Abortion is one of the most common medical procedures undergone by women aged 15–44 in the United States,¹ partly because of the high level of unintended pregnancy. About half of the 6.4 million pregnancies that occurred in 2000 (including those ending in miscarriages) were unintended, and about half of these resulted in abortion.² No comprehensive study has examined abortion incidence and service provision since 2000.³ Updated information is needed, both because it is important to monitor this key reproductive behavior and because changes that have occurred since 2000 may affect the need for and access to abortion services.

The number of abortions in the United States declined from 1.61 million (the all-time high) in 1990 to 1.31 million in 2000. Similarly, the abortion rate declined from 27 per 1,000 women aged 15–44 in 1990 to 21 per 1,000 in 2000, a level comparable to levels of the mid-1970s. Information from the Centers for Disease Control and Prevention suggests that declines in both the number of abortions and abortion rates slowed between 2000 and 2003,⁴ but definitive information is not available on recent trends in incidence.

Trends in abortion incidence may be affected by changes in access to services. Between 1982 and 2000, the number of abortion providers declined by about 38%, from a high of 2,900 to 1,800.^3,5 Some of this decline was due to a shift from hospital-based providers to specialized abortion clinics, a trend that may offer greater accessibility, because, for example, abortion clinics generally charge less than other types of providers,⁶ and many advertise, making them easier to locate. But the decline in the number of providers may mean that some women have a more difficult time locating and affording services. The proportion of counties without an abortion provider increased from 77% in 1978 to 87% in 2000, and the proportion of women of childbearing age residing in these counties increased from 27% to 34%.³

One important change affecting access to abortion services occurred in September 2000, when the Food and Drug Administration approved mifepristone (also known as RU-486) for use for early medication abortion. In early 2001, mifepristone, the main drug used for early medication abortion, accounted for only 6% of all abortions, and most early medication abortions were provided by clinics that also offered surgical abortions.³ As knowledge about and comfort with mifepristone has increased, it likely has been introduced into settings where surgical abortions were previously not provided (e.g., family planning clinics and the practices of family doctors),⁷ possibly increasing access to abortion and reversing the trend of services’ being concentrated in clinics and larger providers.

At the same time, during the last several years, a number of states have implemented restrictions that may have made it more difficult for women to access abortion services and for physicians to perform abortions. For example, between 2000 and 2004, five states enacted laws that impose burdens on abortion providers.⁸ These restrictions range from requiring abortions after 15 weeks to be provided in a licensed surgical center to requiring providers to have expensive ultrasound equipment on-site.

Finally, recent patterns in unintended pregnancy may have affected abortion incidence and services. Numbers and rates of adolescent pregnancies continued to decline between 1995 and 2002, largely because of improved contraceptive use among adolescents,⁹ and fewer adolescents have needed to access abortion services. Overall levels of unintended pregnancy among women aged 20 and older remained stable or increased between 1994 and 2001, but this general picture masks important subgroup differences. Rates of unintended pregnancy and abortion increased for poor and low-income women during this period,² and if these trends continued, the overall incidence of abortion could have increased by 2005.

This article addresses these issues by presenting new information on abortion incidence and access to services in 2005, based on the Guttmacher Institute’s 14th survey of all known abortion providers in the United States. While most state health departments collect abortion statistics, the Guttmacher surveys, which have been conducted periodically since 1974, have produced the most complete available data on the number and geographic distribution of abortions and abortion providers, the types of facilities offering services and other aspects of abortion accessibility.

Methods

Identifying providers

Before fielding the survey, we conducted an extensive update of our list of U.S. facilities where abortions are performed. We began with the providers^* known to have performed abortions in 2000, excluding those known to have stopped or to have closed. We identified possible new providers from a variety of sources: searches of the telephone yellow pages for the entire country, the membership directory of the National Abortion Federation, provider listings on the Internet and miscellaneous other sources. Additional possible providers were identified during the fielding period. The updated list included 2,310 potential providers.

To increase coverage of small providers that offered only early medication abortion, we enlisted the company that is the sole distributor of mifepristone in the United States to mail our questionnaire to providers thought to have purchased the drug for providing abortions. The distributor did not identify the providers, and we were able to include them only if they responded to the survey and supplied their contact information. This strategy allowed us to identify 62 new providers, most of which provided only medication abortion.

