Abortion Surveillance — United States, 2013

Introduction

This report summarizes abortion data for 2013 that were provided voluntarily to CDC by the central health agencies of 49 reporting areas (the District of Columbia; New York City; and 47 states, excluding California, Maryland, and New Hampshire). Data were obtained every year during 2004–2013 from 47 reporting areas (excluding California, Louisiana, Maryland, New Hampshire, and West Virginia) and were used for trend analyses. Since 1969, CDC has conducted abortion surveillance to document the number and characteristics of women obtaining legal induced abortions in the United States (1). Following nationwide legalization of abortion in 1973, the total number, rate (number of abortions per 1,000 women aged 15–44 years), and ratio (number of abortions per 1,000 live births) of reported abortions increased rapidly, reaching the highest levels in the 1980s before decreasing at a slow yet steady pace (2–4). However, the incidence of abortion has varied considerably across demographic subpopulations (5–10). Moreover, during 2006–2008, a break occurred in the previously sustained pattern of decrease (11–14), although this break has been followed in all subsequent years by even greater decreases (15–19). Continued surveillance is needed to monitor long-term changes in the incidence of abortion in the United States.

Methods

Description of the Surveillance System

Each year, CDC requests tabulated data from the central health agencies of 52 reporting areas (the 50 states, the District of Columbia, and New York City) to document the number and characteristics of women obtaining legal induced abortions in the United States. For the purpose of surveillance, a legal induced abortion* is defined as an intervention performed within the limits of state law by a licensed clinician (e.g., a physician, nurse-midwife, nurse practitioner, or physician assistant) that is intended to terminate a suspected or known intrauterine pregnancy that is ongoing and produce a nonviable fetus.

In most states, collection of abortion data is facilitated by the legal requirement for hospitals, facilities, and physicians to report all abortions to a central health agency (20). These central health agencies then voluntarily report the abortion data they have collected through their independent surveillance systems and provide only aggregate numbers to CDC (21). Although reporting to CDC is voluntary, most reporting areas provide their aggregate abortion numbers. During 2004–2013, a total of 47 reporting areas provided CDC a continuous annual record of abortion numbers^† and in 2013, CDC obtained aggregate abortion numbers from 49 reporting areas (excludes California, Maryland, and New Hampshire).

Although CDC obtains aggregate abortion numbers from most of the central health agencies, the level of detail that it receives on the characteristics of women obtaining abortions varies considerably from year to year and by reporting area. To encourage more uniform collection of these details, CDC has collaborated with the National Association of Public Health Statistics and Information Systems to develop reporting standards and provide technical guidance for vital statistics personnel who collect and summarize abortion data within the United States. However, because the collection and reporting of abortion data are not federally mandated, many reporting areas have developed their own data collection forms and therefore do not collect or provide all the information or level of detail included in this report.

Variables and Categorization of Data

Each year, CDC sends suggested templates to the central health agencies for compilation of abortion data in aggregate. Aggregate abortion numbers, without individual-level records, are requested for the following variables:

• Maternal age in years (<15, 15–19 by individual year, 20–24, 25–29, 30–34, 35–39, or ≥40)

• Gestational age in weeks at the time of abortion (≤6, 7–20 by individual week, or ≥21)

• Race (black, white, or other [including Asian, Pacific Islander, other races, and multiple races])

• Ethnicity (Hispanic or non-Hispanic)

• Method type (curettage,^§ intrauterine instillation, medical [nonsurgical] abortion, or hysterectomy/hysterotomy)

• Marital status (married [including currently married or separated] or unmarried [including never married, widowed, or divorced])

• Number of previous live births (0, 1, 2, 3, or ≥4)

• Number of previous abortions (0, 1, 2, or ≥3)

• Maternal residence (the state, reporting area, territory, or foreign country in which the woman obtaining the abortion lived; or, if additional details are unavailable, in-reporting area versus out-of-reporting area)

In addition to sending templates for compiling information on race and ethnicity as separate variables, since 2001 CDC has provided alternative templates for the tabulation of aggregate cross-classified race/ethnicity data. Before 2007, few reporting areas returned these alternative templates; results by these cross-classified race/ethnicity categories (non-Hispanic white, non-Hispanic black, non-Hispanic other, and Hispanic) are thus shown only for 2007–2013.

Finally, both the original and alternative templates provided by CDC request that aggregate numbers for certain variables be cross-tabulated by a second variable. These cross-tabulations include gestational age (separately by maternal age, by method type, by race, by ethnicity, and by race/ethnicity) and maternal age and marital status (separately by race, by ethnicity, and by race/ethnicity).

