Induced abortion, one of the most frequently performed surgical procedures in the United States, is experienced by a substantial proportion of American women. More than one-fifth of all pregnancies end in abortion,¹ a reflection of the fact that almost half of U.S. pregnancies are unintended.² Trends in abortion may reflect a number of factors, such as variations in the underlying incidence of unintended pregnancy and changes in how women resolve unplanned pregnancies and in the availability or accessibility of abortion services. Therefore, regular and accurate estimates of abortion incidence and service provision are essential for monitoring trends in reproductive behavior.

After remaining fairly steady for most of the 1980s, the number of abortions in the United States declined from a high of 1.61 million in 1990 to 1.36 million in 1996, the last year for which comprehensive abortion incidence data were collected. The abortion rate declined from 29.3 per 1,000 women aged 15-44 in 1980 and 27.4 in 1990 to 22.4 in 1996. The abortion ratio (the proportion of pregnancies ending in abortion) also fell during the early and mid-1990s. These declines meant that in the mid-1990s, measures of abortion reached the lowest levels since the 1970s.³

Several major developments since 1996 may have had an impact on unintended pregnancy levels and, therefore, abortion levels. Declines in teenagers' level of sexual activity⁴ and continued increases in their use and effective use of contraceptives⁵ could have reduced adolescent pregnancy and abortion rates and, thus, the overall abortion rate (although only one-fifth of abortions are provided to women younger than 20).⁶ Some states have expanded eligibility for family planning services under Medicaid;⁷however, Title X funds for free and low-cost family planning services have increased only enough to match inflation.⁸ In addition, the number of women of reproductive age covered by Medicaid declined in the late 1990s, while the number with no health insurance increased;⁹ these factors could have inhibited women's access to both family planning and abortion services.

Meanwhile, a continuing decline in the number of providers could result in more limited access to abortion services. The number of U.S. abortion providers fell from a high of 2,900 in 1982 to about 2,000 in 1996, and the proportion of counties without a provider increased from 77% in 1978 to 86% in 1996.¹⁰ A 1997 survey of obstetricians and gynecologists who perform abortions indicated that 57% were aged 50 or older,¹¹ fueling the perception that the number of providers will decline drastically as current providers reach retirement age. However, some evidence indicates that training opportunities for providers have begun to increase.¹²

A development that may have facilitated access to abortion was the introduction of a new method of early medical abortion. Mifepristone (formerly known as RU 486) was approved by the Food and Drug Administration (FDA) in September 2000, giving women seeking early abortion a nonsurgical option. Some providers have also used the cancer drug methotrexate to provide early medical abortion, but mifepristone is the first drug approved specifically for that purpose. Information about the extent of mifepristone utilization has only recently become available and still is quite limited.¹³

Between 1997 and 1998, the Centers for Disease Control and Prevention (CDC) reported a 2% decrease in the number of abortions performed in the United States and no change in the abortion rate. However, CDC data are compiled from state reports, and in 1998, four states did not report data to the CDC. These states (primarily California) accounted for 18% of all abortions tallied by The Alan Guttmacher Institute's (AGI's) 1997 data collection effort.¹⁴More recent CDC statistics on national abortion incidence are not available.

Thus, new data on nationwide abortion incidence and the number, types and locations of abortion service providers are needed. To obtain this information, AGI fielded a national survey of U.S. abortion providers (its 13th) in 2001 and 2002, collecting data primarily for 1999 and 2000. In this article, we present information from this survey on the number of abortions performed and national, regional and state abortion rates. We also examine the number and distribution of providers by location, type and caseload. Furthermore, we include data from previous AGI surveys to permit examination of trends over time. To obtain baseline estimates of mifepristone use in the United States, we report on medical abortions occurring during the first half of 2001. Finally, we report on findings regarding the incidence of abortion by dilation and extraction, a procedure that is the primary target of many efforts to ban so-called partial-birth abortions.*

METHODS

Questionnaire Development

Our survey questionnaire was modeled on the one used in AGI's previous round of data collection, in 1997. We created versions of the questionnaire for each of three major categories of providers: clinics, physicians and hospitals. The clinic and physician questionnaires were virtually identical. All questionnaires asked the number of induced abortions performed at the provider's location in 1999 and 2000. In addition, we asked hospitals the number of inpatient and outpatient procedures performed. We requested information from all providers on minimum and maximum gestations at which both surgical and medical abortions are performed, and we asked nonhospital^† providers about fees charged, sources of payment, distance traveled by clients and antiabortion harassment; results from these questions are presented elsewhere.¹⁵

In regard to early medical abortions and intact dilation and extraction abortions, we asked nonhospital providers the number of procedures performed in 2000 and during the first six months of 2001. We also asked whether they anticipated providing early medical abortions within the next year (if they were not already doing so). For nonhospital providers offering early medical abortion, we ascertained whether they used mifepristone or methotrexate.

