Abortion Incidence and Service Availability In the United States, 2014.

CONTEXT: National and state-level information about abortion incidence can help inform policies and programs intended to reduce levels of unintended pregnancy.
METHODS: In 2015-2016, all U.S. facilities known or expected to have provided abortion services in 2013 or 2014 were surveyed. Data on the number of abortions were combined with population data to estimate national and state-level abortion rates. The number of abortion-providing facilities and changes since a similar 2011 survey were also assessed. The number and type of new abortion restrictions were examined in the states that had experienced the largest proportionate changes in clinics providing abortion services.
RESULTS: In 2014, an estimated 926,200 abortions were performed in the United States, 12% fewer than in 2011; the 2014 abortion rate was 14.6 abortions per 1,000 women aged 15-44, representing a 14% decline over this period. The number of clinics providing abortions declined 6% between 2011 and 2014, and declines were steepest in the Midwest (22%) and the South (13%). Early medication abortions accounted for 31% of nonhospital abortions, up from 24% in 2011. Most states that experienced the largest proportionate declines in the number of clinics providing abortions had enacted one or more new restrictions during the study period, but reductions were not always associated with declines in abortion incidence.
CONCLUSIONS: The relationship between abortion access, as measured by the number of clinics, and abortion rates is not straightforward. Further research is needed to understand the decline in abortion incidence.

Information about abortion incidence in the United States is necessary to estimate accurate pregnancy rates and to determine rates of unintended pregnancy.[1] In 2011, there were 1.06 million abortions, and 21% of pregnancies were terminated.[2] These figures reveal that abortion is not uncommon. Abortion incidence in 2011 was remarkable for several reasons. Between 1990 and 2008, the abortion rate declined an average of 2% per year,[3] but between 2008 and 2011, it dropped 13%.[2] Fewer women had abortions in 2011 than in 2008 because fewer women became pregnant when they did not want to: Over this period, the proportion of pregnancies that were unintended declined from 51% to 45%, and the rate of unintended pregnancy dropped 18%, from 54 to 45 per 1,000 women.[1]

Research suggests that a rise in contraceptive use was responsible for at least some of the decline in unintended pregnancy between 2008 and 2011, and that increased use may have continued into more recent years. Reliance on long-acting reversible contraceptive (LARC) methods—the IUD and the implant—increased 130% between 2007 and 2009,[4] and this trend continued, albeit at a slightly slower pace, through 2012.[5] The proportion of clients obtaining LARC methods at Title X–supported family planning clinics rose from 7% in 2011 to 11% in 2014.[6] Clients at Title X–supported facilities are disproportionately young and low-income,[7] and these populations account for the majority of unintended pregnancies.[1] Thus, even relatively small increases in LARC use among this population could reduce the incidence of unintended pregnancy and abortion.

Abortion incidence can also decline if women who want abortions are unable to obtain them; abortion restrictions have the potential to reduce abortion incidence by impeding access to services. Between 2008 and 2011, some 24 states enacted 106 abortion restrictions.[2] However, no strong evidence exists that these restrictions were the main factor behind the decline in abortion. While some hospitals and physicians’ offices provided abortions, the overwhelming majority of procedures—95%—were accounted for by clinics.[2] Between 2008 and 2011, the number of clinics providing abortions declined by only 1%, and decreases in abortion incidence occurred in almost all states, including states that enacted multiple restrictions and states that enacted none.[2] Some of the laws implemented in 2008–2011 would not be expected to have a measurable impact on abortion incidence—for example, regulations requiring that new information be added to existing counseling materials. Additionally, 62 of the 106 abortion restrictions were enacted in 2011, and many were not implemented or enforced until late in the year. Thus, their impact might not have been evidenced until 2012 (or later).

State efforts to restrict abortion have maintained their momentum, and states have enacted hundreds of new restrictions pertaining to abortion since 2011.[8] Most restrictions, such as mandated ultrasounds and in-person counseling requirements, are intended to discourage women from having abortions, thereby reducing demand for services.[9] However, legislators have increased efforts to restrict the provision of abortion, typically through targeted regulation of abortion providers, or TRAP, laws.[9,10] These laws place burdensome regulations on abortion providers and the facilities in which they work. Perhaps the most high-profile case was Texas, which, in 2013, implemented TRAP laws requiring that physicians who provided abortions have admitting privileges at nearby hospitals and that clinics providing abortions meet the standards of ambulatory surgical centers. Following the implementation of these laws, more than half of the 46 clinics in the state closed, and the abortion rate had declined 13% by April 2014.[11] In June 2016, the Supreme Court overturned both provisions of this law,[12] and in the succeeding four months, the decision was cited three times to block enforcement of abortion restrictions in Alabama,[13] Florida[14] and Indiana.[15] Moreover, the Supreme Court has denied judicial review to pending cases in Mississippi and Wisconsin,[16] and more TRAP laws are likely to be challenged. However, as of August 1, 2016, some 24 states had implemented TRAP laws, and these regulations may have affected access to abortion in 2013 and 2014.[17]

When unable to access abortion services in the face of restrictive laws, some women may turn to self-induced abortion.[18] One study estimated that as many as 100,000 women aged 18–49 residing in Texas had ever attempted to end a pregnancy on their own,[18] and a media analysis found that interest in self-induced abortion—as measured via Google searches—was higher in states with restrictive abortion laws than in states without them.[19] If substantial numbers of women were able to have abortions outside of a health care setting, the estimated incidence of abortions occurring in clinical settings would be too low.