In this report, medical abortions and abortions performed by curettage are further categorized by gestational age. For medical abortion, early medical abortion is defined as the administration of medication or medications (typically mifepristone followed by misoprostol) to induce an abortion at ≤8 weeks’ gestation^¶; medical abortion at >8 weeks’ gestation is defined as the administration of medication or medications (typically vaginal prostaglandins) to induce an abortion at >8 weeks’ gestation. For curettage, abortions are categorized as having been performed at ≤13 weeks’ gestation or at >13 weeks’ gestation because of differences in technique used before and after 13 weeks (23). Finally, because intrauterine instillations cannot be performed early in gestation, abortions reported to have been performed by intrauterine instillation at ≤12 weeks’ gestation are excluded from calculation of the percentage of abortions by known method type.**

Measures of Abortion

Four measures of abortion are presented in this report: 1) the total number of abortions in a given population, 2) the percentage of abortions obtained by women in a given population, 3) the abortion rate (number of abortions per 1,000 women aged 15–44 years or other specific group within a given population), and 4) the abortion ratio (number of abortions per 1,000 live births within a given population). Although total numbers and percentages are useful for determining how many women have obtained an abortion, abortion rates adjust for differences in population size and reflect how likely abortion is among women in particular groups. Abortion ratios measure the relative number of pregnancies in a population that end in abortion compared with live birth. Abortion ratios are influenced both by the proportion of pregnancies in a population that are unintended and the proportion of unintended pregnancies that end in abortion. Abortion ratios also are influenced by the proportion of intended pregnancies that end in abortion; however, intended pregnancies account for a very small percentage of abortions (<5%) (26).

U.S. Census Bureau estimates of the resident female population of the United States, compiled by CDC, were used as the denominator for calculating abortion rates (27–36). Overall abortion rates were calculated from the population of women aged 15–44 years living in the reporting areas that provided data. For adolescents aged <15 years, abortion rates were determined on the basis of the number of adolescents aged 13–14 years; similarly, for women aged ≥40 years, abortion rates were determined on the basis of the number of women aged 40–44 years. For the calculation of abortion ratios, live birth data were obtained from CDC natality files (37) and included births to women of all ages living in the reporting areas that provided abortion data.

Data Presentation and Analysis

This report provides state-specific and overall abortion numbers, rates, and ratios for the 49 areas that reported to CDC for 2013 (excludes California, Maryland, and New Hampshire). In addition, this report describes the characteristics of women who obtained abortions in 2013. Because the completeness of reporting on the characteristics of women varies by year and by variable, this report only describes the characteristics of women obtaining abortions in areas that met reporting standards (i.e., reported at least 20 abortions overall, provided data categorized in accordance with surveillance variables, and had <15% unknown values for a given characteristic). Cells with a value in the range of 1–4 have been suppressed to maintain confidentiality. In addition, abortion rates and ratios have been omitted for groups with <20 abortions because results are considered unstable (38).

Although most of the data are presented by the reporting area in which the abortions were performed, 48 reporting areas in 2013 also provided the number of abortions by maternal residence.^†† However, two of these reporting areas (Illinois and Wisconsin) reported certain characteristics for in-state residents but not for out-of-state residents. Three other reporting areas (Iowa, Louisiana, and Massachusetts) provided only the total number of abortions for out-of-state residents without specifying individual states or areas of residence from which these women came. As a result, abortion statistics in this report by area of residence should be interpreted with caution and might underestimate abortions, especially for reporting areas from which many women travel to other states to obtain abortion services.

To evaluate overall trends in the number, rate, and ratio of reported abortions, annual data are presented for the 47 areas that reported every year during 2004–2013. Linear regression analysis was used to assess the overall rate of change among these areas during the entire 10-year period of analysis (2004–2013) and during the first and second halves of the period of analysis (2004–2008 and 2009–2013). The percentage change in abortion measures from the most recent past year (2012 to 2013) and from the beginning to the end of the 10-year period of analysis (2004 to 2013) also were calculated for these same 47 areas.

For the analysis of certain additional variables (i.e., abortions by maternal age and gestational age), annual data are presented for areas that met reporting standards every year during 2004–2013; the percentage change was calculated from the beginning to the end of the 10-year period of analysis (2004 to 2013), from the first and second halves of this period (2004 to 2008 and 2009 to 2013), and from the most recent past year (2012 to 2013). For other variables (i.e., race/ethnicity, method for performing an abortion, marital status, number of previous abortions, and number of previous live births), annual data are not presented; areas were included if they met reporting standards for the years needed for percentage change calculations. To evaluate trends in the use of different methods for performing an abortion, reporting areas were included only if they met reporting standards and if they specifically included medical abortion as a method on their reporting form.

Some of the 49 areas that reported for 2013 are not included in certain trend analyses. As a result, summary measures for comparisons over time might differ slightly from the point estimates presented for all areas that reported for 2013.

Abortion Mortality

CDC has reported data on abortion-related deaths periodically since information on abortion mortality first was included in the 1972 abortion surveillance report (19,39). An abortion-related death is defined as a death resulting from a direct complication of an abortion (legal or illegal), an indirect complication caused by a chain of events initiated by an abortion, or an aggravation of a pre-existing condition by the physiologic or psychologic effects of abortion (40). All deaths determined to be related causally to induced abortion are classified as abortion related regardless of the time between the abortion and death. In addition, any pregnancy-related death in which the pregnancy outcome was induced abortion regardless of the causal relation between the abortion and the death is considered an abortion-related death. An abortion is defined as legal only if it is performed by a licensed clinician within the limits of state law.

Since 1987, CDC has monitored abortion-related deaths through its Pregnancy Mortality Surveillance System (41,42). Sources of data for abortion-related deaths have included state vital records; media reports, including computerized searches of full-text newspaper and other print media databases; and individual case reports by public health agencies, including maternal mortality review committees, health care providers and provider organizations, private citizens, and citizen groups. For each death that possibly is related to abortion, CDC requests clinical records and autopsy reports. Two medical epidemiologists independently review these reports to determine the cause of death and whether the death was abortion related. Discrepancies are discussed and resolved by consensus. Each death is categorized by abortion type as legal induced, illegal induced, spontaneous, or unknown type.