Identifying Providers

Before fielding the survey, we conducted an extensive update of our list of facilities in the United States (excluding Puerto Rico and U.S. territories) where abortions are performed. We began with all of the providers^‡ known to have performed abortions in 1996, excluding those that stopped providing abortions or closed before January 1, 1999. To this list, we added possible new providers obtained from a variety of sources, including telephone yellow pages for the entire country, Planned Parenthood affiliates, the membership directory of the National Abortion Federation and World Wide Web listings of abortion providers. The updated list contained 2,287 possible providers.

In addition, the clinic and physician questionnaires inquired whether providers knew of facilities not offering surgical abortion that had begun offering medical abortion. Another question asked about hospital satellite facilities that performed abortions. During follow-up of these questions and of survey nonrespondents, as well as the investigation of mail returns, 155 additional possible providers were identified and included in the survey universe, bringing the total to 2,442. Seven of the additional providers were identified through the question about providers who performed only medical abortion; however, all seven reported performing surgical abortions as well.

Survey Fielding

In July 2001, we mailed questionnaires to all potential providers. Those who did not respond were sent two additional mailings at three-week intervals; a fourth mailing was sent to doctors' offices and hospitals. In addition, we contacted state health statistics agencies, requesting all available data reported by providers to each state health agency on the number of abortions performed in 1999 and 2000. For the states that supplied us with data by provider, we used the health agency figures for providers who did not respond to any of our mailings.

The remaining nonrespondents were contacted by telephone and asked to complete and return the questionnaire. Providers who could not or would not do so were asked a small number of key questions, including the number of abortions they had performed and gestational limits. To obtain facility data or record a final refusal, staff members made up to 35 attempts to contact each provider by phone, mail or fax. In total, staff members made more than 6,000 attempts to reach more than 900 providers. Follow-up continued through June 2002.

Of the 2,442 facilities surveyed, 962 responded to the mailed questionnaire, and 662 faxed or mailed a response or provided information during telephone follow-up; health department data were used for 449. (Each of these three groups of respondents included both providers who reported having performed abortions during the survey period and those who did not.) After additional follow-up with other sources, we determined that 32 more providers had closed or performed no abortions during the survey period, and that 14 were providers for whom we had already obtained data. For 71 of the remaining 323 potential providers, we obtained estimates of the number of abortions performed in 1999 and 2000 from knowledgeable sources in their communities, and for an additional 183 facilities that we knew had provided abortions, we made our own estimates. For three-fifths of these 183 estimates, we projected the number of abortions using data from previous surveys; such projections were almost always based on past information from the facilities themselves (and not on previous estimates).

We did not attribute any abortions to the remaining 69 facilities, for which no data or estimates were available; therefore, we did not count them as providers in 1999 or 2000. However, we cannot be sure that no abortions were provided at these facilities, although we were unable to obtain any indication that they were. For 15 of these providers, data were available for 1996; these providers performed a total of 1,594 abortions in that year.

Of the abortions reported for 2000, 77% were reported by the providers, 10% came from health department data, 11% were estimated by knowledgeable sources and 2% were projections or other estimates. These figures were similar to 1996 results. Out of 2,442 potential providers, a total of 1,931 performed abortions at some time between January 1999 and June 2001. Of those that did not, 245 indicated that they were not abortion providers, 82 had stopped providing abortions before the survey period began or had begun providing after the survey period ended, 76 had closed completely and 39 were duplicates; as indicated above, we were unable to ascertain whether 69 provided any abortions in the study period.