Questionnaire content and fielding

The questionnaire was modeled on the instrument used in the previous survey, which was conducted in 2001–2002 and collected data for 1999, 2000 and the first half of 2001.³ All providers were asked the number of induced abortions they performed in 2004 and 2005, the minimum and maximum gestations at which they will perform surgical abortions and medication abortions, and whether they offered early medication abortion. Clinics and physician providers (but not hospital providers) were also asked the number of early medication abortions performed (with separate items for methotrexate and mifepristone), gestational limits for medication abortion, charges, distance traveled by clients and the proportion of provider services accounted for by abortions.^† We asked fewer questions of hospitals because the individuals answering the questionnaires in these settings typically have access to less information about clients. While some of the information we present is restricted to nonhospital facilities, the results represent the experience of most women having abortions, since these providers performed 95% of all abortions in 2000.³

In July 2006, we mailed questionnaires to all potential providers we had identified. Respondents could return the survey via mail or respond through a secure Web site. Providers that did not respond to the first mailing were sent two additional mailings at three-week intervals. In September and October 2006, the distributor of mifepristone sent questionnaires to approximately 1,200 providers (most of which were likely already in our database).

We also contacted state health statistics agencies, requesting all available data on the number of abortions reported for 2004 and 2005. Forty-six states and the District of Columbia provided at least some information.^‡ A few states publish abortion data by individual provider, but we used these figures only if providers did not respond to our mailings or, in a few instances, if the number from the state was very different from a provider’s report and we suspected that the provider-supplied information was inaccurate.

Intensive telephone follow-up of nonrespondents was carried out from October 2006 to July 2007, with particular effort made to obtain the total number of surgical and medication abortions performed in 2004 and 2005. In total, more than 6,200 contacts were made with approximately 1,000 providers.

Of the 2,310 facilities surveyed, 916 responded to the mailed questionnaire (including 99 that responded via the Internet), and 636 responded via fax, mail, Internet or phone during follow-up; for 274 facilities, health department data were used. After additional follow-up with other sources, we determined that 22 nonresponding providers had closed or performed no abortions during the survey period, and 25 were duplicates of providers on our list. We were unable to confirm that abortions were provided by 48 facilities, and we did not count them as providers. For 59 of the remaining 389 potential providers, we obtained estimates of the number of abortions performed in 2004 and 2005 from knowledgeable sources in their communities; we made our own estimates for 330 facilities.

The level of internal estimation was higher than in prior years (in the 2001–2002 survey, abortions for 183 of 2,442 potential providers were estimated) because health department data for New York and California were less complete. In prior years, health departments in both states collected information on all abortions performed in hospitals, and we relied on these data when hospitals failed to respond to the survey. The New York State health department was able to release information only on abortions performed at hospitals in 2004, and we used this information to estimate abortions performed for 59 hospital facilities in 2005. Since 2002, the California state health department has limited abortion reporting to inpatient procedures. For hospitals in California that responded to our survey, we calculated the ratio of inpatient procedures reported by the health department to all procedures reported on our survey. For many of these hospitals, inpatient procedures accounted for most, if not all, abortions. However, in facilities with larger caseloads, inpatient procedures accounted for smaller proportions of all abortions. We calculated ratios of inpatient to outpatient procedures by caseload. We categorized the 112 nonresponding hospitals in California according to their caseloads reported in the 2001 survey and applied the appropriate ratios to inpatient procedures reported to the state health department. Finally, for the remaining 158 facilities that did not respond, we projected the number of abortions, usually on the basis of abortion counts reported by providers in the previous survey or by informants who could provide estimates only for years other than 2004 or 2005.

Of the abortions reported for 2005, 76% were reported by providers, 12% came from health department data, 9% were estimated by knowledgeable sources and 3% were projections or other estimates. These figures are almost identical to our 2000 results, when 77% of abortions were reported by providers, 10% came from health departments, 11% were external estimates and 2% were estimated internally.