This report provides data from the Pregnancy Mortality Surveillance System on induced abortion-related deaths that occurred in 2012, the most recent year for which data are available. Data on induced abortion-related deaths that occurred during 1972–2011 already have been published (19), and possible abortion-related deaths that occurred during 2013–2016 are under investigation. For 1998–2012, abortion surveillance data reported to CDC cannot be used alone to calculate national case-fatality rates (number of legal induced abortion-related deaths per 100,000 reported legal induced abortions in the United States) because certain states^§§ did not report abortion data every year during this period. Thus, national legal induced abortion case-fatality rates were calculated with denominator data from a more complete source on the total number of abortions performed in the United States (15). Because rates determined on the basis of a numerator of <20 deaths are highly variable (38), national legal induced abortion case-fatality rates were calculated for consecutive 5-year periods during 1973–2012.

Results

U.S. Totals

Among the 49 reporting areas that provided data for 2013, a total of 664,435 abortions were reported. Of these abortions, 652,582 (98.2%) were obtained from the 47 reporting areas that provided data every year during 2004–2013.^¶¶ These same 47 areas had an abortion rate of 12.5 abortions per 1,000 women aged 15–44 years and an abortion ratio of 200 abortions per 1,000 live births (Table 1). For all three measures of abortion, large decreases resulted in the lowest levels reported during the entire period of analysis. From 2012 to 2013, a 5% decrease occurred in the total number of reported abortions (from 688,149), the abortion rate (from 13.2 abortions per 1,000 women aged 15–44 years), and the abortion ratio (from 210 abortions per 1,000 live births). From 2004 to 2013, among the same 47 areas that reported every year during this period, the total number of reported abortions decreased 20% (from 817,906), the abortion rate decreased 21% (from 15.9 abortions per 1,000 women aged 15–44 years), and the abortion ratio decreased 17% (from 241 abortions per 1,000 live births) (Figure 1). Among these same 47 areas, the annual rate of decrease fitted from the regression analysis was greater during 2009–2013 than during 2004–2008 for all three measures of abortion. During 2009–2013, the number of reported abortions decreased by 32,002 abortions per year, the abortion rate decreased by 0.64 abortions per 1,000 women per year, and the abortion ratio decreased by 7.1 abortions per 1,000 live births per year. In contrast, during 2004–2008, the number of reported abortions increased by 896 abortions per year, the abortion rate decreased by 0.01 abortions per 1,000 women per year, and the abortion ratio decreased by 2.5 abortions per 1,000 live births per year.

Occurrence and Residence

Abortion numbers, rates, and ratios for 2013 have been calculated by reporting area of occurrence and the residence of the women who obtained the abortions (Table 2). By occurrence, a considerable range existed in the abortion rate (from 3.6 abortions per 1,000 women aged 15–44 years in Mississippi to 24.3 abortions per 1,000 women in New York [city and state combined]) and the abortion ratio (from 49 abortions per 1,000 live births in South Dakota to 414 abortions per 1,000 live births in New York [city and state combined]).*** Similarly, a considerable range existed by residence^††† in the abortion rate (from 4.7 abortions per 1,000 women aged 15–44 years in South Dakota to 23.6 abortions per 1,000 women aged 15–44 years in New York [city and state combined]) and the abortion ratio (from 60 abortions per 1,000 live births in Utah and South Dakota to 401 abortions per 1,000 live births in New York [city and state combined]). Because of variation that occurred among reporting areas in the percentage of abortions obtained by out-of-state residents (from 0.4% in Hawaii to 57.3% in the District of Columbia), abortion rates and ratios calculated by maternal residence might provide a more accurate reflection of the state-specific distribution of women obtaining abortions. However, because states vary in the level of detail they collect on maternal residence, 11.9% of abortions were reported to CDC without exact information on maternal residence.

Maternal Age

Among the 47 areas that reported by maternal age for 2013, women in their 20s accounted for the majority (58.6%) of abortions and had the highest abortion rates (21.8 and 18.2 abortions per 1,000 women aged 20–24 and 25–29 years, respectively) (Figure 2) (Table 3). Women in the youngest (<15 years) and oldest (≥40 years) age groups accounted for the smallest percentages of abortions (0.3% and 3.6%, respectively) and had the lowest abortion rates (0.6 and 2.5 abortions per 1,000 women aged <15 and ≥40 years, respectively). Among the 42 reporting areas that provided data by maternal age every year during 2004–2013, this pattern across age groups was stable, with the majority of abortions and the highest abortion rates occurring among women aged 20–29 years and the lowest percentages of abortions and abortion rates occurring among women in the youngest and oldest age groups (Table 4). During 2004–2013, abortion rates decreased among all age groups <40 years, although the decreases for adolescents (54% and 46% for adolescents aged <15 and 15–19 years, respectively) were greater than the decreases for women in older age groups (8%–27% for women aged 20–39 years). For all age groups, decreases in the abortion rate were greater from 2009 to 2013 than from 2004 to 2008, and these decreases continued from 2012 to 2013.