Some providers were undoubtedly missed because we were unable to identify them; the number can be estimated by surveying a random sample of physicians or hospitals not on our list of possible providers. Results from past underreporting surveys of this kind suggest that the actual number of abortions in 2000 might have been 3-4% greater than the number we counted and that we may have missed as many as half of the providers of fewer than 30 abortions.¹⁶ (We did not adjust the number of abortions or providers for this estimated undercount.) The number of abortions missed could be greater if our list omitted facilities with large abortion caseloads, but such omissions are unlikely, since large providers usually advertise and are known by referral sources. It is unlikely that we missed providers who were offering only medical abortion, because mifepristone became available only in November 2000, and because distributor reports suggest that the bulk of mifepristone shipments have been to existing providers.¹⁷

The state, regional and national data reported here are based on the location at which abortions occurred. In some cases, data based on women's place of residence may be quite different. For example, according to the most recent CDC data, 36% of abortions performed in Delaware in 1998 and 64% of abortions performed in the District of Columbia were obtained by nonresidents.^§18

RESULTS

Abortion Incidence

The number of abortions in the United States declined 3% between 1996 and 2000, from 1.36 million to 1.31 million (Table 1). This was the lowest number of abortions since 1976. The abortion rate also declined through 2000, reaching 21.3 abortions per 1,000 women 15-44 in that year. This figure represents a 5% drop over the four-year interval and is the lowest rate since 1974. The abortion ratio declined to 24.5 abortions per 100 pregnancies ending in abortion or live birth in 2000; this also represents a 5% drop since 1996 and the lowest figure since 1974. Including estimated miscarriages, 21% of all pregnancies in 2000 ended in abortion (not shown).^**

The number of abortions and abortion rates vary widely by region and state of occurrence (Table 2). Six states that account for 40% of women aged 15-44—California, Florida, Illinois, New Jersey, New York and Texas—accounted for 55% of all abortions in 2000. Rates were highest in New Jersey and New York, and were relatively high (above 30 per 1,000 women 15-44) in California, Delaware, Florida and Nevada. The states with the fewest abortions—South Dakota, North Dakota and Wyoming—are largely rural states and have relatively small populations. The lowest rates were in Kentucky, South Dakota and Wyoming; Idaho, Mississippi, Missouri, Utah and West Virginia also had low rates (seven or fewer per 1,000 women 15-44). Among the 25 states with the largest populations of women 15-44, the lowest abortion rate was in Kentucky.

Between 1996 and 2000, the abortion rate declined in every region of the country, but changes varied by region and, even more so, by state. The abortion rate declined in 35 states and the District of Columbia; the greatest percentage decreases occurred in Kentucky and Wyoming. Percentage changes are most meaningful in states with the greatest number of abortions, since small absolute changes in states with few abortions can result in large percentage shifts. Among the states reporting at least 10,000 abortions in 1996, the largest declines occurred in Massachusetts and Missouri. The abortion rate increased in 15 states. The largest percentage increase occurred in Delaware, and the largest increase among states with at least 10,000 abortions in 1996 occurred in Kansas (15%).

There was no clear state or regional pattern in time trends in abortion rates. Some states with rate increases between 1992 and 1996 had declines in the later period, and vice versa. The correlation between changes in state abortion rates in these two periods was low (r=-0.10).

Trends in Provider Numbers

A total of 1,819 providers performed at least one abortion in 2000—11% fewer than in 1996 (Table 3). In comparison, the number of providers declined by 14% from 1992 to 1996. The number of providers in 2000 was 37% lower than the all-time high of 2,908 in 1982 (not shown).

Between 1996 and 2000, the number of providers grew in nine states and fell in 38 and the District of Columbia; in the remaining three states, the number of providers did not change (Table 3). California and New York—the states with the largest numbers of providers—saw the largest absolute decreases between 1996 and 2000. The biggest absolute increases were in Connecticut, Hawaii and Pennsylvania. The increase in Pennsylvania may have resulted from the use of state health department data to identify hospitals that performed small numbers of abortions. In Connecticut, most of the increase was due to the identification of several physicians who performed a small number of abortions in 2000; some of these may have performed abortions in 1996, although we did not record them at that time. In Hawaii, most of the new providers were physicians as well.