Some abortion providers were excluded because we were unable to identify them. A past underreporting survey, based on a random sample of physicians and hospitals, suggested that the number of abortions in 1992 was 3–4% greater than the number we counted, and that we may have missed a number of small providers.¹⁰ This problem may have become more pronounced for this survey period because of the introduction and integration of mifepristone for early medication abortion at facilities that previously did not offer abortion services. Although questionnaires were distributed to health care professionals believed to have used mifepristone, some of these practitioners may have been reluctant to identify themselves as abortion providers, especially if they performed few abortions. Thus, we likely missed some providers who were offering small numbers of medication abortions. However, it is highly unlikely that facilities with larger caseloads were excluded or missed, since they typically are known by other providers in their communities and advertise in the yellow pages or on the Internet.

Analysis

We distinguish between four types of providers: hospitals, abortion clinics, other (nonspecialized) clinics and physicians’ offices. Abortion clinics are nonhospital facilities where half or more of patient visits are for abortion services; other clinics are sites where fewer than half of patient visits are for abortion services, including physicians’ offices that provide 400 or more abortions per year. Physicians’ offices are facilities that perform fewer than 400 abortions per year and have names suggesting that they are physicians’ private practices.

In addition to the number of abortions performed, the majority of the 1,182 nonhospital abortion providers reported on the number of early medication abortions they performed (72%), gestational limits (73%), charges (67%) and distance traveled by clients (61%). Because nonhospital facilities and facilities that perform 400 or more abortions per year were more likely to respond to the survey than were hospitals and small facilities, we weighted results to reflect the national proportions according to facility type and caseload. To account for item-specific nonresponse, we used different weights for variables measuring early medication abortion, gestational limits, charges and distance traveled. Unless otherwise noted, all abortion data presented include both surgical and medication procedures.

We used Census Bureau data on the population of women aged 15–44 for July 1, 2004, and July 1, 2005, as denominators for calculating abortion rates for the entire United States and for each state and the District of Columbia.^§ We estimated the national abortion ratio by combining 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, 2004, and July 1, 2005 (in order to match conception times for pregnancies ending in births with those for pregnancies ending in abortions).^**

We examined the overall distribution of providers by county and metropolitan area. The previous provider survey was based on 1999 definitions, but in 2000, the Census Bureau revised its definition of metropolitan areas.¹¹ In general, however, these continue to represent urbanized areas with populations of 50,000 or more. The current analysis is based on metropolitan area definitions from 2003.^††

Results

Abortion incidence

The number of abortions declined by 8% between 2000 and 2005, from 1.31 million to 1.21 million (Table 1). The last year in which the number of abortions was lower was 1976. Similarly, the abortion rate of 19.4 per 1,000 women aged 15–44 in 2005 represented a 9% decline over five years and was the lowest rate since 1974. Abortion rates declined faster between 2000 and 2005 than they had between 1996 and 2000 (5%). The abortion ratio indicates that 22% of pregnancies (excluding those ending in miscarriages) ended in abortion in 2005; this figure, too, represents the continuation of a long-term decline.

Abortion measures vary substantially by state and by region (Table 2, page 10). Abortion rates were highest in the District of Columbia, New York and New Jersey (34–54 per 1,000 women aged 15–44). Delaware, Florida, Maryland, California and Nevada also had relatively high rates (27 or more per 1,000 women). Rates were lowest (less than five per 1,000) in rural states that are less densely populated: Wyoming, Kentucky and Mississippi. Similarly, South Dakota, Idaho, Utah, West Virginia and Missouri had relatively low abortion rates (lower than seven per 1,000). These rates reflect the state in which the abortion occurred and may differ from rates at which residents obtained abortions, as some women cross state lines for abortion services. (For example, 25% of abortions in Delaware in 2003 were obtained by out-of-state residents.⁴)

As in prior years, the Northeast had the highest abortion rate, followed by the West, the South and the Midwest. Declines in the rate between 2000 and 2005 were most pronounced in the Midwest and the West (12%); the abortion rate dropped the least (3%) in the Northeast.