In contrast to the percentage of abortion numbers and abortion rates, abortion ratios in 2013 were highest among adolescents aged ≤19 years and lowest among women aged 30–39 years (Figure 2) (Table 3). Among the 42 reporting areas that provided data by maternal age for every year during 2004–2013, abortion ratios decreased among women in all age groups except adolescents aged <15. For all age groups, decreases in the abortion ratio were greater from 2009 to 2013 than from 2004 to 2008, and these decreases continued from 2012 to 2013 (Table 4).

Adolescents

Among the 46 areas that reported maternal age by individual year among adolescents for 2013, adolescents aged 18–19 years accounted for the majority (67.2%) of adolescent abortions and had the highest adolescent abortion rates (11.5 and 15.6 abortions per 1,000 adolescents aged 18 and 19 years, respectively); adolescents aged <15 years accounted for the smallest percentage (2.9%) of adolescent abortions and had the lowest adolescent abortion rate (0.6 abortions per 1,000 adolescents aged 13–14 years) (Table 5). Among the 40 reporting areas that provided data for adolescents by individual year of maternal age every year during 2004–2013, the percentage of abortions accounted for by adolescents aged 19 years increased, whereas the percentage of abortions accounted for by adolescents aged <19 years decreased. Among adolescents of all ages, abortion rates decreased, but the decreases were smaller for adolescents aged 19 years as compared with adolescents <19 years (Table 6). For all age groups, decreases in the abortion rate were greater from 2009 to 2013 than from 2004 to 2008, and these decreases continued from 2012 to 2013.

In 2013, the abortion ratio for adolescents decreased with increasing age and was lowest among adolescents aged 19 years (Table 5). Among the 40 reporting areas that provided data for adolescents by individual year of maternal age for every year during 2004–2013, abortion ratios decreased among adolescents of all ages except for those aged <15 years, for whom the abortion ratio increased slightly (2%) (Table 6).

Gestational Age

Among the 40 areas that reported gestational age at the time of abortion for 2013, the majority (66.0%) of abortions were performed by ≤8 weeks’ gestation and 91.6% were performed at ≤13 weeks’ gestation (Table 7). Few abortions were performed between 14 and 20 weeks’ gestation (7.1%) or at ≥21 weeks’ gestation (1.3%). Among the 30 reporting areas that provided data on gestational age every year during 2004–2013, the percentage of abortions performed at ≤13 weeks’ gestation was stable (Table 8). However, within this gestational age range, a shift occurred toward earlier gestational ages at time of abortion, with the percentage of abortions performed at ≤8 weeks’ gestation increasing 4% and the percentage of abortions performed at 9–13 weeks’ gestation decreasing 10%. For the entire period of analysis, abortions performed at >13 weeks’ gestation accounted for a small percentage of abortions (≤8.5%).

Among abortions performed at ≤13 weeks’ gestation and reported by individual week of gestation for 2013, 37.8% were performed at ≤6 weeks’ gestation (Table 9). Among the remaining abortions performed between 7 and 13 weeks’ gestation, the percentage contribution was progressively smaller for each additional week of gestation: 19.6% were performed at 7 weeks’ gestation and 3.1% were performed at 13 weeks’ gestation. Among the 30 areas that reported by exact week of gestation for abortions performed at ≤13 weeks’ gestation every year during 2004–2013, the percentage of abortions shifted toward earlier gestational ages. Abortions performed at ≤6 weeks’ gestation increased 16% and those performed at 7–12 weeks’ gestation decreased in the range of 3%–16% (Table 10).

Method Type

Among the 43 areas that reported by method type for 2013 and included medical abortion on their reporting form for medical providers, 67.9% of abortions were performed by curettage at ≤13 weeks’ gestation, 22.2% were performed by early medical abortion (a nonsurgical abortion at ≤8 weeks’ gestation), and 8.6% were performed by curettage at >13 weeks’ gestation; all other methods were uncommon (Table 11). Among the 33 reporting areas that included medical abortion on their reporting form and provided these data for the relevant years of comparison (2004 versus 2013, 2004 versus 2008, 2009 versus 2013, and 2012 versus 2013),^§§§ use of early medical abortion increased 5% from 2012 to 2013 (from 21.3% of abortions to 22.3%); from 2004 to 2013, use of early medical abortion increased 110% (from 10.6% of abortions to 22.3%). Large increases in medical abortion occurred both from 2004 to 2008 (from 10.6% of abortions to 15.2% [43% increase]) and from 2009 to 2013 (from 17.1% of abortions to 22.3% [30% increase]). In contrast, use of curettage at ≤13 weeks’ gestation decreased 15% from 2004 to 2013 (from 79.7% of abortions to 67.9%). Curettage at >13 weeks’ gestation consistently accounted for approximately 8.0% of all abortions, and all other methods consistently accounted for a small percentage of abortions (<0.1%–1.3%) during 2004–2013.

Race/Ethnicity

Among the 29 areas that reported cross-classified race/ethnicity data for 2013, non-Hispanic white women and non-Hispanic black women accounted for the largest percentages of abortions (37.3% and 35.6%, respectively) and Hispanic women and non-Hispanic women in the other race category accounted for smaller percentages (19.0% and 8.1%, respectively) (Table 12). Non-Hispanic white women had the lowest abortion rate (7.2 abortions per 1,000 women aged 15–44 years) and ratio (121 abortions per 1,000 live births) and non-Hispanic black women had the highest abortion rate (27.0 abortions per 1,000 women aged 15–44 years) and ratio (420 abortions per 1,000 live births). Data for 2013 are also reported separately by race and by ethnicity (Tables 13 and 14).

