates, and it can also serve as a measure of access to reproductive health care. Between 2011 and 2014, the U.S. abortion rate declined from 16.9 to 14.6 abortions per 1,000 women aged 15–44, the lowest rate ever recorded and the continuation of a decades-long trend.¹ Still, in 2014, almost one in five pregnancies ended in abortion, and given abortion rates in that year, an estimated one in four U.S. women will have an abortion in their lifetime.² These statistics demonstrate that abortion is not uncommon. 

The Supreme Court of the United States recognized the constitutional right to abortion in 1973 in Roe v. Wade. In the decades since, the Court has continued to affirm the fundamental right to abortion, including in 1992 in Planned Parenthood v. Casey and in 2016 in Whole Women’s Health v. Hellerstedt.^3,4 Despite existing precedents, states have continued to find ways to restrict or ban abortion, enacting more than 227 restrictions between January 2014 and June 2019.⁵ More than a dozen cases challenging some of the most extreme restrictions—such as bans on abortions after six weeks’ gestation—currently have the potential to reach the Supreme Court, and the outcomes could pose significant challenges to the legal framework protecting abortion rights.⁶ If the Court undermines or overturns Roe v. Wade, this will likely exacerbate existing disparities in abortion access and may allow individual states to explicitly or effectively ban abortion altogether.⁷ Although prior research has not found state policy to be the primary driver of the decline in the national abortion rate,¹ abortion bans would undoubtedly prevent many individuals from obtaining abortion care in clinical settings.

Documenting changes in the number of health care facilities that provide abortion is also an important activity, as the number of facilities can directly affect the availability and accessibility of care. In 2014, the vast majority (95%) of abortions were provided by clinic facilities, while 4% were provided by hospitals and 1% by private physicians’ offices.¹ Between 2011 and 2014, the number of clinics providing abortions had declined by 6%. These declines were steepest in the Midwest and South (22% and 13%, respectively), regions that had also enacted the most abortion restrictions.⁸ Still, the 2014 study did not identify a clear association between changes in clinic numbers and state abortion rates between 2011 and 2014; for example, the declines in abortion rates in some states that had lost one-third or more of their clinic facilities mirrored, or were smaller than, the national decline. Updated national data suggest that the overall number of facilities providing abortion did not change much between 2014¹ and 2017,⁹ but state and regional patterns of clinic closures may reveal meaningful patterns in availability of and access to services over time.

Some individuals may self-manage their abortion outside of clinical settings. National estimates of the incidence of self-managed abortion are limited and cannot be developed using traditional surveillance at the facility level. However, facility-level reports of the number of patients seen following an attempt to self-manage an abortion may reveal interesting patterns when those individuals seek follow-up care. In 2014, 12% of nonhospital facilities reported that they had seen one or more patients who had attempted to self-manage an abortion, and these proportions were highest in the South (21%) and the Midwest (16%).¹

This report, which summarizes findings from the Guttmacher Institute’s most recent Abortion Provider Census (APC), provides estimates of abortion incidence and the number of clinic facilities providing abortion care in 2017, nationally and by state. We also examine patterns in abortion rates and abortion legislation in the 10 states that experienced the largest proportionate declines and increases in clinic numbers. The last APC gathered data for 2014, and our analysis focuses on changes since that time. Finally, we examine the incidence of medication abortion and facilities that provided only medication abortion in 2017. Taken together, the data from this report provide the most comprehensive accounting of abortion incidence and of the landscape of clinical abortion service provision in the United States.

Methods

Data for this study are from the Guttmacher Institute’s most recent APC. This was the 18th national census of its kind; the first assessed abortion incidence in 1973.¹⁰ Conducted between January 2018 and March 2019, the APC surveyed all health care facilities known or suspected to have been providing abortions in the United States in 2016 and 2017. We modeled the data collection instruments on the questionnaires used in our prior study, which collected data for 2013 and 2014. Several new questions were added, and other changes were made to improve clarity. The instrument sent to clinics and to physicians’ offices had more questions than the questionnaire for hospital facilities, in order to account for differences in service provision. We asked contacts at all facilities the number of abortions provided at their site in 2016 and 2017.