Within each region, state abortion rates varied. Despite an overall decline in abortion in the Northeast, the abortion rate increased for Connecticut and, to a lesser extent, Maine and New Hampshire. In the Midwest, abortion rates increased by 6–8% in Iowa and Missouri, but declined by 19–24% in Illinois and Nebraska. Maryland was the only state in the South where the abortion rate increased; several states in the region had substantial declines. Although the District of Columbia has the highest abortion rate in the United States, its rate declined by 20% between 2000 and 2005. Finally, in the West, the abortion rate increased in Alaska and New Mexico. It declined substantially in Oregon and Wyoming, although it was very low to begin with in the latter.

Trends in provider numbers

In all, 1,787 facilities provided abortions in the United States in 2005 (Table 3); that number represents a decline of just 2% from 2000. Provider numbers dropped at a much higher rate in the prior eight years—14% in 1992–1996 and 11% in 1996–2000.

Some 87% of U.S. counties, accounting for 35% of women, had no abortion provider in 2005. Women in the Northeast and the West, where the populations are concentrated in metropolitan areas, were less likely to live in a county without a provider (17% and 15%, respectively) than were women in the South and the Midwest (47% and 50%, respectively).

Between 2000 and 2005, the number of providers decreased in 26 states and the District of Columbia, increased in 15 states and remained stable in nine. The largest absolute increases occurred in the two states with the largest numbers of providers: New York had 27 more providers in 2005 than in 2000, and California had 24 more. Both states saw a decline in the number of specialized clinics (not shown), and in New York, the number of nonspecialized clinic providers increased, as did the number of physician providers, most of which performed fewer than 30 abortions in 2005. The overall increase in California’s number of providers was due largely to an increase in hospitals that provided a small number of abortions. Similarly, in Georgia, the 31% increase in provider numbers (a gain of eight facilities) was due to an increase in the number of hospitals providing a small number of abortions per year. While the increase in hospital providers in Georgia may be real, it may also reflect that the abortion data from the state health department was more detailed in 2005 than in prior years. The largest absolute decreases in abortion providers occurred in North Carolina (18), Pennsylvania (17) and Hawaii (12). Declines in Pennsylvania and North Carolina were mostly in the number of hospitals with small abortion caseloads, while the decline in Hawaii was due to fewer physicians’ performing, or reporting that they perform, abortions.

In several states where provider numbers increased, such as California, Georgia and New York, abortion rates decreased between 2000 and 2005. Three of the nine states where abortion rates increased (Maine, Maryland and New Hampshire) had a decrease in provider numbers, and three (Alaska, Colorado and Connecticut) had an increase of two or more. In states that had few providers to begin with, such as Arkansas, Mississippi and North Dakota, declines in the number of providers likely further restricted access to abortion services and, in turn, contributed to lower abortion rates.

Provider types and caseloads

•Clinics. The 381 specialized abortion clinics performed 69% of all abortions in 2005 (Table 4). A large majority of these reported 1,000 or more abortions during the year, and a few performed 5,000 or more. The 435 nonspecialized clinic providers performed 25% of abortions. Even though a majority of patient visits to these facilities are for other services, some nonspecialized clinics are similar to abortion clinics in that they have large abortion caseloads.

Between 2000 and 2005, abortion clinics declined both in number (by 15%, from 4473) and as a proportion of all providers (from 25% to 21%—not shown). Seventy-seven abortion clinics closed during this period, and only 29 new clinics opened. In addition, while 36 providers switched their focus and were reclassified as abortion clinics, 54 facilities were removed from that category because they changed their focus or did fewer abortions in 2005. The number of nonspecialized clinic providers increased by 13%, from 386 in 2000,³ partly because some clinics that previously had not offered abortion services began to offer medication abortions; this number also increased as a proportion of the total number of providers, from 21% to 24%. Overall, the number of clinics that performed 400 or more abortions fell from 668 to 617 between 2000 and 2005.³

As in 2000, 80% of all abortions took place in facilities that performed 1,000 or more abortions.