Provider changes may be reflected in state abortion occurrence rates, especially if the number of abortions in a state is relatively small, but the impact may be exaggerated or muted by the size of the state (area and population), by the size of providers that discontinue or initiate services and by other factors. For example, a small net increase of two providers in Delaware is probably reflected in the parallel increase in the number of abortions that occurred in that state. However, it is hard to tell whether the latter change was a real increase or a shift in where Delaware residents had abortions because the change was small compared with changes in abortion levels in nearby areas. And while the number of providers decreased in both Kansas and Missouri from 1996 to 2000, a shift in service provision within the Kansas City area from Missouri to Kansas contributed to a sizable decrease in the abortion rate in Missouri and an increased rate in Kansas, although the same population was probably being served. The state-level correlation between the percentage change in the abortion rate between 1996 and 2000 and the percentage change in the number of providers during the same period was only -0.02 (not shown). In addition, the state-level percentage change in provider counts between 1996 and 2000 was not highly correlated (r=-0.08) with the change between 1992 and 1996.

Geographic Distribution of Providers

Abortion providers were located in 404 of the 3,141 U.S. counties in 2000. Overall, 87% of counties had no provider of abortions (Table 3). More than 90% of counties in the Midwest and South had no abortion provider; outside of these regions, the only states with no provider in at least 90% of counties were Idaho, Montana, Utah and Wyoming.

Although the vast majority of counties had no provider, only 34% of women aged 15-44 in 2000 lived in counties with no abortion providers, because many of these have relatively small populations. However, nearly half of women in the Midwest (49%) and South (45%) lived in counties that lacked abortion services. In 19 states in these regions, at least half of women lived in counties without an abortion provider. However, in six states in the same regions—Delaware, Florida, Illinois, Maryland, Michigan and Texas (and the District of Columbia)—fewer than one-third lived in counties with no provider. Fewer than one in five women in the Northeast and West lived in counties without an abortion provider; the proportion was less than one-third in 13 states in these regions and more than one-half in only three.

These measures may overestimate or underestimate the availability of services. On the one hand, many counties with no provider are adjacent to others where services may be available. On the other hand, facilities that perform few abortions may not be well known to the general public, so the existence of a small provider in a county does not guarantee the availability of services. Thus, additional useful indicators of service availability are the presence or absence of providers in an entire metropolitan area†&134; and the proportion of counties without a provider large enough to be likely to advertise its services and accept self-referred patients.¹⁹ (For the purposes of this analysis, we use 400 or more abortions provided per year as the criterion for this category of provider.)

Between 1990 and 1999, the number of counties defined as metropolitan grew as some cities and urbanized areas became large enough to qualify as metropolitan areas.^‡‡ In Table 4, where we present analyses by metropolitan status, we show figures for 2000 using two definitions of metropolitan status. This allows us to give accurate figures for 2000 (based on the 1999 metropolitan classification), while also showing true trends since 1978 (based on the 1990 classification).

The proportion of counties with no abortion provider in 2000 (87%) changed little compared with that in 1996 (86%), but remained higher than the proportion in 1978 (77%). In addition, the proportion of counties with no provider of 400 or more abortions per year has changed little over time, indicating that the drop in counties with providers has been concentrated in those where providers perform fewer than 400 abortions per year.

Most abortion providers are located in metropolitan areas: 94% of all providers and 99% of those who performed 400 or more abortions in 2000 (not shown). Even so, 61% of counties in metropolitan areas had no abortion provider, and 70% had no large provider (Table 4). Of nonmetropolitan counties, 97% had no provider, and virtually all lacked a provider of at least 400 abortions per year.

Overall, the proportion of women living in a county without a provider increased from 27% in 1978 to 30% in 1985 and 34% in 2000. However, figures based on comparable metropolitan classifications indicate that the proportion of women with no provider in their county increased from 1978 to 1996 in both metropolitan and nonmetropolitan counties, but changed only slightly between 1996 and 2000. There was no change during the 1990s in the proportion of women in metropolitan areas living in counties with no large provider, although the levels were slightly greater than those in 1978 and 1985. Almost all women in nonmetropolitan counties have lived without a large abortion provider.