This study summarizes findings from the Guttmacher Institute’s most recent Abortion Provider Census, reflecting abortion incidence and the number of facilities providing abortion in 2013 and 2014. Because of the important role of clinics, we examine the abortion policy context in states that experienced the greatest proportionate changes in the number of these types of facilities.

METHODS

Survey Content and Fielding

Data for this study come from the Guttmacher Institute’s 2014 Abortion Provider Census, which surveyed the known universe of abortion-providing facilities in the United States.[2] Questionnaires were similar to the instrument used in 2011; some modifications were made for clarity, and several new items were added. Nonhospital facilities (clinics and physicians’ offices) received a longer questionnaire than hospitals because of differences in service provision. Regardless of facility type, all respondents were asked the number of induced abortions that were provided at their location in 2013 and 2014; hence, the state in which a patient obtained an abortion was not necessarily her state of residence. Past iterations of this survey used a simple “yes/no” item to determine whether hospitals had provided terminations, and this was intended to capture terminations for any reason. However, to ensure that we were capturing all abortions, and not only elective ones, the item was expanded to include specific reasons for providing an abortion: for health indications (such as fetal anomaly or health of the mother), because a patient requested one and for other reasons. This strategy was also adopted during nonresponse follow-up, when hospital facilities were contacted by telephone.

In addition to the foregoing items, nonhospital facilities were asked about the proportion of services accounted for by abortion and the number of early medication abortions provided. The survey instrument defined early medication abortions as procedures performed up to nine weeks’ gestation; to ensure that we captured all such abortions (and not just those done with mifepristone), we asked providers to identify the type of medication used, distinguishing among mifepristone (with misoprostol), methotrexate and misoprostol alone. The 2014 survey included two new items for nonhospital facilities. We asked respondents to estimate the total number of days on which the facility was unable to provide abortion services in 2014 because of efforts to come into compliance with local or state laws or regulations. This item was intended to capture the potential impact of TRAP and other types of laws on abortion services. In addition, we asked if the facility had treated any patients for missed or failed abortion as a result of women’s attempts to end a pregnancy on their own and, if yes, to estimate the total number of patients seen for this reason. This item was intended to serve as a crude measure of the prevalence of self-induced abortion.

The survey universe comprised all facilities known to have provided abortions in 2011, as well as possible new abortion-providing facilities, as identified via Internet searches, telephone directories, media articles and membership directories of organizations that work with abortion service providers. We mailed the first questionnaire to all facilities in the universe in March 2015, and we followed with two mailings at four-week intervals to nonrespondents. Intensive follow-up of nonrespondents was conducted via telephone, fax and e-mail between June 2015 and April 2016 to obtain completed questionnaires; these efforts prioritized obtaining the total number of abortions over other survey items. During this phase of data collection, more than 11,800 contacts were made with approximately 1,800 facilities, including administrators at facilities that had closed.

We also requested abortion data from health department agencies in the District of Columbia and the 45 states in which they were available. Reliability and type of information available varied across states. Where possible, we collected the number of abortions by facility, but most commonly, we obtained information by county of occurrence or facility type; in some states, we were able to obtain only the total number of abortions. This information was used to supplement and validate information obtained from abortion-providing facilities and was sometimes used to generate estimates for nonresponding facilities.

Of the 2,313 facilities in the universe of potential abortion-providing facilities, we collected data from 868 (38%) via the questionnaire and from 463 (20%) during telephone follow-up. Health department data were used to determine caseloads for 460 facilities (20%). We estimated abortion figures for 390 facilities (17%). For 265 of these, we based our estimates on abortion numbers from previous census results and service patterns of other abortion-providing facilities in the community; for another 125, we used information from key informants (e.g., grass-roots organizations and individuals who were knowledgeable about reproductive health services in a community) and the facilities’ websites. The majority of nonresponding facilities for which we made estimates (291 of the 390) were hospitals and physicians’ offices, and both types of facilities typically have small caseloads;[2] 75 of the 390 facilities were estimated to have provided no abortions during the survey period. The remaining 132 facilities (6%)[*] in the overall universe were found to be closed or to have stopped providing abortion services.

Of the abortions counted in 2014, some 68% were reported via questionnaire, and an additional 20% during nonresponse follow-up. Five percent of abortions were reported by health departments, and 7% were estimated using historical data and information from key informants and facilities’ websites.[†] Notably, the extent to which data had to be estimated was not uniform across states. The highest proportions of missing data were seen in the District of Columbia (12%), Florida (21%), Georgia (11%), Hawaii (19%), Michigan (15%), New Jersey (16%), New Mexico (21%) and Oklahoma (18%).

Undoubtedly, some facilities that provide abortion care, especially those with small caseloads, are not known to us and are excluded from our census. To address this issue, we used the American Medical Association’s master list to obtain a random sample of 2,000 physicians who identified their specialty as obstetrics and gynecology,[‡] and who provided a phone number; the latter was necessary so that we could conduct phone follow-up. Notably, 49% of all obstetrician-gynecologists on the master list did not provide a phone number, and our strategy assumed that these physicians did not differ from those who did with regard to abortion provision. We mailed a survey to each of the sample’s physicians asking, among other things, whether they had provided abortions in 2014 and, if so, how many. If a physician did not respond to any of three mailings, we made up to three attempts to contact the individual by phone. During fielding, we determined that 74 physicians in the sample were already known to us to provide abortion services, and 432 physicians were deceased, retired or unreachable because of incorrect contact information; 551 did not respond to mailings or telephone follow-up. The information we obtained from the responding 943 physicians was used to estimate the potential number of physicians’ offices and abortions missed by our survey.