Among the 20 areas^¶¶¶ that reported by race/ethnicity for 2007 (the first year with available data), 2012, and 2013, abortion rates decreased substantially for all three major racial/ethnic groups: for non-Hispanic white women, the abortion rate decreased 26% (from 9.3 abortions per 1,000 women in 2007 to 6.9 in 2013), for non-Hispanic black women it decreased 25% (from 36.8 abortions per 1,000 women in 2007 to 27.5 in 2013), and for Hispanic women it decreased 33% (from 20.8 abortions per 1,000 women in 2007 to 14.0 in 2013). Abortion ratios also decreased from 2007 to 2013. For non-Hispanic white women, the abortion ratio decreased 23% (from 145 abortions per 1,000 live births in 2007 to 112 in 2013), for non-Hispanic black women it decreased 17% (from 514 abortions per 1,000 live births in 2007 to 425 in 2013), and for Hispanic women it decreased 13% (from 209 abortions per 1,000 live births in 2007 to 181 in 2013).

Marital Status

Among the 42 areas that reported by marital status for 2013, 14.8% of all women who obtained an abortion were married and 85.2% were unmarried (Table 15). The abortion ratio was 46 abortions per 1,000 live births for married women and 387 abortions per 1,000 live births for unmarried women. Among the 31 reporting areas**** that provided these data for the relevant years of comparison (2004 versus 2013, 2004 versus 2008, 2009 versus 2013, and 2012 versus 2013), the percentage of abortions among unmarried women increased 3% from 2004 to 2013 (from 82.6% to 85.3%). The increase from 2004 to 2008 (2%) was larger than the increase from 2009 to 2013 (<1%). Among married women, the abortion ratio decreased 22% from 2004 to 2013 (from 58 to 45 abortions per 1,000 live births), with a larger decrease occurring from 2009 to 2013 (15%) than from 2004 to 2008 (5%). Among unmarried women, the abortion ratio decreased 18% from 2004 to 2013 (from 472 to 389 abortions per 1,000 live births), with a larger decrease also occurring from 2009 to 2013 (11%) than from 2004 to 2008 (5%).

Previous Live Births and Abortions

Data from the 41 areas that reported the number of previous live births for women who obtained abortions in 2013 indicate that 40.2%, 45.6%, and 14.1% of these women had zero, one to two, or three or more previous live births, respectively (Table 16). Among the 33 reporting areas^†††† that provided these data for the relevant years of comparison (2004 versus 2013, 2004 versus 2008, 2009 versus 2013, and 2012 versus 2013), the percentage of women obtaining abortions who had no previous live births changed by <1% from 2004 to 2013; by contrast, the percentage decreased for women who had one to two previous live births (4%) and increased for women who had three or more previous live births (13%). Among the areas included in this comparison, 39.9%, 47.5%, and 12.6% of women had zero, one to two, or three or more previous live births, respectively, in 2004; 40.1%, 45.7%, and 14.2% of women had zero, one to two, or three or more live births, respectively, in 2013.

Data from the 39 areas that reported the number of previous abortions for women who obtained abortions in 2013 indicate that the majority (55.0%) had no previous abortions, 36.2% had one to two previous abortions, and 8.8% had three or more previous abortions (Table 17). Among the 30 reporting areas^§§§§ that provided data for the relevant years of comparison (2004 to 2013, 2004 versus 2008, 2009 versus 2013, and 2012 versus 2013), the percentage of women who had zero or one to two previous abortions did not change appreciably over time: 54.9%, 54.8%, and 54.9% had zero previous abortions in 2004, 2012, and 2013, respectively, and 37.1%, 36.0%, and 36.0% had one to two previous abortions in 2004, 2012, and 2013, respectively. In contrast, among these 30 areas, the percentage of women who had three or more previous abortions increased from 2004 to 2013 but did not change appreciably from 2012 to 2013: 8.0% had three or more previous abortions in 2004, as compared with 9.3% in 2012 and 9.1% in 2013.

Maternal Age and Marital Status by Race/Ethnicity

In certain reporting areas, abortions that were categorized by maternal race and race/ethnicity were further categorized by maternal age and by marital status (Tables 18 and 19). A consistent pattern existed for abortions by maternal age across all race/ethnicity groups, with the smallest percentage of abortions occurring among adolescents aged <15 years (0.2%–0.4%) and the largest percentage occurring among women aged 20–24 years (27.8%–34.1%) (Table 19). A consistent pattern also existed for abortions by marital status across all race/ethnicity groups, with a higher percentage of abortions occurring among women who were unmarried (68.6%–91.7%) than among those who were married (8.3%–31.4%) (Table 19). However, for abortions among unmarried women, the percentage was higher for non-Hispanic black women (91.7%) than for non-Hispanic white (82.8%) or Hispanic women (84.1%) (Table 19).