•Hospitals. One-third of identified abortion providers (604 facilities) were hospitals. Many hospitals provide abortions only in cases of fetal anomaly or serious risk to the woman’s health, and a majority (62%) performed fewer than 30 abortions during 2005. It is difficult to identify hospitals where abortions are performed only occasionally; our survey likely missed many such hospitals, especially in states where the health department does not release provider-specific data. Twenty hospitals reported 400–999 abortions, and only seven reported 1,000 or more. Together, hospitals accounted for 5% of all abortions, the same proportion as in 2000.³

•Physicians. One-fifth of providers were physicians’ offices. The 367 identified physicians’ offices represent a decline of 4% from the number located in 2000.³ Fifty-three percent of these facilities reported fewer than 30 abortions; as with hospitals, other such small providers may have been missed. Physicians’ offices performed nearly 22,000 abortions, or 2% of the total.

Early medication abortion

Early medication abortion services, which can use mifepristone or methotrexate, have expanded substantially since our last survey, which took place shortly after mifepristone became available. We estimate that 1,026 facilities (57% of abortion providers) performed one or more early medication abortions in 2005—70% more than had done so in 2001 (Table 5). In 2005, clinics were more likely to offer early medication abortion (78–81% did so) than were other types of providers (29–55%). However, since mid-2001, facilities other than abortion clinics were more likely to introduce this service. Similarly, we estimate that among providers with the smallest abortion caseloads, the number providing this service almost tripled between early 2001 and 2005.

A substantial number of clinics and physicians’ offices provided medication but not surgical abortions in 2005 (not shown). We identified 49 physicians’ offices, 67 nonspecialized clinics and three abortion clinics that offered medication abortions only; 13% of physicians’ offices known to perform abortions were in this group, as were 15% of nonspecialized clinics. Many of these facilities were not previously surveyed, and about a quarter were identified because they responded to the mailing by the distributor of mifepristone. Although a majority of the new providers were in areas that were also served by surgical providers, 11 were in nonmetropolitan areas, and 12 were in cities with no other services. Estimates related to facilities offering only early medication abortion are conservative, as we expect that some providers contacted in the distributor mailing did not respond and perform medication abortions only.^‡‡

We estimate that 161,100 early medication abortions were performed in nonhospital facilities in 2005. Mifepristone was used for 142,600, or approximately 90%, of these procedures (not shown). More than half of early medication abortions were provided by abortion clinics, and most of the rest by other clinics. While more than half of physicians’ offices performed medication abortions, the caseloads were small, averaging only 24 abortions per provider during the year.

The proportion of abortions performed medically is driven by provider practices and protocols, as well as patient preferences. Early medication abortions accounted for 14% of nonhospital abortions, or 13% of all abortions (not shown). We also calculated the proportion of "eligible" early abortions that were performed medically. Most protocols indicate that early medication abortion is recommended only up to 63 days of pregnancy,^12,13 but data indicating the distribution of early medication abortions by gestation are not available. We therefore used the total number of abortions before nine weeks of gestation as the denominator, and estimated that early medication procedures represented 22% of such abortions.

Eleven percent of procedures in abortion clinics and 22% in other clinics were early medication abortions. Medication abortions accounted for almost half of abortions in the nonhospital facilities with the smallest caseloads. About 20% of these facilities offered only medication abortions (not shown).

Accessibility of abortion

The number of abortions and the abortion rate are, in part, dependent on the accessibility of services, and some women may be unable to obtain an abortion because of circumstances such as distance, gestational limits and cost.

•Location and distance. Metropolitan statistical areas are made up of adjacent counties and are helpful to consider in measuring access to services. For example, transportation is likely to be available between the counties within a metropolitan area, so the proportion of women in counties without providers (shown in Table 3) may overstate the difficulty of accessing services.

Although abortion services tend to be concentrated in cities, 69% of counties in metropolitan areas lack a provider (Table 6, page 13). By comparison, almost all nonmetropolitan counties (97%) have no abortion provider, and virtually all such counties have no provider of 400 or more abortions. Like providers, the U.S. population is concentrated in metropolitan areas, but 24% of metropolitan women and 92% of their nonmetropolitan counterparts lack a provider in their county.

Thirty-seven percent of the 362 metropolitan statistical areas specified in 2005 lacked a provider, compared with 31% of 276 in 2000. (The 2000 proportion is similar—32%—if the new definition of metropolitan statistical areas is used.) An additional 6% of metropolitan areas had providers that reported fewer than 50 abortions in 2005.