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

Analysis

Census Bureau data on the population of women aged 15–44 on July 1, 2013, and July 1, 2014, were used as denominators for calculating abortion rates for the entire United States and for each state and the District of Columbia.[20] We estimated the national abortion ratio as the proportion of pregnancies (excluding those ending in miscarriages) that ended in abortion. 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 2013 and 2014 (to match conception times for births with those for abortions).[21-23]

We distinguished among four types of abortion-providing facilities: abortion clinics, nonspecialized clinics, hospitals and physicians’ offices. Abortion clinics are defined as non-hospital facilities in which half or more of patient visits are for abortion services, regardless of annual abortion case-load. Nonspecialized clinics are nonhospital sites in which fewer than half of patient visits are for abortion services. Physicians’ offices are defined as facilities that provide fewer than 400 abortions per year and have names suggesting that they are private practices. Physicians’ offices that provide 400 or more abortions per year were categorized as nonspecialized clinics; because of their relatively large case-load, we assume that their service provision more closely mirrors that of a nonspecialized clinic.

Eighty percent of nonhospital facilities provided information on early medication abortion, 60% provided information on self-induced abortion and 63% answered items about lost service days. Response rates to these measures varied by facility type and caseload, and we constructed weights to account for these differences.

Our analysis takes a particularly close look at states that experienced the largest changes in clinics of both types between 2011 and 2014. Specifically, we examined the 10 states that experienced the proportionately largest declines in clinics and the 10 that exhibited the largest increases, and compared three measures: the percentage change in abortion rate between 2011 and 2014, the number of abortion restrictions enacted between 2012 and 2014, and whether the state had a TRAP law. Information on state laws came from the Guttmacher Institute.[24-27] Appendix Table 1 (Supporting Information) provides a list of laws and the states in which they were implemented.

RESULTS

Abortion Incidence

The number of abortions and the abortion rate declined steadily between 2011 and 2014, by 3–6% per year (Table 1). In 2014, there were 926,200 abortions, and the abortion rate was 14.6 abortions per 1,000 women aged 15–44. This is the lowest rate since abortion was legalized nationally in 1973.[3] Nineteen percent of pregnancies (excluding miscarriages) ended in abortion in 2014, a decrease of 11% from the 2011 figure of 21%.

TABLE 1

Number of reported abortions, abortion rate and abortion ratio, United States, 1995–2014

Year	No. (in 000s)	Rate*	Ratio†
1995	1,359.4	22.5	25.9
1996	1,360.2	22.4	25.9
1997	[1,335.0]	[21.9]	[25.5]
1998	[1,319.0]	[21.5]	[25.1]
1999	1,314.8	21.4	24.6
2000	1,313.0	21.3	24.5
2001	[1,291.0]	[20.9]	[24.4]
2002	[1,269.0]	[20.5]	[23.8]
2003	[1,250.0]	[20.2]	[23.3]
2004	1,222.1	19.7	22.9
2005	1,206.2	19.4	22.4
2006	[1,242.2]	[19.9]	[22.9]
2007	1,209.6	19.4	21.9
2008	1,212.4	19.4	22.5
2009	[1,151.6]	[18.5]	[22.2]
2010	1,102.7	17.7	21.7
2011	1,058.5	16.9	21.2
2012	[1,011.0]	[16.1]	[20.4]
2013	958.7	15.2	19.4
2014	926.2	14.6	18.8

Abortions per 1,000 women aged 15–44 as of July 1 of each year.

Abortions per 100 pregnancies ending in abortion or live birth; for each year, the ratio is based on births occurring during the 12-month period starting in July of that year.

Note: Figures in brackets were estimated by interpolation of numbers of abortions and adjustments made to state health department reports. Sources: Number of abortions, population data and birth data, 1995–2011—references 2 and 3. Number of abortions, 2012—2010–2011 Guttmacher Abortion Provider Census and interpolations. Population data, 2012–2014—reference 20. Birth data, 2012–2015—references 21–23.

Overall, the abortion rate declined 14% between 2011 and 2014, while the number of abortions dropped by 12% (Table 2). The rate declined in almost all states. Increases were generally small (1–4% in Arkansas, Kansas, Michigan, Mississippi, North Carolina and Vermont), but the District of Columbia had an increase of 15%. Three states experienced a decline in the abortion rate that was at least twice the national decline (Delaware, 41%; Hawaii, 33%; and Texas, 28%).