Weeks of Gestation by Maternal Age, Race/Ethnicity, and Method Type

In certain reporting areas, abortions that were categorized by weeks of gestation were further categorized by maternal age, race, and race/ethnicity (Tables 20 and 21). In every subgroup for these three variables, the largest percentage of abortions was obtained at ≤8 weeks’ gestation. However, by maternal age, 44.9% of adolescents aged <15 years and 57.1% of adolescents aged 15–19 years obtained an abortion by ≤8 weeks’ gestation compared with 63.9%–72.2% of women in older age groups (Figure 3) (Table 20). Conversely 20.6% of adolescents aged <15 years and 11.9% of adolescents aged 15–19 years obtained an abortion after 13 weeks’ gestation compared with 7.1%–8.7% for women in older age groups. By race/ethnicity, 60.3% of non-Hispanic black women obtained an abortion at ≤8 weeks’ gestation compared with 68.7%–71.7% of women from other racial/ethnic groups. Non-Hispanic black women obtained the highest percentage of abortions after 13 weeks’ gestation, but differences across racial/ethnic groups were less apparent than differences across age groups (9.4% for non-Hispanic black women compared with 7.1%–8.0% for women in the remaining race/ethnicity groups).

Among abortions categorized by weeks of gestation and method type, curettage accounted for the largest percentage of abortions within every gestational age category (Table 22). At ≤8 weeks’ gestation, curettage accounted for a smaller percentage of abortions (67.2%) than at any other stage of gestation. At 9–20 weeks’ gestation, curettage accounted for 96.7%–99.1% of all abortions; this percentage then decreased to 90.8% of abortions at ≥21 weeks’ gestation. By contrast, at ≤8 weeks’ gestation early medical abortion accounted for 32.8% of abortions, but at all subsequent points in gestation, medical abortion accounted for only 0.8%–8.2% of reported abortions. Throughout gestation, abortions performed by intrauterine instillation or hysterectomy/hysterotomy were rare (<0.01%–0.8% of abortions).

Abortion Mortality

Using national data from the Pregnancy Mortality Surveillance System (41), CDC identified four abortion-related deaths for 2012 (Table 23). These deaths were identified either by some indication of abortion on the death certificate, by reports from a health care provider or public health agency, or from a media report. Investigation of these cases indicated that all four deaths were related to legal abortion and none to illegal abortion.

The annual number of deaths related to legal induced abortion has fluctuated from year to year over the past 39 years (Table 23). For example, 10 legal induced abortion-related deaths occurred in 1994, four deaths in 1995, and nine deaths in 1996. Because of this variability and the relatively small number of legal induced abortion-related deaths every year, national legal abortion case-fatality rates were calculated for consecutive 5-year periods during 1973–2012. The national legal induced abortion case-fatality rate for 2008–2012 was 0.65 legal induced abortion-related deaths per 100,000 reported legal abortions. This case-fatality rate was similar to the rate for most of the preceding 5-year periods but lower than the case-fatality rate of 2.09 legal induced abortion-related deaths per 100,000 reported legal abortions for the 5-year period (1973–1977) immediately following nationwide legalization of abortion in 1973. Possible abortion-related deaths that occurred during 2013–2016 are under investigation.

Discussion

For 2013, a total of 664,435 abortions were reported to CDC. Of these abortions, 652,582 (98.2%) were from 47 reporting areas that submitted data every year during 2004–2013, thus providing the information necessary for evaluating trends. These 47 areas had an abortion rate of 12.5 abortions per 1,000 women aged 15–44 years and an abortion ratio of 200 abortions per 1,000 live births. Compared with 2012, this represents a 5% decrease in the total number (from 688,149), rate (from 13.2 abortions per 1,000 women), and ratio (from 210 abortions per 1,000 live births) of reported abortions. Combined with decreases from the previous 4 years (16–19), all three measures of abortion reached their lowest level for the entire period of analysis (2004–2013).

In addition to highlighting changes that occurred among all women of reproductive age, this report underscores important maternal age differences in abortion trends. During 2004–2013, abortion rates for women in their 20s were consistently higher than for any other age group and women in their 20s accounted for the majority of abortions (56%–59%); therefore, they have contributed substantially to overall changes. Conversely, during 2004–2013 women aged ≥40 years had consistently low abortion rates and accounted for a small percentage of abortions (≤3.7%); therefore, they have had a much smaller contribution to overall abortion trends. Nonetheless, among women aged ≥40 years, the abortion ratio continues to be as high as that for women in their 20s. Given the small proportion of abortions that are performed later in gestation among women aged ≥40, which potentially might be completed for maternal medical indications or fetal anomalies, the continuing high abortion ratio among these older women suggests that unintended pregnancy is a problem that women encounter throughout their reproductive years.

The adolescent abortion trends described in this report are important for monitoring progress that has been made toward reducing adolescent pregnancies in the United States. During 1991–2011, the pregnancy rate for adolescents aged 15–19 years decreased 55% to an historic low (43). This decrease was associated with a larger decrease in adolescent abortions (64%) as compared with births (49%) (43). More recent data indicate that the birth rate for adolescents aged 15–19 years decreased 29% from 2011 to 2015 (44–48). The 12% decrease from 2012 to 2013 in the abortion rate for adolescents aged 15–19 years suggests that adolescent pregnancies in the United States continue to decrease and that this decrease continues to be accompanied by substantial decreases in adolescent abortions as well as live births.