These circumstances suggest that some women travel long distances to obtain an abortion. Nonhospital providers estimate that 8% of their clients travel more than 100 miles to access abortion services, 19% travel 50–100 miles and 73% travel less than 50 miles. These figures are comparable to those for 2001.6 Providers in the Northeast report women traveling the shortest distances; only 3% of clients were estimated to travel more than 100 miles, and 86% less than 50 miles. Women in the South and the Midwest have to travel the farthest: Providers estimate that 10% and 9% of clients, respectively, travel more than 100 miles to access services. Finally, providers in the West estimate that 5% of clients travel 100 or more miles to obtain services, and 18% travel 50–100 miles.

•Gestational limits. Most providers have lower and upper gestational limits for abortion services, and some women may have difficulty finding a provider if they seek an abortion too early or too late in pregnancy. For example, some providers will not perform an abortion if they cannot see the gestational sac on an ultrasound scan, which usually is not possible until 4–5 weeks after a woman’s last menstrual period.

Forty percent of providers in 2005 offered abortions at four or fewer weeks since the woman’s last menstrual period, about the same proportion as in 2001 (37%).⁶ The proportion of providers offering services increases with gestation and peaks—at 96%—at eight weeks. Sixty-seven percent of facilities offered at least some second-trimester abortion services (13 weeks or later). Twenty percent of providers offered abortions after 20 weeks, and only 8% at 24 weeks; comparable figures for 2001 were 24% and 13%, respectively.⁶

•Charges. Nonhospital providers were asked to report the usual charges a woman would incur for an abortion (with local anesthesia) at 10 and 20 weeks, including fees for required services (e.g., laboratory tests, additional visits); we did not distinguish between surgical and medication procedures.^§§ Since the majority of abortions are performed during the first trimester,4 and a majority of clinics charge a standard fee for any first-trimester abortion, the cost at 10 weeks represents the charge incurred by most women having an abortion. The mean charge for an abortion at 10 weeks’ gestation was $523, but charges ranged widely among providers (Table 7). The median charge, which is less influenced by the unusually high charges of a few providers, was $430. On average, abortion clinics reported the lowest median charge ($400), and private physicians’ offices reported the highest ($550). The cost of procedures varied by type of facility as well as the provider’s caseload. At 10 weeks, the larger the caseload, the less charged for the procedure. Facilities performing fewer than 30 abortions charged substantially more than those performing 5,000 or more.

Abortion procedures at later gestations are more complex, require an increased level of provider skill and take longer to perform. (For example, some procedures late in the second trimester occur over two or three days.) Later abortions are therefore more costly. Both the median ($1,260) and the mean ($1,339) charges for abortions performed at 20 weeks are roughly three times those for abortions performed at 10 weeks. Private physicians’ offices reported the lowest median cost ($1,000), and abortion clinics the highest ($1,350).

The amount that clinics charge for an abortion is not the same as the amount that women pay. When we weighted the charges on the basis of the number of abortions (rather than the number of providers), we found that the average woman obtaining an abortion at 10 weeks’ gestation paid $413 in 2006. The 2001 mean, adjusted for inflation (using the Consumer Price Index), was $424 in 2006 dollars.¹⁴ Thus, the amount that the average woman paid for a first-trimester abortion declined by $11 between 2001 and 2006.

Discussion

The long-term decline in abortion incidence continued through 2005. Between 2000 (the most recent year with comprehensive national data) and 2005, the number of abortions declined by 8%, and the abortion rate by 9%. Data are not yet available to determine the reasons for the continuing decline, but they likely include a range of circumstances, such as better contraceptive use, lower levels of unintended pregnancy, more women carrying unintended pregnancies to term and greater difficulties accessing abortion services in some geographic areas.

The number of abortion providers declined between 2000 and 2005, although at a slower rate than it had in previous periods. No clear pattern emerges between changes in numbers of providers and in abortion rates. Future research should seek to better understand state-level changes in provider numbers, provider types and abortion rates, particularly within the context of state restrictions on abortion services.