TABLE 2

Number of reported abortions and abortion rate, selected years; and percentage change in rate, 2011–2014—all by region and state in which the abortions occurred

Region and state	Number			Rate*
	2011	2013	2014	2011	2013	2014	% change, 2011–2014
U.S. total	1,058,490	958,700	926,190	16.9	15.2	14.6	−14
Northeast	272,020	195,330	240,320	24.6	17.7	21.8	−11
Connecticut	14,640	13,320	13,140	21.3	19.4	19.2	−10
Maine	2,360	2,190	2,220	9.9	9.3	9.5	−4
Massachusetts	24,030	20,890	21,020	17.8	15.3	15.3	−14
New Hampshire	3,200	2,320	2,540	12.9	9.5	10.4	−19
New Jersey	46,990	43,670	44,460	27.1	25.3	25.8	−5
New York	138,370	119,170	119,940	34.2	29.4	29.6	−13
Pennsylvania	36,870	32,370	32,030	15.1	13.4	13.3	−12
Rhode Island	4,210	3,550	3,580	19.8	16.9	17.0	−14
Vermont	1,370	1,490	1,400	11.7	12.8	12.1	3
Midwest	153,380	143,650	138,940	11.7	10.9	10.6	−9
Illinois	44,580	43,400	42,270	17.0	16.6	16.3	−4
Indiana	9,430	8,600	8,180	7.3	6.7	6.3	−14
Iowa	5,640	4,900	4,380	9.7	8.4	7.5	−23
Kansas	6,940	7,470	7,240	12.5	13.4	12.9	4
Michigan	29,190	28,800	29,120	15.3	15.2	15.4	1
Minnesota	11,140	10,030	9,760	10.7	9.5	9.3	−13
Missouri	5,820	5,470	5,130	5.0	4.7	4.4	−12
Nebraska	2,570	2,230	2,280	7.2	6.2	6.3	−13
North Dakota	1,250	1,180	1,260	9.5	8.4	8.7	−8
Ohio	28,590	24,560	22,730	12.9	11.1	10.3	−20
South Dakota	600	600	550	3.9	3.8	3.5	−11
Wisconsin	7,640	6,410	6,050	7.0	5.9	5.6	−21
South	356,790	330,790	308,060	15.2	13.9	12.9	−16
Alabama	9,550	8,420	8,020	10.0	8.8	8.3	−16
Arkansas	4,370	4,470	4,590	7.6	7.8	8.0	4
Delaware	5,090	3,190	3,010	28.4	17.8	16.7	−41
District of Columbia	4,750	5,650	5,820	28.5	32.2	32.7	15
Florida	84,990	77,020	75,990	23.7	21.1	20.6	−13
Georgia	34,910	33,550	33,000	16.8	16.0	15.7	−7
Kentucky	3,970	3,700	3,530	4.6	4.3	4.1	−11
Louisiana	12,210	9,890	10,150	13.1	10.5	10.8	−18
Maryland	34,260	29,360	28,140	28.6	24.5	23.4	−18
Mississippi	2,220	2,210	2,290	3.7	3.7	3.8	3
North Carolina	28,600	30,550	29,960	14.6	15.5	15.1	3
Oklahoma	5,860	5,770	5,330	7.9	7.6	7.0	−12
South Carolina	6,620	6,350	6,040	7.1	6.8	6.4	−10
Tennessee	16,720	15,180	13,880	13.1	11.8	10.7	−18
Texas	73,200	67,530	55,230	13.5	12.1	9.8	−28
Virginia	27,110	25,890	21,080	16.3	15.4	12.5	−23
West Virginia	2,390	2,070	2,020	7.0	6.1	6.0	−14
West	276,300	245,290	238,860	18.5	16.2	15.6	−16
Alaska	1,820	1,570	1,470	12.4	10.7	10.0	−20
Arizona	16,100	13,580	12,870	12.7	10.5	9.8	−22
California	181,730	164,190	157,350	23.0	20.5	19.5	−15
Colorado	14,710	12,870	13,160	14.2	12.0	12.1	−14
Hawaii	5,580	4,250	3,760	21.1	15.9	14.0	−33
Idaho	1,680	1,350	1,320	5.4	4.3	4.2	−23
Montana	2,220	1,840	1,690	12.3	10.0	9.1	−26
Nevada	11,290	9,810	10,970	20.6	17.5	19.4	−6
New Mexico	5,180	4,200	4,650	13.0	10.5	11.7	−10
Oregon	10,690	9,130	9,330	14.1	11.9	12.0	−15
Utah	3,290	3,170	2,960	5.4	5.0	4.6	−14
Washington	21,880	19,190	19,230	16.0	13.8	13.7	−14
Wyoming	120	140	120	1.1	1.3	1.1	−4

Abortions per 1,000 women aged 15–44. Note: Numbers of abortions are rounded to the nearest 10. Sources: See Table 1.

Abortion rates dropped in all four regions of the country, but declines were steeper in the West and the South (16% each) than in the Midwest (9%) and the Northeast (11%). Following previous patterns, the Northeast maintained the highest abortion rate, followed by the West, the South and the Midwest. The five highest abortion rates were in the District of Columbia, New York, New Jersey, Maryland and Florida; rate declines in the four states ranged from 5% to 18%.

Facility Type and Numbers

In 2014, a total of 1,671 facilities provided at least one abortion (Table 3). Abortion clinics accounted for 16% of all facilities but for 59% of all abortions (Figure 1), largely because a majority of these facilities had caseloads of 1,000 or more per year. Nonspecialized clinics represented 31% of all known abortion-providing facilities and accounted for 36% of abortions. While many of these clinics primarily serve contraceptive and family planning clients, about half provided 400 or more abortions per year. Thirty-eight percent of facilities were hospitals; however, most of these provided fewer than 30 abortions per year, and these types of facilities accounted for 4% of all abortions. Finally, private physicians’ offices made up 15% of abortion facilities and provided 1% of procedures.