The universe for data collection included all facilities known to have provided abortions in 2014, as well as additional or new facilities identified through Web-based searches, media reports, and reviews of directories of organizations and associations that work with abortion-providing facilities. The universe also included facilities identified in a national survey of obstetrician-gynecologists conducted by the Guttmacher Institute in 2015 to capture information on abortion and referrals provided in private practice settings.¹¹ We sent two mailings, one in January 2018 and a second to all nonresponding facilities four weeks later. We conducted intensive telephone, fax and e-mail follow-up for 13 months (from early March 2018 through March 2019) to acquire completed questionnaires. We gave priority to obtaining the total number of abortions for 2016 and 2017 over other questions on the survey. During this period of follow-up, we documented more than 16,775 phone calls, e-mails and faxes, including contacts for facilities that we discovered had closed during the survey period.

We also collected state health department data for 2016 and 2017 to supplement information obtained from abortion-providing facilities, and we sometimes used this information to generate estimates for nonresponding facilities. We requested abortion occurrence data from state health department offices in 45 states and the District of Columbia. The quality, completeness and type of health department data available varied widely across states. Where possible, we collected the number of abortions by facility, but most commonly, health departments provided data by county of occurrence or by facility type. In some states, we were able to obtain only the total number of abortions.

Response Rates and Completeness of Data

We collected data directly from 59% of the 2,277 facilities in our universe of potential abortion-providing facilities; we received questionnaires from 954 entities (42%) and acquired data during telephone and e-mail follow-up from an additional 386 entities (17%). We used health department data to determine the abortion caseloads of 423 facilities (19%). For another 446 facilities (20%), we estimated abortion totals using a variety of techniques and information sources. In some cases, estimates were based on 2013 and 2014 caseloads, with adjustments for trends in service provision at other facilities in the same state or metropolitan area. In addition, to estimate caseloads at a small number of facilities, we combined information from a range of sources, including the internet,* other researchers and key informants in communities who had in-depth knowledge about reproductive health service provision.

The majority (72%) of facilities that did not provide data and for which we made estimates were hospitals and physicians’ offices, which have small abortion caseloads (based on responding facilities). We determined that 71 of the 446 facilities for which we estimated caseloads provided no abortions during the survey period. The remaining 68 facilities (3%) in the overall survey universe were found to have closed, to have stopped providing abortions before 2016 or to have been added in error.

The large majority (89%) of the total number of abortions that we counted in 2017 were reported by the facilities, 75% via questionnaire and an additional 14% from nonresponse follow-up. Two percent of the abortions we counted were obtained from health departments, and we estimated 9% of abortions using historical data and information from key informants and information obtained via the internet.† This distribution is comparable to that in the 2014 APC, where 68% of counted abortions were reported via questionnaire, 20% from nonresponse follow-up, 7% from estimates and 5% from health department data.¹ 

The degree to which data were estimated varied across states. States for which we had to estimate more than 10% of abortions (the highest proportion of missing data) were Rhode Island (31%), Florida (25%), New York (19%), New Jersey (17%), North Carolina (15%), Nevada (14%), New Mexico (11%) and Texas (11%).‡

We obtained approval for this study from the Guttmacher Institute’s federally registered institutional review board.

Analysis

We distinguished among four types of abortion-providing facilities: specialized abortion clinics, nonspecialized clinics, hospitals and physicians’ offices. Specialized abortion clinics are defined as nonhospital facilities in which half or more of patient visits were for abortion services, regardless of annual abortion caseload. Nonspecialized clinics are nonhospital facilities in which fewer than half of patient visits were for abortion services. Physicians’ offices are defined as facilities that provided fewer than 400 abortions per year and had names suggesting that they were private practices. Physicians’ offices that provided 400 or more abortions per year are categorized as clinics (either specialized or nonspecialized), as their caseload suggests service provision more closely mirroring that of a clinic. Similarly, as in the past, we classified hospital-affiliated clinics as either specialized or nonspecialized clinics; traditional hospital-based sites of care, such as operating rooms, emergency departments, labor and delivery wards, and maternal-fetal medicine departments, are classified as hospitals.