Our survey uncovered a shift in provider types. The number of clinics specializing in abortion services declined by 15%, and these facilities accounted for a smaller proportion of providers in 2005 than they did in 2000. The number of other clinics that offer abortions increased by 13% and accounted for a larger share of both providers and abortions than they had when last surveyed. This shift was due to the closing of abortion clinics, the introduction of abortion services into clinics that previously did not offer these services and the reclassification of abortion clinics as nonspecialized clinics. Specialized abortion clinics, which typically provide several hundred to several thousand abortions per year, are often more accessible to women with unwanted pregnancies because they advertise their services and usually charge lower fees than do other provider types. Many nonspecialized clinics also advertise abortion services and provide large numbers of abortions, for fees that are not substantially higher than those charged by specialized clinics. Future research will need to examine if the shift in provider type has impacted access to abortion services.

Early medication abortion, particularly procedures using mifepristone, has become more integrated into abortion services. In 2005, early medication abortions accounted for 14% of nonhospital abortions (compared with 6% in early 20013), and more than half of abortion providers offered early medication abortion. While abortion clinics were the most likely to offer this service, other clinics were the most likely to have introduced it. Early medication abortion requires less training and equipment than surgical abortion and can be more easily provided by family planning clinics and physicians’ offices. At least 10% of nonhospital abortion providers offered only early medication abortion services, and these facilities were most likely to be physicians’ offices and nonspecialized clinics. Indeed, if not for new providers offering only early medication abortion, the total number of providers would have decreased by 8% instead of 2% between 2000 and 2005. At the same time, our data suggest that most facilities offering only early medication abortion are located in areas where surgical abortion is provided, so it is unclear if these services have substantially increased access to abortion services. The availability of mifepristone starting in late 2000 does not appear to have led to an increased number of abortions or a higher abortion rate.

Other measures of accessibility to abortion services show little change. Some 87% of counties still lack a provider, and the proportion of women who have to travel more than 100 miles to access abortion services did not change. The proportion of providers offering very early abortions (those performed within four weeks after a woman’s last menstrual period) remained stable, and the proportion offering late abortions may have declined somewhat.

Limitations

While these data constitute the most complete information about abortion services in the United States, they are not without shortcomings. Some abortion providers were not included in our survey, either because we did not know about them or because they declined to respond. We expect that most of these providers perform fewer than 30 abortions a year, and failure to include them has had minimal impact on the overall number of abortions. At the same time, one important finding is that a nonnegligible minority of providers offer only early medication abortions and provide only a small number per year. Mifepristone has made it easier for health care providers, including those that do not specialize in obstetrics and gynecology, to provide abortion services. Even though we were able to send questionnaires to providers believed to have purchased mifepristone, we expect that some did not respond to the survey. Thus, we probably have underestimated the number of early medication abortions, as well as the total number of providers.

Other shortcomings include item nonresponse and inaccurate data reporting. When conducting nonresponse follow-up, we were sometimes able to obtain data only on numbers of abortions in 2004 and 2005; information about cost of services, early medication abortion and gestational limits was not always provided. Our analytic strategy assumes that nonhospital facilities that did not respond to specific items resembled those that did, but if this condition is not true, information about these aspects of abortion services may be inaccurate. Finally, providers have different ways of recording and retrieving information. Clinics that lack electronic records or do not monitor abortion services may have reported estimates rather than actual incidence, and this, too, increases the inaccuracy of our data.

Conclusions

Slightly more than one in five pregnancies end in abortion, indicating that unwanted pregnancy is still too common in the United States. More needs to be done to help women and their partners prevent unintended pregnancy. For example, more women and couples need access to resources and services that will help them to better plan when they want to have children and how to use contraceptive methods effectively until that time. In addition, it is important to remove barriers to abortion services—particularly for lower income women, who have above-average rates of unintended pregnancy.^2,15 Continued integration of mifepristone into settings where abortion is currently not available should increase access to abortion services, but this, alone, is not enough. Policy changes are also needed. Medicaid coverage of abortion services would better allow lower income women to access services. The elimination of Targeted Regulation of Abortion Providers, or TRAP, laws might prevent facilities from discontinuing abortion services, and might encourage potential providers to offer them. Until these structural barriers are reduced or eliminated, access to abortion will remain restricted.