TABLE 3

Number of abortion-providing facilities, by type of facility and caseload, 2011 and 2014; and percentage change between these years

Facility type and caseload	2011	2014	% change
Total	1,720	1,671	−3
1–29	610	659	8
30–399	534	477	−11
400–999	227	247	9
1,000–4,999	329	269	−18
≥5,000	20	19	−5
Abortion clinics	329	272	−17
1–29	0	1	na
30–399	20	23	15
400–999	50	51	2
1,000–4,999	244	180	−26
≥5,000	15	17	13
Nonspecialized clinics	510	516	1
1–29	50	52	4
30–399	216	198	−8
400–999	158	178	13
1,000–4,999	81	86	6
≥5,000	5	2	−60
Hospitals	595	638	7
1–29	400	463	16
30–399	172	154	−10
400–999	19	18	−5
1,000–4,999	4	3	−25
≥5,000	0	0	na
Physicians’ offices*	286	245	−14
1–29	160	143	−11
30–399	126	102	−19
400–999	na	na	na
1,000–4,999	na	na	na
≥5,000	na	na	na

Offices that reported 400 or more abortions a year were classified as non-specialized clinics. Note: na=not applicable. Source: 2011 data—reference 2.

FIGURE 1

Percentage of abortion-providing facilities accounted for by each facility type, and percentage of abortions that are performed in each type of facility, 2014

The total number of abortion facilities declined 3% between 2011 and 2014, but the drop was particularly marked among abortion clinics (17%), especially those with caseloads of 1,000–4,999 per year (26%). The number of nonspecialized clinics that provided abortions remained stable. The 7% increase in the number of hospitals providing abortions was likely due to the fact that the 2014 survey did a better job of capturing abortions provided for reasons of fetal or maternal indications. The number of physicians’ offices providing abortions declined by 14% between 2011 and 2014.

Six percent fewer clinics were providing abortions in 2014 than in 2011 (Table 4). The number of clinics decreased in 25 states, remained stable in 14 states and the District of Columbia, and increased in 11 states. In 2014, five states—Mississippi, Missouri, North Dakota, South Dakota and Wyoming[*]—had only one clinic that provided abortion services.

TABLE 4

Number of abortion-providing facilities and number of clinics, 2011 and 2014, and percentage change between these years; number of counties and percentage with no clinic, 2014; and percentage of women aged 15–44 living in counties with no clinic, 2014—all by region and state

Region and state	All facilities			Clinics			Counties, 2014		% of women in counties with no clinic, 2014*
	2011	2014	% change	2011	2014	% change	No.	% with no clinic
U.S. total	1,720	1,671	−3	839	788	−6	3,142	90	39
Northeast	453	476	5	186	212	14	217	60	23
Connecticut	41	59	44	21	25	19	8	13	5
Maine	11	9	−18	5	4	−20	16	81	55
Massachusetts	40	43	8	12	14	17	14	43	14
New Hampshire	13	12	−8	5	4	−20	10	60	30
New Jersey	64	79	23	24	41	71	21	33	23
New York	225	218	−3	94	95	1	62	44	10
Pennsylvania	47	42	−11	20	20	0	67	85	48
Rhode Island	4	5	25	2	3	50	5	80	36
Vermont	8	9	13	3	6	100	14	64	38
Midwest	173	142	−18	124	97	−22	1,055	95	55
Illinois	37	40	8	26	24	−8	102	92	40
Indiana	12	11	−8	10	9	−10	92	95	66
Iowa	18	13	−28	17	12	−29	99	89	42
Kansas	3	4	33	3	4	33	105	97	56
Michigan	41	29	−29	30	20	−33	83	89	40
Minnesota	15	11	−27	7	6	−14	87	95	59
Missouri	5	2	−60	4	1	−75	115	99	94
Nebraska	5	5	0	3	3	0	93	97	41
North Dakota	1	1	0	1	1	0	53	98	73
Ohio	26	17	−35	18	12	−33	88	93	56
South Dakota	2	2	0	1	1	0	66	98	77
Wisconsin	8	7	−13	4	4	0	72	96	67
South	357	336	−6	245	214	−13	1,422	93	51
Alabama	8	9	13	6	5	−17	67	93	59
Arkansas	5	4	−20	3	3	0	75	97	77
Delaware	8	6	−25	4	3	−25	3	33	18
District of Columbia	9	9	0	5	5	0	1	0	0
Florida	88	86	−2	72	71	−1	67	70	20
Georgia	28	28	0	19	17	−11	159	96	58
Kentucky	3	3	0	2	2	0	120	98	74
Louisiana	7	5	−29	7	5	−29	64	92	63
Maryland	34	41	21	21	25	19	24	67	24
Mississippi	2	2	0	1	1	0	82	99	91
North Carolina	36	37	3	21	16	−24	100	90	53
Oklahoma	5	5	0	3	3	0	77	96	54
South Carolina	9	7	−22	3	3	0	46	93	71
Tennessee	14	11	−21	9	7	−22	95	96	63
Texas	62	44	−29	46	28	−39	254	96	43
Virginia	35	34	−3	21	18	−14	133	92	78
West Virginia	4	5	25	2	2	0	55	98	90
West	737	717	−3	284	265	−7	448	80	17
Alaska	9	8	−11	4	3	−25	29	90	37
Arizona	17	12	−29	15	9	−40	15	80	19
California	512	512	0	160	152	−5	58	43	5
Colorado	42	36	−14	24	21	−13	64	78	27
Hawaii	33	29	−12	6	4	−33	5	40	5
Idaho	4	5	25	2	3	50	44	95	68
Montana	8	5	−38	7	5	−29	56	93	55
Nevada	14	13	−7	8	8	0	17	88	9
New Mexico	12	11	−8	7	9	29	33	91	48
Oregon	29	27	−7	15	15	0	36	78	30
Utah	9	6	−33	4	2	−50	29	97	62
Washington	45	50	11	32	33	3	39	64	15
Wyoming	3	3	0	1	1	0	23	96	96