The 856 metropolitan counties make up 276 metropolitan areas (on the basis of the 1999 metropolitan classification). Eighty-six of these areas (31%) have no abortion provider, and an additional 12 reported fewer than 50 abortions in 2000 (not shown). If women in these areas sought abortions at the same rate as the overall U.S. population, as many as 250-2,640 women in each metropolitan area would seek abortion services. The two largest areas without a provider are within 100 miles of each other in eastern Pennsylvania (Scranton-Wilkes-Barre-Hazleton and Lancaster), and have 124,000 and 99,000 women aged 15-44, respectively. Three other areas have populations of 80,000 or more women 15-44: Provo-Orem, Utah; Lafayette, Louisiana; and Canton-Massillon, Ohio. However, all regions of the country are represented on the list of metropolitan areas with no provider of 50 or more abortions.^§§ Ten of these metropolitan areas are located in or include Texas, seven are in Pennsylvania and six each are in Alabama, Indiana and Ohio. In some cases, active community opposition has made it difficult to establish abortion facilities in unserved cities. For example, when a provider in Lancaster, Pennsylvania, made plans to offer abortions in September 1998, antiabortion advocates initiated legislation that led the local zoning board to reverse its initial stance and deny the clinic a permit for surgical procedures.²⁰

Types of Providers

• Clinics. In 2000, clinics made up 46% of all abortion providers (Table 5); this proportion was up from 43% in 1996 (figures cited here and below for 1996 are not shown in the table). Most abortions in 2000 were performed at clinics (93%); this figure increased from 90% in 1996. (Physicians' offices where more than 400 abortions were provided have been categorized as clinics.)

Slightly more than half of clinics (25% of all providers) were specialized abortion clinics, defined as those where at least half of patient visits are for abortion services. Such clinics provided 71% of abortions in 2000, about the same proportion as in 1996 (70%). Caseloads are largest at abortion clinics: Three-fourths provided at least 1,000 abortions in 2000, while only 7% of other providers did so. The remaining clinics, in which the majority of patients receive services other than abortion, made up 21% of providers and performed 22% of all abortions in 2000.

• Hospitals. One-third of abortion providers in 2000 were hospitals, nearly the same proportion as in 1996. However, the proportion of abortions performed in hospitals decreased from 7% to 5% during the four years. More than half of hospitals performing abortions (18% of all providers) performed fewer than 30; 24% performed five or fewer abortions (not shown—these hospitals were most likely performing abortions only in cases of fetal anomaly or serious risk to the woman's life or health).

Four-fifths (82%) of hospitals that provided abortions were private; 69% were nonprofit and 13% were for-profit. The remaining hospital providers were under the jurisdiction of either a state (5%), a county (5%), a city (3%) or a hospital district, a public entity created by a state and covering a specific community (6%). Eighty-eight percent of hospital abortions were outpatient procedures, nearly the same proportion as in 1996 (91%) and 1992 (89%). In 2000, some 8,000 abortions involved hospitalization (not shown).

• Physicians. One-fifth (21%) of providers were physicians' offices (defined here as providers that appear from their name to be physicians' offices and reported performing fewer than 400 abortions in 2000), representing a decline from 23% in 1996. Forty-one percent of these practices (9% of all providers) performed fewer than 30 abortions in 2000. In total, these offices performed 27,500 abortions, and their share of abortions fell from 3% in 1996 to 2% in 2000.

Provider Caseloads and Types of Procedures

A majority (62%) of abortion providers performed fewer than 400 abortions in 2000. However, most abortions were obtained at large facilities where 1,000 or more abortions were performed (80%), nearly the same proportion as in 1996 (79%). Large providers were predominantly abortion clinics; 65% of abortions in 2000 were performed in abortion clinics that had caseloads of 1,000 or more procedures per year. Between 1996 and 2000, the number of providers declined in each size category except the largest (5,000 or more); thus, abortions were increasingly concentrated among a small number of very large providers.

• Early medical abortion. Mifepristone received FDA approval in September 2000, and distribution of the drug to providers began in November 2000.²¹ Thus, the first six months of 2001 represent the initial period in which the method was available to American women outside of clinical trials. During that period, one-third of all abortion providers in the 2000 provider universe performed at least one early medical abortion—that is, an abortion in the first trimester using mifepristone or methotrexate (Table 6); medical abortions with mifepristone and methotrexate are always prescribed with misoprostol.