We used Census Bureau data on the population of women aged 15–44 on July 1, 2016, and July 1, 2017, to calculate the national abortion rate, as well as the rates for each state and the District of Columbia.¹² We estimated the national abortion ratio as the proportion of pregnancies that ended in abortion (and excluded those ending in miscarriages). To do this, we combined our abortion counts with National Center for Health Statistics data on the number of U.S. births in the one-year periods beginning on July 1 of 2016 and 2017.^13,14

More than two-thirds of nonhospital facilities (68%) offered information about the number of early medication abortions they provided, specified as abortions provided by mifepristone with misoprostol, by methotrexate or by misoprostol alone up to 10 weeks since the last menstrual period (LMP). "Early" medication abortion was indicated on the survey to preclude the reporting of medication used for cervical preparation prior to surgical abortion. The response rate on this measure varied by facility type and caseload, and we constructed weights to account for these differences. We also estimated the proportion of eligible abortions that were provided using medication. As of June 2016, the U.S. Food and Drug Administration (FDA) regimen permits use of mifepristone up to 10 weeks’ gestation. To estimate the number of eligible abortions, we used the 2015 Centers for Disease Control and Prevention (CDC) abortion surveillance report (the most recent CDC data available) to calculate the proportion of all abortions performed at nine weeks or less and applied this proportion to our 2017 abortion count.¹⁵

Changes to the Survey

The 2014 APC questionnaire included an item asking nonhospital facilities if any patients had been treated for missed or failed abortions due to self-induction in 2014, and if so, how many.¹ We included a similar item on the 2017 questionnaire but removed the yes/no screener and asked only for the total number of patients treated for missed or failed self-managed abortions in 2017. Among nonhospital facilities, 55% provided information on this item, but 106 facilities indicated "don’t know," compared with 10 in 2014. It was unclear whether a "don’t know" response on the 2017 survey indicated that respondents were unsure of the total number of patients or that they were unsure whether they had seen any patients for this reason. Of the 106 facility responses indicating "don’t know," 67 were completed by four respondents at central administrative offices; they indicated "don’t know" for each facility in the network, including some with abortion totals as few as one or two in 2017. We were able to follow up with three out of four of these respondents, and they indicated that clinic staff may have seen patients who attempted self-induction but were unsure; for example, they noted that it was possible some miscarriages they treated may have been self-managed. Thus, as we did for the 2014 data, we assumed that a "don’t know" response meant facility staff were unsure whether patients had been treated for self-induction; these facilities were coded as missing data for this question. As a sensitivity analysis, we examined this measure assuming that a response of "don’t know" was equivalent to having seen no patients who attempted to self-manage. We constructed weights to take response patterns by facility size into account, and our findings are based on weighted data.

In the findings and discussion, we focus on the 2017 abortion data; estimates for 2016 are included in tables reporting numbers and rates of abortions.

Findings

Abortion Incidence

The total number of abortions, the abortion rate and the abortion ratio in the United States all declined between 2014 and 2017 (Table 1). In 2017, 862,320 abortions were provided in clinical settings, a 7% decline from 2014. The 2017 abortion rate of 13.5 abortions per 1,000 women aged 15–44 represented an 8% decline from 2014. Just under one in five pregnancies (births and abortions), 18.4%, ended in abortion in 2017.

While abortion incidence and rates declined in most states, the degree of change varied substantially (Table 2). Declines in abortion rates were largest in Delaware, Arkansas, West Virginia, Alabama and Virginia; most of these states also had abortion rates substantially lower than the national rate in 2014, so even a small change in this measure can seem large. Abortion rates increased in Mississippi, New Jersey, Minnesota, Georgia, Maryland and Wisconsin.§ While abortion rates declined in all four regions, the drop was steepest in the West (14%). Indeed, states considered to be supportive of abortion rights in 2017—including large states such as California and New York—accounted for 43% of all U.S. abortions in that year but 55% of the decline since 2014.**¹⁶

Areas with the highest abortion rates in 2017 were the District of Columbia, New Jersey, New York, Maryland and Florida. Rates were lowest in Wyoming, South Dakota, Kentucky, Idaho and Missouri. Notably, our study measures abortion by state of occurrence and does not account for individuals crossing state lines for abortion care; in the five states with the lowest rates, 28% or more of individuals go out of state to obtain abortions.¹⁷

Abortion Facilities

In 2017, 1,587 health care facilities were known to have provided abortions (Table 3), a 5% decline from 2014. Changes in the overall number of facilities over time varied by facility type. The number of hospitals providing abortions declined by 19%, from 638 to 518. This decrease was largely attributable to California, where 114 hospitals that provided 633 abortions in 2014 reported zero procedures in 2017 (data not shown). California hospitals accounted for the same proportion of abortions in the state in both years (5%).