The findings in this report indicate that the number, rate, and ratio of reported abortions have declined across all race/ethnicity groups, but that well-documented disparities persist (3–10). Comparatively high abortion rates and ratios among non-Hispanic black women have been attributed to higher unintended pregnancy rates and a greater percentage of unintended pregnancies ending in abortion (49,50). Data from certain reports suggest that differences in abortion indicators between non-Hispanic black women and women of other groups narrowed during 1994–2008 (4,9), but have remained steady since 2008 (10). Data reported to CDC for 2013 and in previous years suggest similar declines in abortion rates among non-Hispanic white and black women. Higher abortion rates among Hispanic compared with non-Hispanic white women have been attributed to higher pregnancy rates, including intended and unintended pregnancies, among Hispanic women (49,50). However, abortion ratios for these two groups have been more comparable: Hispanic women have had a slightly higher percentage of pregnancies that are unintended but are no more likely than non-Hispanic white women to end unintended pregnancies in abortion (49,50). Differences between non-Hispanic white and Hispanic women in abortion rates narrowed only slightly from 2007 to 2013, with large declines occurring among both groups of women.

The findings in this report indicate women are obtaining abortions earlier in gestation, when the risks for complications are lowest (51–54). Among the areas that reported data every year during 2004–2013, the percentage of abortions performed at ≤8 weeks’ gestation increased 4%. Moreover, among the areas that reported abortions at ≤13 weeks’ gestation by individual week, the distribution continued to shift toward earlier weeks of gestation with the percentage of early abortions performed at ≤6 weeks’ gestation increasing 16%. Nonetheless, the overall percentage of abortions performed at ≤13 weeks’ gestation changed little during 2004–2013, and findings from this and other reports suggest that delays in obtaining an abortion are more common among certain groups of women (55–57). Because of the small but persistent percentage of women who obtain abortions at >13 weeks’ gestation, a better understanding is needed of the factors that cause delays in obtaining abortions (55,57–60).

The trend of obtaining abortions earlier in pregnancy has been facilitated by changes in abortion practices. Research conducted in the United States during the 1970s indicated that surgical abortion procedures performed at ≤6 weeks’ gestation as compared with 7–12 weeks’ gestation were less likely to result in successful termination of the pregnancy (61). However, subsequent advances in technology (e.g., improved transvaginal ultrasonography and sensitive pregnancy tests) have allowed very early surgical abortions to be performed with completion rates exceeding 97% (54,62–64). Likewise, the development of medical abortion regimens has allowed for abortions to be performed very early in gestation, with completion rates for regimens that combine mifepristone and misoprostol reaching 96%–98% (65). In 2013, 66.0% of all reported abortions were performed at ≤8 completed weeks’ gestation, and thus the women receiving these abortions were eligible for early medical abortion on the basis of gestational age; 32.8% of abortions at ≤8 weeks’ gestation and 22.2% of all abortions were reported as medical abortions. Moreover, the use of early medical abortion has continued to rise: from 2004 to 2013, the percentage of all reported abortions accounted for by this method increased 110%. In early 2016, the Food and Drug Administration (FDA) updated its approval for use of mifepristone for early medical abortions, extending the gestational age limit to 70 days (≤9 completed weeks) on the basis of extensive safety and efficacy data (66). The definition of early medical abortion will change in subsequent years to include this additional week of gestation in accordance with FDA revised labeling.

The annual number of deaths related to legal induced abortion has fluctuated from year to year over the past 39 years. Because of this variability and the relatively small number of abortion-related deaths every year, national legal abortion case-fatality rates were calculated for consecutive 5-year periods during 1973–2012. The national legal induced abortion case-fatality rate for 2008–2012 was similar to the case-fatality rate for most of the preceding 5-year periods, but was much lower than the case-fatality rate for the 5-year period (1973–1978) that immediately followed nationwide legalization of abortion in 1973.

Limitations

The findings in this report are subject to at least four limitations. First, because reporting requirements are established by the individual reporting areas (21), the collection of data varies and CDC is unable to obtain the total number of abortions performed in the United States. During the period covered by this report, the total annual number of abortions reported to CDC was consistently approximately 70% of the number recorded by the Guttmacher Institute (15,67), which uses numerous active follow-up techniques to increase the completeness of the data obtained through its periodic national census of abortion providers (15). Although most reporting areas collect and send abortion data to CDC, this information is submitted to CDC voluntarily. Consequently, during 2004–2013, five of the 52 reporting areas did not provide CDC data on a consistent annual basis, and for 2013 CDC did not obtain any information from California, Maryland, or New Hampshire.^¶¶¶¶ In addition, whereas most reporting areas that send abortion data to CDC have laws requiring medical providers to submit a report for every abortion they perform to a central health agency, in New Jersey and the District of Columbia, medical providers submit this information voluntarily (68). As a result, the abortion numbers these areas report to CDC are incomplete.***** Moreover, even in states that legally require medical providers to submit a report for all the abortions they perform, enforcement of this requirement varies and as a consequence several other reporting areas tend to provide CDC with incomplete numbers.^{†††††}

Second, because reporting requirements are established by the individual reporting areas, many states use reporting forms that do not follow the technical standards and guidance CDC developed in collaboration with the National Association of Public Health Statistics and Information Systems. Consequently, many reporting areas do not collect all the information CDC compiles on the characteristics of women obtaining abortions (e.g., maternal age, race, and ethnicity). Although missing demographic information can reduce the extent to which the statistics in this report represent all women in the United States, five nationally representative surveys of women obtaining abortions in 1987, 1994–1995, 2001–2002, 2008, and 2014 (5–8,10) have produced percentage distributions for most characteristics that are nearly identical to the percentage distributions reported by CDC. The exception is the percentage distribution of abortions by race/ethnicity. In particular, the percentage of abortions accounted for by non-Hispanic black women is higher in this report than the percentage determined on the basis of a recent nationally representative survey of women obtaining abortions (10). Differences might be attributable both to the high degree of imprecision for this variable that reduces the reliability of national survey results (8,9) and because the number of states that report to CDC by race/ethnicity continues to be somewhat lower than for other demographic variables. Importantly, some reporting areas that have not reported to CDC, or have not reported cross-classified race/ethnicity data (e.g., California, Florida, and Illinois), have sufficiently large populations of minority women that the absence of data from these areas reduces the representativeness of CDC data.