About half of abortion clinics (51%) and nonspecialized clinics (45%) provided early medical abortion, as did one in five (19%) hospital abortion providers. Large providers were the most likely to offer early medical abortion during this initial time period: At least 60% of those performing 1,000 or more abortions per year offered medical abortion, compared with at most 23% of providers performing fewer than 400 abortions. All of the providers offering early medical abortions during our survey period also performed surgical abortions.

Nonhospital facilities made up 81% of sites where early medical abortions were provided in the first half of 2001 (not shown). These sites provided an estimated 35,000 early medical abortions in that time period; 72% of these were performed with mifepristone, and the rest with methotrexate (Table 6). Roughly three-quarters of medical abortions were provided at abortion clinics.

Early medical abortions represented an estimated 6% of abortions performed in nonhospital facilities during the first half of 2001. Providers with annual caseloads of fewer than 30 abortions reported a higher proportion of medical abortions than those with larger caseloads (although larger providers reported a greater number of early medical abortions). We did not ask hospitals the number of medical abortions they provided, but if 6% of all abortions at hospitals were early medical abortions, an estimated 2,000 additional early medical abortions were performed, for a total of 37,000 early medical abortions in the first half of 2001.

Of providers performing medical abortions, 54% used only mifepristone, and 18% used only methotrexate (not shown). A smaller proportion of physician offices (54%) than of clinics (75-77%) performed medical abortions with mifepristone; larger providers were more likely than smaller ones to perform medical abortions with mifepristone.

Among nonhospital facilities that did not offer medical abortion in the first half of 2001, 30% reported that they "probably will" offer it in the future, 23% said "maybe" and 47% said they "probably won't" (not shown). Providers with larger caseloads were more likely than those with smaller caseloads to report that they would offer the method, as were providers in the Northeast and Midwest.

• Dilation and extraction abortions. Abortions performed by dilation and extraction*^;† are quite rare: Eighteen providers reported 1,274 such abortions in 2000, and 16 providers reported 742 for the first half of 2001; an additional provider reported performing dilation and extraction abortions in both 2000 and 2001, but could not say how many. Assuming that the provision of dilation and extraction abortions by providers who responded to the question reflects the experience of nonrespondents of similar type and size, an estimated total of 31 providers performed the procedure 2,200 times in 2000, and 0.17% of all abortions performed in that year used this method. While these data confirm that the absolute number of abortions performed by dilation and extraction is very small, this figure should be interpreted cautiously, because projections based on such small numbers are subject to error.²²

DISCUSSION

Between 1996 and 2000, the U.S. abortion rate fell 5%, a decline less than half as steep as that seen between 1992 and 1996 (12%). The number of abortion providers continued to decline between 1996 and 2000, at a rate slightly lower than that during 1992-1996. The 1996-2000 period saw the continuing consolidation of abortion provision at clinics, particularly specialized clinics; only 7% of abortions in 2000 were performed in nonclinic facilities. This trend may be partially due to increasing legal constraints on the circumstances under which abortions may be performed, such as zoning rules and state licensing and inspection requirements. Specialized clinics may be better able to deal with new restrictions than physicians' offices and nonspecialized clinics, which may not be willing or able to undertake the expenses and time required to comply with them. This factor may be most relevant in South Carolina and Mississippi, where new licensing laws have created burdensome requirements for small providers; at least one South Carolina provider has reportedly closed in response to the new regulations.²³

Another factor that may have contributed to the decline in the number of providers since 1996 is harassment. Despite the reported decline in severe forms of harassment of abortion providers,²⁴ several high-profile incidents of violence have occurred since 1996. In addition to the murder of Buffalo abortion provider Barnett Slepian and the death of a police officer in a Birmingham, Alabama, clinic bombing in 1998, two doctors were shot and wounded in 1997.²⁵ These incidents may have increased providers' fear of physical threats and, thus, contributed to the drop in the number of providers.

The decrease in providers was concentrated among those with small caseloads. Because many hospitals and physicians who did not perform abortions in 2000 performed few abortions in 1996, this decline probably had little impact on abortion incidence nationally, although it may have had a significant impact on abortion accessibility for residents of some rural areas and small towns.