The number of specialized abortion clinics—facilities where half or more of patients received abortion care—declined by 7%, from 272 in 2014 to 253 in 2017. The number of nonspecialized clinics increased by 7% over the same period, from 517 to 555. Clinic facilities play a critical role in abortion provision, and, in 2017, specialized abortion clinics provided 60% of all abortions, while nonspecialized clinics provided 35%.

Changes in the number of clinics (both specialized and nonspecialized) providing abortion care varied substantially by region and by state (Table 4). (Information on the total number of facilities providing abortion care, by state, is available in Appendix Table 1.) The number of clinic facilities declined by 9% in the South, the largest of the four regions. Indeed, nine of the 17 states in the South lost at least one such clinic during the study period. The Midwest also saw a 6% decline in the number of clinic facilities providing abortion care—i.e., six fewer clinic facilities in 2017 than in 2014. Indiana, Iowa, Ohio and Wisconsin all had fewer clinics in 2017 than in 2014.

The number of clinic facilities increased by 16% in the Northeast and by 4% in the West. Four of the nine states in the Northeast had an increase in clinic facilities, as did five of 13 in the West. Despite this overall increase, a few states in each of these regions also saw a decline in the number of clinic facilities between 2014 and 2017.

The overall increase in clinic facility numbers masks some amount of turnover. Of the 808 clinic facilities providing abortions in 2017, 127 had not been providing this care in 2014. Similarly, of the 789 clinic facilities providing abortions in 2014, 103 had stopped doing so or had closed by 2017.*†

In 2017, 89% of U.S. counties did not have a clinic facility that provided abortion care, and 38% of women aged 15–44 lived in these counties (Table 4); these figures are comparable to those found in 2014—90% and 39%, respectively.¹ In five states, fewer than 10% of women lived in a county without a clinic facility: California, Connecticut, Hawaii, Nevada and New York. In Mississippi and Wyoming, more than 90% of women lived in a county without such a clinic.

Changes in Clinic Numbers, Abortion Rates and Policy Context

To explore whether decreases in the numbers of clinics across states might be related to declining abortion rates (and vice versa), we examined changes in rates in the 10 states with the largest proportionate increases and decreases in clinic numbers. We also examined the number of new laws enacted in all 20 states, as these restrictions could have made it harder for health care facilities to provide abortions.

Five of the 10 states with the largest proportionate declines in clinics had one fewer facility in 2017 than in 2014: Kentucky, West Virginia, Rhode Island, Wisconsin and Louisiana (Table 5). In fact, Kentucky and West Virginia were left with only one clinic in 2017. Thirteen of the 25 clinics that were no longer providing abortion in these 10 states had been specialized abortion clinics (which had provided an average of 957 abortions each year in 2014), and the other 12 had been nonspecialized clinics (data not shown).

The change in abortion rates varied among states with the largest proportionate declines in clinic numbers (Table 5). Abortion rates increased or stayed the same in Wisconsin and New Mexico. Declines in abortion rates were lower than or comparable to the national level in six of the states and larger than average in only two, West Virginia (–26%) and Iowa (–15%). All but two of the states—New Mexico and Rhode Island—had enacted multiple abortion restrictions over the study period.

Six of the 10 states with the largest proportionate increases in clinic facilities had one or two additional facilities in 2017. Larger increases were seen in Maine (12 clinics), Massachusetts (5), New York (18) and Washington (7). The proportion of women in counties without a clinic decreased from 55% to 24% in Maine and from 37% to 32% in Alaska.¹ The overwhelming majority of newly providing clinics in these 10 states (50 in total) were nonspecialized clinics; only three were abortion clinics (data not shown).