Similar to the case for race/ethnicity, the absence of medical abortion as a specific category on the reporting form used by some states (68) might reduce the accuracy of CDC’s estimates of the use of this method relative to other abortion methods. Furthermore, even in states with medical abortion on their reporting form, it is possible that this method is disproportionately undercounted. A higher percentage of the abortions provided in physician’s offices and smaller caseload facilities are early medical abortions (14,69), and these practices might be difficult to locate in the wider medical community without active surveillance efforts (69). Nonetheless, a recent comparison of CDC data with mifepristone sales data^§§§§§ suggests that CDC’s Abortion Surveillance System accurately describes the use of early medical abortion relative to other abortion methods in the United States (70). In 2013, however, professional clinical practice guidelines were updated midyear to extend the gestational age eligibility for early medical abortion to 70 days (≤9 completed weeks) (66,71). Findings in this report do not reflect this midyear practice change, only classifying abortions as early medical abortions when performed at ≤8 weeks’ gestation.

Third, abortion data are compiled and reported to CDC by the central health agency of the reporting area in which the abortion was performed rather than the reporting area in which the woman lived. Thus, the available population (27–36) and birth data (37), which are organized by the states in which women live, differs in some cases from the population of women who undergo abortions in a given reporting area. This likely results in an overestimation of abortions for reporting areas in which a high percentage of abortions are obtained by out-of-state residents and an underestimation of abortions for states where residents frequently obtain abortions out of state. Limited abortion services, more stringent legal requirements for obtaining an abortion, or geographic proximity to services in another state might influence where women obtain abortion services. To adjust for these reporting biases, CDC attempts to categorize abortions by residence in addition to geographic occurrence. However, in 2013, CDC was unable to identify the reporting area, territory, or country of residence for 11.9% of reported abortions.

Finally, reporting areas provide CDC with aggregate numbers rather than individual-level records. Because CDC does not obtain individual-level records, stratified analyses by socioeconomic status cannot be done.

Public Health Implications

Ongoing surveillance of legal induced abortion is important for several reasons. First, abortion surveillance is needed to guide and evaluate the success of programs aimed at preventing unintended pregnancies. Although pregnancy intentions are difficult to assess (72–79), abortion surveillance provides an important measure of pregnancies that are unwanted. Second, routine abortion surveillance is needed to assess trends in clinical practice patterns over time. Information in this report on the number of abortions performed through different methods (e.g., medical or curettage) and at different gestational ages provides the denominator data that are necessary for analyses of the relative safety of abortion practices. Finally, information on the number of pregnancies ending in abortion are needed in conjunction with data on births and fetal losses to more accurately estimate the number of pregnancies in the United States and determine rates for various outcomes of public health importance (e.g., adolescent pregnancies) (43,80).

According to the most recent national estimates from 2010, 18% of all pregnancies in the United States end in abortion (81). Multiple factors influence the incidence of abortion. These factors include access to health care services, including contraception (82–84); the availability of abortion providers (14,15,85–87); state regulations, such as mandatory waiting periods (88), parental involvement laws (89), and legal restrictions on abortion providers (90,91); increasing acceptance of nonmarital childbearing (92,93); shifts in the racial/ethnic composition of the U.S. population (94,95); and changes in the economy and the resulting impact on fertility preferences and use of contraception (96,97). However, because unintended pregnancy precedes nearly all abortions (26),^¶¶¶¶¶ efforts to reduce the incidence of abortion need to focus on helping women, men, and couples avoid pregnancies that they do not desire.

Providing women and men with the knowledge and resources necessary to make decisions about their sexual behavior and use of contraception can help them avoid unintended pregnancies. Recent data indicate that the proportion of pregnancies in the United States that were unintended decreased from 51% in 2008 to 45% during 2011–2013, after slight increases from 2001 to 2008 (98). One factor that might have contributed to this decrease is the increase that occurred during the same period in the use of the most effective forms of reversible contraception, specifically intrauterine devices and hormonal implants, which are as effective as sterilization at preventing unintended pregnancy (99,100–102). Although use of intrauterine devices and implants has increased in recent years, use of these methods remains low in comparison with use of oral contraceptives and condoms, both of which are less effective at preventing pregnancy (99,101). Research has indicated that providing contraception for women at no cost can increase use of these methods and reduce abortion rates (82–84). Cost, as well as insufficient provider reimbursement and training, inadequate client-centered counseling or youth friendly services, and low client awareness of available contraceptive methods, are common barriers to accessing contraception (103). Removing these barriers can help improve contraceptive use, thereby reducing the number of unintended pregnancies (103,104) and consequently the number of abortions performed in the United States.