Population counts are for July 1, 2014.[20] Source: 2011 data—reference 2.

The Northeast was the only region that had more clinics providing abortion services in 2014 than in 2011; the number increased by 14%, largely because of an increase from 24 to 41 in New Jersey. Most of the New Jersey facilities were nonspecialized clinics that started offering early medication abortion (not shown). The Midwest experienced the largest decline in clinics—22%, or 27 clinics. Most of this decrease was accounted for by Michigan (which lost 10 clinics), Ohio (six) and Iowa (five). The South also experienced a substantial decline in clinics (13%); more than half of this was accounted for by the loss of 18 clinics in Texas. In the West, the number of clinics fell by 7%.

In 2014, some 90% of U.S. counties had no clinics that provided abortion care, and 39% of women aged 15–44 lived in those counties; these numbers were essentially unchanged from the 2011 figures of 89% and 38%, respectively.[2] Access to abortion services appeared to be best in California, Connecticut, the District of Columbia, Hawaii and Nevada, where fewer than 10% of women lived in a county without a clinic. More than 90% of women residing in Mississippi, Missouri or Wyoming lived in a county without a clinic.

Of the 943 physicians from whom we obtained information, 58, or 6%, reported providing an average of 25 abortions per year. Assuming the respondents were representative of the universe of 33,379 practicing obstetrician-gynecologists in the United States,[28] our main survey missed 2,069 physicians, who collectively provided 51,725 abortions in 2014. If this estimate is correct, our study undercounted the total number of abortions by 5%.

Early Medication Abortion

In 2014, an estimated 272,400 early medication abortions were provided in nonhospital facilities, representing a 14% increase since 2011 (Table 5). Early medication abortions accounted for 31% of all nonhospital abortions, compared with 24% in 2011. Half or more of all abortions (50–68%) provided by facilities with annual caseloads of fewer than 400 procedures were early medication abortions.

TABLE 5

Number of early medication abortions provided at nonhospital facilities, 2011 and 2014, and percentage change between these years; and these abortions as a percentage of all nonhospital abortions—all by facility type and by caseload

Facility type and caseload	Number			% of all nonhospital abortions
	2011	2014	% change	2011	2014
Total	239,400	272,400	14	24*	31
Facility
Abortion clinics	123,400	126,300	2	18	23
Nonspecialized clinics	111,100	140,200	26	33	42
Hospitals	u	u	u	u	u
Physicians’ offices	4,900	5,900	20	35	46
Caseload
1–29	1,000	1,100	10	43*	50
30–399	31,700	37,000	17	54*	68
400–999	41,700	66,000	58	30*	44
1,000–4,999	142,500	144,100	1	21	26
≥5,000	22,600	24,200	7	17	19

Previously published figure was incorrect and has been adjusted to exclude abortions performed in hospitals. Notes: Early medication abortions are those performed with mifepristone, methotrexate or misoprostol alone. Numbers of abortions are rounded to the nearest 100. u=unavailable. Source: 2011 data—reference 2.

We assumed that the majority of facilities provided medication abortion up to nine weeks’ gestation; this was the gestational limit defined on the survey. Using gestational data from the Centers for Disease Control and Prevention, we estimated that 45% of abortions up to nine weeks’ gestation in 2014 were early medication procedures (not shown), up from approximately 36% in 2011.[*] While other drugs can be administered in early medication abortion, the overwhelming majority—97%—were done with mifepristone; the remaining procedures used methotrexate or misoprostol alone.

The use of early medication abortion increased across all facility types and sizes between 2011 and 2014. While abortion clinics saw only a 2% increase, nonspecialized clinics and physicians’ offices reported increases of 26% and 20%, respectively. Though nonspecialized clinics accounted for 36% of all abortions, they provided 51% of all early medication abortions.

For 2014, we estimated that 900 nonhospital facilities—87% of all nonhospital provider sites— provided early medication abortion (not shown). The overwhelming majority of both abortion clinics (99.8%) and nonspecialized clinics (88%) offered this service. At least 23% of all nonhospital facilities offered only early medication abortions; most of these facilities were non-specialized clinics, and 38% of such clinics offered only early medication abortion. Most facilities that provided only early medication abortions were located in areas that were also served by facilities offering surgical abortion, though 11 were the sole abortion-providing facility in the metropolitan statistical area in which they were located.