Abortion rates declined in all 10 states that had more clinic facilities in 2017 than they did in 2014; declines of more than twice the national rate occurred in Delaware (–37%) and South Carolina (–17%). Half of these states had not enacted any abortion restrictions during the study period; among the five that had, Oklahoma and South Carolina had enacted the greatest number of restrictions (eight each). A correlation analysis revealed no association between percentage change in rates and percentage change in clinic providers between 2014 and 2017, either nationally (r=–.04, p=.764) or among the 20 states in Table 5 (r=–.05, p=.894; data not shown).

Medication Abortion

In 2017, 339,640 medication abortions were provided in nonhospital facilities (Table 6), a 25% increase from 2014.¹ Medication abortion accounted for 39% of all abortions. Assuming that health care providers followed the FDA-recommended regimen that allows mifepristone to be administered up to 10 weeks’ gestation, we estimate that 60% of all eligible abortions were early medication abortions (data not shown). The majority of medication abortions were provided by specialized clinics and at high-volume facilities (those with annual caseloads of more than 1,000 abortions).

In 2017, a minimum of 25% of all nonhospital facilities (including physicians’ offices) and 30% of clinics provided only early medication abortion, representing a slight increase from 2014, when these figures were 23% and 26%, respectively.¹ In particular, a higher proportion of nonspecialized clinics (41%) than specialized abortion clinics (4%) offered only early medication abortion. Facilities with abortion caseloads of 30–399 abortions had the highest proportion offering only early medication abortion (41%).

Self-Managed Abortion

In 2017, 18% of nonhospital facilities reported that they had seen one or more patients for a missed or failed abortion due to self-induction (data not shown), up from 12% in 2014.¹ Reports of self-managed abortion were highest in the South (25%) and the West (21%), compared with 10% in the Midwest and 14% in the Northeast. The majority of these facilities (54%) had seen only one or two such patients, but four facilities (all high-volume) reported 50 or more.

The sensitivity analysis (which assumed that a response of "don’t know" to the question was comparable to having seen no patients who attempted to self-manage) suggested that 15% of nonhospital facilities had seen at least one of these patients; the regional distributions were 18% in the South, 19% in the West, 11% in the Northeast and 9% in the Midwest.

Discussion

Between 2014 and 2017, abortions provided in clinical settings in the United States continued to decline. The 2017 rate of 13.5 abortions per 1,000 women aged 15–44 is the lowest recorded since abortion was legalized nationally in 1973 and is 54% lower than the peak rate of 29.3 per 1,000 in 1980.¹⁸ The decline was seen across all four regions and most states.

One factor that can contribute to declines in abortion is a reduction in the number of facilities providing this care. While hospitals and physicians’ offices constituted a substantial share of abortion-providing facilities, the overwhelming majority of abortions, 95%, were provided by clinics. Thus, our analysis focused on trends in clinic facilities during the study period. Nationally, there was a slight (2%) increase in the total number of clinic facilities, but this increase was not evenly distributed by region because the overall number of clinics increased only in the Northeast and in the West. The substantial increase seen in Maine was due to the introduction of abortion services via telemedicine.¹⁹ Adoption of this technology allowed for a jump in the number of clinical sites providing abortion care in a state where a larger than average proportion of residents live in rural areas. Similar developments may have contributed to the increase in clinics in Alaska and Washington.²⁰

The overall number of clinics in the Midwest and the South declined. Texas had the largest drop, losing seven clinic facilities between 2014 and 2017. The 2014 figure included clinics that provided any abortions within that year, but a number of these facilities had stopped providing abortion care at some point in 2014 due to state abortion restrictions—in particular, the requirement that physicians providing abortions have admitting privileges at a nearby hospital.²¹ While this and other Texas restrictions have since been struck down by the U.S. Supreme Court,³ a number of clinics in the state have not reopened or reintroduced abortion care. The Texas legislature passed 10 additional abortion restrictions between 2014 and 2017, indicating continued attempts to restrict access to abortion care. Similarly, a state restriction passed in Ohio in 2013 that required facilities providing abortions to have a transfer agreement with a public hospital was amended to be even more stringent in 2015, requiring that the public hospital be within 30 miles of the facility; this development contributed to the continued decline of clinic facilities in that state.²²

In 2014, four states had only one clinic providing abortion care—Mississippi, Missouri, North Dakota and South Dakota.¹ In 2017, Kentucky and West Virginia were also down to one clinic facility, while Missouri has fluctuated between one and three clinical sites (see Limitations). All of these states are located in the Midwest and the South.