For most American women, access to abortion is directly tied to where they live. Only 3% of nonmetropolitan counties have a provider, and almost none of those providers performed more than 400 abortions in 2000. Of metropolitan counties, only 30% have a large abortion provider. Surprisingly, although the proportion of nonmetropolitan counties with a provider has declined, the proportion of women in nonmetropolitan counties with a provider appears to have increased slightly, probably because of population shifts toward counties with providers. In metropolitan areas, the proportion of women living in counties with providers has changed little.

The Northeast and West are characterized by higher abortion rates and greater access to providers than are the Midwest and South, and also by more supportive laws regarding abortion.²⁶ In some states, abortion decreases may be due to regulatory requirements placed on women seeking abortion. For example, in Wisconsin, the imposition of a two-day delay law may have contributed to the 21% decline in the abortion rate (although women there may increasingly have gone to Illinois, particularly Chicago, to obtain abortions). In other states, rates may decline because many women travel out of state to have abortions.²⁷ This may occur when the barriers to obtaining an abortion—such as gestational limits or other restrictions, or expense—are lower in neighboring states.

During the first six months of 2001, early medical abortion (largely mifepristone) accounted for a small but nonnegligible proportion of all abortions. As of April 2002, 69% of National Abortion Federation members offered the method.²⁸ The growing acceptance of mifepristone raises the possibility that the decrease in surgical abortion providers may be offset by an increase in the number of providers that offer medical abortion, particularly in areas with no current providers. However, the information available from this early phase of provision suggests that the availability of this new procedure has not reduced travel distances for abortions²⁹ or increased the overall abortion rate. In addition, our findings show that mifepristone is being used mostly by existing (surgical) abortion providers rather than by new providers.

In the past, the U.S. abortion rate has been distinctly higher than the rate in other industrialized countries. Although the U.S. rate (21.3 per 1,000 women 15-44) is still higher than those in many western European countries, it is now within the range of rates in a few other developed countries, such as Sweden (18.7) and Australia (22.2).³⁰ Furthermore, U.S. rates vary by women's ethnicity and socioeconomic standing; the rate among white non-Hispanic women is in the middle range of other developed countries, but other ethnic groups have higher rates. Moreover, poor and near-poor women have rates roughly twice as high as their wealthier counterparts.³¹

This article has documented current levels of abortion and abortion service provision. More research needs to be done both to understand why abortion service provision is changing and the impact on women of the small number and geographic concentration of providers. In addition, further work is needed to determine the causes of declines in the abortion rate. Increasing use of emergency contraception appears to have been a major contributor in recent years: An estimated 51,000 pregnancies were averted by emergency contraception in 2000, accounting for 43% of the decrease in abortions since 1994.³² Contraceptive use trends through 1995—improvements in use (e.g., a shift to greater use of long-acting, highly effective methods) and reductions in the proportion of women using no method—may have continued. The abortion rate decline between 1994 and 2000 was greatest among teenagers.³³ Both a decline in sexual activity among adolescents and increased use of contraceptives at first intercourse contribute to decreasing pregnancy and abortion rates among adolescents.³⁴

It is also important to understand better the societal and personal factors that can have an impact on sexual and contraceptive behavior and the ways women deal with unintended pregnancies, as well as the factors that affect women's ability to obtain abortions when they seek them. The impacts of various influences may also change over time. For example, one previous study found no consistent relationship between economic conditions (as measured by income, employment and government benefits) and abortion rates at the state level.³⁵ However, new data indicate that trends in abortion rates were similar among lower- and higher-income women between 1987 and 1994, but have diverged since then.³⁶ This may indicate that increased economic pressures are discouraging greater numbers of lower-income women from having children, or that it is more difficult for them to avoid unintended pregnancy because of decreased access to contraceptive services.

With more than one in five U.S. pregnancies ending in abortion, it is clear that American women are becoming pregnant far more often than they desire. More than half of these pregnancies occur among women who had difficulty using contraceptive methods effectively or who experienced method failure, and nearly half occur among the minority of sexually active women who use no contraceptives, reflecting the high rate of pregnancy among this group.³⁷ The challenge of reducing U.S. abortion rates without increasing unintended births requires action on several fronts, but foremost among these are increasing (and increasing the effectiveness of) contraceptive use by sexually active women and their partners, improving access to contraceptive services for those who are disadvantaged and ensuring the availability of a broader range of more-effective and user-friendly contraceptive methods.