State Abortion Policy Context

Abortion restrictions were associated with a decrease in the number of abortion and nonspecialized clinics, but fluctuations in clinic numbers—whether decreases or increases—were not clearly associated with abortion rates. Eight of the 10 states that experienced the largest proportionate declines in the number of clinics over the period 2011–2014 implemented at least one new abortion restriction between 2012 and 2014, and six enacted three or four (Table 6). Four states—Arizona, Michigan, Ohio and Texas—implemented TRAP regulations during this period, and four had TRAP laws in place prior to 2012.[29] Hawaii and Iowa experienced substantial declines in the number of clinics providing abortions, but enacted no new abortion restrictions and, along with Montana, had no TRAP laws.

TABLE 6

Number of clinics providing abortions in 2011 and 2014, and percentage change between these years; percentage change in abortion rates, 2011–2014; number of new abortion restrictions enacted in 2012–2014; and status of TRAP laws—all among states with the largest proportionate changes in numbers of clinics

State	No. of clinics			% change in abortion rate, 2011–2014	No. of restrictions enacted in 2012–2014	TRAP laws
	2011	2014	% change			Existed prior to 2012	Enacted in 2012–2014
Total	839	788	−6	−14	84	46	9
States with largest declines
Missouri	4	1	−75	−12	3	X
Utah	4	2	−50	−14	1	X
Arizona	15	9	−40	−22	4		X
Texas	46	28	−39	−28	3	X	X
Hawaii	6	4	−33	−33	0
Michigan	30	20	−33	1	3		X
Ohio	18	12	−33	−20	3		X
Iowa	17	12	−29	−23	0
Louisiana	7	5	−29	−18	4	X
Montana	7	5	−29	−26	1
States with largest increases
Vermont	3	6	100	3	0
New Jersey	24	41	71	−5	0
Idaho	2	3	50	−23	0
Rhode Island	2	3	50	−14	0	X
Kansas	3	4	33	4	4
New Mexico	7	9	29	−10	0
Connecticut	21	25	19	−10	0	X
Maryland	21	25	19	−18	0
Massachusetts	12	14	17	−14	0
Washington	32	33	3	−14	0

Note: TRAP=targeted regulation of abortion providers.

Seven of the states that experienced the largest proportionate declines in clinics also experienced a larger than average decline in the abortion rate. However, the states with the two largest proportionate declines in clinic numbers—Missouri and Utah—experienced declines in the abortion rate that were comparable to the national average. And while Michigan had 33% fewer clinics in 2014 than in 2011, the state’s abortion rate increased slightly.

Patterns among the 10 states with the largest proportionate increases in the number of clinics further demonstrated that changes in clinic numbers and abortion rates were not clearly related. The abortion rate declined in eight of these states and increased slightly (3–4%) in two. Only one of these 10 states enacted new abortion restrictions between 2012 and 2014; Kansas enacted four new provisions, even as one new facility opened. Two states—Rhode Island and Connecticut—had implemented TRAP laws prior to 2012.

Other Potential Impacts o f Restrictions

Overall, 4% of nonhospital facilities (located in 15 states) reported being unable to provide care on one or more days in order to be in compliance with local or state regulations (not shown). These closures presumably occurred in response to state inspections, as well as TRAP and other types of laws. Facilities in the South were the most likely to report lost days (15%),[*] followed by those in the Midwest (5%). Only 1% of nonhospital facilities in the Northeast and the West reported lost service days.

Twelve percent of nonhospital facilities reported treating at least one patient who had attempted to end her pregnancy on her own in 2014 (not shown). This figure included facilities in 30 states, but there were regional variations: The proportion of facilities that had treated such patients was higher in the South and the Midwest (21% and 16%, respectively) than in the West (10%) and the Northeast (8%). More than two-thirds of facilities were able to quantify this caseload, and three-quarters reported having treated 10 or fewer of these patients in 2014.

DISCUSSION

The long-term decline in abortion continued through 2014; in that year, the abortion rate and ratio were both lower than those documented in 1973, the year abortion was legalized nationwide. The drop in abortion was found in all but six states and the District of Columbia, though there was substantial variation across states. It is beyond the scope of this descriptive study to explore the larger dynamics responsible for these patterns, but we suggest several potential factors that may have contributed to some of the observed patterns.

Abortion incidence can be affected by service availability. The overwhelming majority of abortions—95%—are provided by abortion clinics and nonspecialized clinics, and 6% fewer clinics provided abortions in 2014 than in 2011. Declines in the numbers of clinics were most pronounced in the Midwest (22%) and the South (13%), regions of the country where the majority of new abortion restrictions were enacted during the study period.[26,27] However, regional patterns in clinic numbers did not always correspond with changes in abortion rates, as rate declines were steepest in the West and the South (both 16%) and lowest in the Midwest (9%). And while the Northeast had more clinics providing abortion care in 2014 than in 2011, that region’s abortion rate declined 11%. These patterns demonstrate that the relationship between abortion access, as measured by numbers of clinics, and abortion rates is not straightforward. These trends may obscure more nuanced relationships between clinics with the largest caseloads and abortion rates. For example, abortion clinics provided the majority of abortions in 2014 (59%), but the number of such clinics with annual caseloads of 1,000–4,999 declined by 26%.