Our analysis did not identify any clear patterns between changes in states’ number of clinic facilities and trends in state abortion rates. Iowa and West Virginia were the only states that lost a substantial share of clinics and also had larger than average declines in abortion rates. In 2014, 13–14% of abortions in both states were provided to nonresidents.¹⁷ It is possible that the declines in abortion in these states reflect, at least in part, a decline in the number of individuals who crossed state lines to obtain care, since there were fewer clinics.

All 10 states that had a meaningful increase in clinic numbers also showed declines in their abortion rates. Most of the new facilities, or facilities that had not previously provided abortions in these states, were nonspecialized clinics, suggesting that the concurrent expansion of abortion care and decrease in abortion rates was taking place in the context of an increase in comprehensive health care.

While the decline in the number of clinics providing abortion care in some states likely prevented some patients from obtaining wanted abortions, other factors also contributed to the decline in the abortion rate. Fertility rates declined in virtually all states between 2014 and 2017,¹⁴ suggesting that the drop in abortions was not compensated for by an increase in births. Rather, declines in reported abortions could be related to at least two other factors: self-managed abortion and a decline in pregnancy rates.

Our analysis found an increase in the proportion of nonhospital facilities that had treated at least one patient for a missed or failed abortion due to self-induction, from 12% in 2014 to 15–18% in 2017. For more than a decade, the drug misoprostol has been available over or behind the counter in other countries and has been brought to the United States, and researchers have documented its use for self-managed abortion in the United States.^23–25 More recently, drugs similar to those used in the U.S. medication abortion regimen—a highly effective combination of mifepristone and misoprostol—have become available on the internet,²⁶ as have websites providing accurate information about how to safely and effectively self-manage abortion using drugs obtained outside of a clinical setting. In particular, Aid Access, an international organization that provides medication abortion pills via mail order to people living in the United States, launched their website in March 2018 (after the study period) and reported filling 2,500 prescriptions in that year.²⁷ The majority of patients obtaining abortions are poor or low-income, many lack health insurance that will cover the procedure,²⁸ and many live in states with numerous abortion restrictions.⁸

These factors, along with the increased accessibility of resources to help individuals safely self-manage their abortions outside of a clinical setting, likely account for some of the decline in abortions that we have documented. However, one national survey of U.S. adult women, conducted in 2017, found that only 1.4% reported ever having attempted to end a pregnancy on their own.²⁴ Moreover, 24% of these instances had occurred prior to 2000, and only 28% were reported to have been successful. Abortion is underreported on surveys of this type, and the actual incidence may be higher,^29,30 but it is nonetheless unlikely that even a substantial increase in self-managed abortion can account for the majority of the decline in abortion incidence nationally during the study period.

The decline in births and abortions also means that fewer people were getting pregnant. Improved contraceptive use is one factor that could have contributed to this change. The most recent national data suggest that between 2014 and 2016, the proportion of women aged 15–44 using long-acting reversible contraceptive methods increased by 23%, from 13% to 16%; levels of sterilization were 25% and 26%, respectively.^31,32 Greater reliance on highly effective methods appears to have been balanced by a drop in the use of hormonal methods such as the pill and the injectable (Depo) which, combined, declined from 29% to 25% of all contraceptive use. Still, it is possible that a decline in contraceptive failures could have reduced the incidence of unintended pregnancy. Additionally, state-level efforts to increase access to long-acting reversible contraceptive methods^33–35 may have had a measurable impact, particularly in states with higher-than-average abortion rates.

Medication abortion has come to play a pivotal role in abortion care. Even while the overall number of abortions declined, the number of medication abortions and the proportion of all abortions that were medication abortions increased. Some of this increase was due to a June 2016 change in the FDA medication abortion regimen: Prior to that time, state laws in North Dakota, Ohio and Texas mandated that facilities adhere to a less-effective regimen that relied on outdated guidelines on the maximum gestation at which this method could be used.³⁶ For the years in which these state laws were in effect, the required regimen increased the cost of medication abortion and restricted the number of patients who were eligible to use it.^37,38 The continued increase in clinics that provide only early medication abortion might also have contributed to the increased incidence in this category of abortion. This category of facilities may include clinics that do not have the staff or equipment to provide procedural abortions, and their ability to provide medication abortion has likely increased access to care. At the same time, it is also possible that some of the increase can be attributed to early medication abortion’s being the only option available to patients in some areas.

Limitations

This study had a number of shortcomings. Although 89% of the abortions we counted in 2017 were based on information provided by individuals working at or for the facilities from which the data were collected, we estimated 9% of abortions. This problem was particularly pronounced in eight states, including larger ones such as New York and Florida. It is possible that we consistently underestimated or overestimated these caseloads, which would mean that our count is inaccurate. In addition, our study was only able to measure abortions that occurred in clinical settings. If the number of individuals who were able to successfully self-manage their abortions has increased substantially, then the decline we recorded may have been artificially high. While we found an increase in the proportion of nonhospital facilities that had seen at least one patient who had attempted to self-manage, this estimate is based on the assumption that facilities that did not provide information on this measure were similar to those that did.

We are aware that our study did not capture all facilities that provide a small number of abortions per year—hospitals and private physicians’ offices in particular. A 2016 study of abortion provision in private practices of obstetrician-gynecologists suggests that our study is missing as many as 1,200 of these facilities and as many as 12,000 abortions (or 1% of the total abortions estimated to have occurred in 2017).¹¹ Still, we are confident that we have accounted for most, if not all, clinic facilities, and these provide the overwhelming number of abortions. 

Our study measured abortion by state of occurrence, but many patients cross state lines to obtain care. For example, CDC data for 2015 suggest that 28% of abortions reported to have occurred to residents of Idaho and 83% of those to residents of Wyoming were obtained in other states.^1,17 Similarly, though the District of Columbia (DC) had the highest abortion rate in the country in 2017, the majority of abortions provided in DC in 2014 were for nonresidents, most commonly individuals from Maryland or Virginia.¹⁷

The abortion policy landscape changes rapidly, and some of the information in this report may already be out of date. For example, while Missouri had three clinics providing abortion care in 2017, two have since closed due to a law mandating that physicians at medication-only abortion facilities have admitting privileges at a local hospital and a contract with an obstetrician-gynecologist.³⁹ Similarly, we found some amount of turnover among clinics from one study period to the next, and the number of facilities providing abortion care is constantly changing.

Conclusions

Declines in abortion were seen in all four regions of the United States, including in states with policy landscapes that were both restrictive toward and supportive of abortion rights. However, access to abortion, when measured by the number of clinic facilities in a state, has become more polarized across regions of the country. The overall number of clinics increased in the Northeast and the West but declined in the Midwest and the South; in addition, more states (all in the Midwest or the South) have only one clinic remaining. These patterns demonstrate that the existence of more clinic facilities does not necessarily translate to an increase in abortion rates. Rather, an increase in clinic numbers likely represents greater access to health care in general, enabling patients to travel shorter distances, obtain abortion care in nonspecialized settings and perhaps obtain contraceptive care more easily.

Medication abortion plays an integral role in abortion care, having accounted for 39% of all abortions in 2017 and more than half of abortions occurring prior to 10 weeks’ gestation. The availability of mifepristone not only allows some patients to choose between types of abortion procedure, but also lends itself to innovations in health care delivery models, such as telemedicine. For this reason, the landscape of abortion provision and access in the United States may change as these innovations spread. In addition, the increased availability of highly effective and affordable abortion pills via the internet has the potential to substantially increase access to abortion, for which future surveillance efforts will need to account. It will also be important to ensure that policies and funding promote access to all methods of abortion, so that people seeking this care are able to obtain the care that is best for them.

As abortion service delivery and utilization continue to be restricted at the state level, documentation of abortion incidence, abortion rates and numbers of service sites is necessary to establish baselines and measure trends in a changing health care landscape.

FOOTNOTES