Six of the 10 states that experienced the greatest proportionate declines in the number of clinics enacted three or four new abortion restrictions between 2012 and 2014, and four of these states implemented new TRAP laws. At the same time, however, two of the 10 states did not enact new restrictions during this period, and three had no TRAP laws. Thus, restrictions alone do not account for declines in clinic numbers. Michigan and Ohio both experienced a 33% decline in the number of clinics providing abortion services, and both implemented multiple restrictions, including TRAP laws, between 2012 and 2014. However, some Michigan facilities located near the Ohio border were able to both meet the TRAP requirement and expand services to meet increased demand from women in the neighboring state.[30] This may have been one reason why the abortion rate in Michigan increased slightly.

Hawaii and Iowa experienced larger than average declines in both numbers of clinics and abortion rates, and neither had TRAP laws or enacted new abortion restrictions. The decline in abortion incidence in Iowa may have been due to reduced levels of unintended pregnancy. In 2006, Iowa expanded access to family planning for low-income women, and between 2007 and 2013, a privately funded initiative sought to increase access to LARC methods among the same population.[31] Areas of the state with the most exposure to these programs experienced larger declines in abortion than did those with less exposure.[31] We are not aware of similar programs or initiatives in Hawaii, and it is unclear what accounted for the changes there. The loss of two abortion-providing clinics may have reduced access to the procedure. However, only 5% of women in Hawaii lived in a county without a clinic in 2014, so this seems unlikely. Alternately, because 19% of abortions in Hawaii were estimated using information from sources other than the facilities where the abortions were performed, we may have underestimated the number of procedures that occurred, and the actual change in abortion incidence may be smaller (or larger).

At least some of the decline in abortion across the United States likely is due to declines in unintended pregnancy. Use of LARC methods increased 36% among all women using contraceptives between 2009 and 2012,[5] and 48% among clients at Title X–supported family planning facilities.[6] Clients at the latter facilities are disproportionately young adults and lower income women, two groups who are at the highest risk for unintended pregnancy.[1] Additionally, reliance on LARC methods has been increasing since 2002,[4] and this may have had a cumulative effect. For example, the most commonly used IUD is effective for five (or more) years, and the implant for three; if a majority of women who received LARC methods in 2011 were still using them in 2014, they would have been protected from unintended pregnancy during the entire study period.

Limitations

We are aware of several shortcomings of our study. We obtained responses from only 58% of facilities that we believed provided abortions in 2014. Health department data, which can be of variable quality, were used to determine caseloads for 20% of facilities, and we estimated caseloads for 17%. Moreover, some states required more estimation than others. The overwhelming majority of the abortions that we counted—88%—were based on information given to us by providers, but the fact that 12% were estimated using other sources may introduce inaccuracy into the data.

Undoubtedly, some abortion-providing facilities were not counted because we were unable to identify them. Our survey of a random sample of obstetrician-gynecologists suggests that some 2,000 physicians who provide abortions in their private practice were not captured in our study; hence our estimate of abortions may be 5% lower than it would have been if these providers had been included.

While we are confident that our study captured most, if not all, clinics that provide abortion services, we are aware that some of the information in this study is already out of date. The media have closely monitored clinic closures (and reopenings) in Texas; as of June 2016, some 18–21 clinics were providing abortion services there,[32] whereas we counted 28. In part, this difference is due to our inclusion of facilities that provided at least one abortion in 2014; a number of clinics closed at some point in that year, many in response to the TRAP laws,[11,33] and more facilities closed in 2015.[34,35] Substantial numbers of clinics may have closed in other states as well, while other facilities may have opened (or reopened).

Some evidence suggests that women who live in states that have restrictive abortion laws are turning to self-induced abortion,[18,19] but our study did not capture abortions that occurred outside of medical settings.[36] Overall, 12% of nonhospital facilities had seen at least one patient who had attempted to end her pregnancy on her own in 2014, and reports of self-induction were greater in the South and the Midwest, regions that had higher numbers of abortion restrictions.[26,27] If reliance on self-induced abortion was higher in 2014 than in 2011, the observed decline in abortion would not be as large as our findings suggest.

Our abortion counts and rates are tabulated by state of occurrence. The five states with the lowest abortion rates were Wyoming, South Dakota, Mississippi, Kentucky and Idaho. Data from the Centers for Disease Control and Prevention suggest that in 2012, some 14% of abortion patients who lived in South Dakota went out of state for services, as did more than 90% of abortion patients who lived in Wyoming.[37] Thus, the abortion numbers and rates in this study do not always reflect abortion incidence among women who reside in a given state.

Conclusions

Abortion is an important indicator of unintended pregnancy, but it is unclear whether the most recent decline in abortion is due to fewer women’s having unintended pregnancies, more women’s being unable to access abortion services or some combination of these dynamics. That there were fewer clinics providing abortions in 2014 than in 2011 could be attributed to several factors. In some states, increased abortion restrictions likely contributed to the decline in abortions, but in others, the decline may have been driven by a drop in demand. For example, increased reliance on LARC methods likely contributed to the decline in unintended pregnancy that occurred between 2008 and 2011.[1] If women continued to experience fewer unintended pregnancies in 2014, there may have been less need for abortion services and, in turn, decreased need for providers. Future research should aim to elucidate patterns in abortion incidence and numbers of abortion-providing facilities and identify reasons for change. In addition, because state legislatures continued to pass new abortion restrictions in 2015 and 2016, subsequent research will need to monitor the accessibility of abortion services.