The Alan Guttmacher Institute   search  
home home about contact e-lists support agi buy help
publications article archive state center media center tablemaker
Issues in Brief

The Limitations of U.S. Statistics

In the spring of 1995, a new issue arose in the bitter abortion debate that has divided Americans for 25 years. It was the use in some instances of a procedure to terminate late-term pregnancies known as intact dilatation and extraction, or D&X. Abortion opponents dubbed the procedure "partial-birth abortion."

A bill to ban the procedure except when it is necessary to save a woman's life was adopted by the U.S. Congress but vetoed by President Bill Clinton in the spring of 1996. The Senate, but not the House of Representatives, lacked the two-thirds of votes ne cessary to override the veto. Following the veto, President Clinton, surrounded by several women who had undergone D&X to terminate wanted pregnancies that had gone seriously wrong and threatened their health, announced that he would have signed the m easure if it had contained an exception to preserve the woman's health as well as her life.

However, the continuing controversy has ignited claims and counterclaims about the frequency of late abortions, regardless of the procedure used, and the justification for making abortion decisions in the late stages of pregnancy. In the general confusion , little distinction is made between abortions performed before the fetus is able to survive, with or without artificial help, and postviability procedures.

The Supreme Court has ruled repeatedly over a 20-year period that prior to fetal viability, a woman has a constitutional right to decide to terminate her pregnancy in consultation with her physician. Even after fetal viability, she must be able to do so t o preserve her life or health. Although it is very difficult to determine when a particular fetus is viable, 23-24 weeks of gestation is generally considered a minimum for viability.

There are few authoritative data to support claims regarding how many late-term abortions are performed, and little understanding of the complexities involved in securing such data. This Issues in Brief describes how abortion data are obtained, what they consist of and why they have inherent limitations.

Who Collects Nationwide Information on Abortion?
Termination of pregnancy is one of the most frequently performed surgical procedures in the United States. There were 1.5 million abortions in 1992, and the incidence is estimated to have declined to 1.4 million in 1994. Although abortion incidence is the subject of some academic research and much anecdotal reporting, nationally valid data are available from only two sources: the federal Centers for Disease Control and Prevention (CDC) and The Alan Guttmacher Institute (AGI).

The CDC initiated its enumeration of abortions in 1969, two years after Colorado became the first state to liberalize its abortion law. AGI began its collection of information in 1974, the year after the Supreme Court handed down its ruling in Roe v. Wade that the Constitution protects a woman's decision to terminate her pregnancy prior to the viability of the fetus. After viability, ruled the Court, abortion could be prohibited except when it was necessary to preserve the woman's life or he alth. The CDC collects abortion statistics yearly. AGI, dependent on private funding, has had to limit its data collection in recent times to every four years.

How Are Abortion Data Collected?
The data collected by the CDC and AGI are complementary, but have different emphases. The CDC, consistent with its federal function, focuses particular attention on the safety of the procedure, while AGI concerns itself with the availability of abortion services throughout the country.

The methods of data collection differ as well. The CDC collects most of its information indirectly, mainly through reports from state health departments. Reports for the 45 states that collect information on abortion and the District of Columbia vary in completeness, with some lacking information on as many as 40-50% of the abortions that occur in the state.

The CDC also conducts limited surveys of abortion providers or makes estimates for the states that do not collect abortion information (Alaska, California, Iowa, New Hampshire and Oklahoma). For information on the type of abortion procedure used and the characteristics of women having abortions, the CDC relies on the reports of the approximately 40 states that collect these data.

AGI, on the other hand, directly surveys all known providers of abortion services, which numbered 2,380 nationwide at last count. As a result, the number of abortions reported by AGI is accepted as the more accurate and is somewhat higher than that report ed by the CDC—by 15%, on average (although the difference has narrowed somewhat in recent years). Most likely, some abortions also go unreported to AGI—primarily office procedures performed by physicians for their own patients—but this numb er is believed to be very small. Underreporting may become more prevalent if methods of nonsurgical abortion become widely available in private physicians' offices.

How and When Are These Data Reported?
As with the collection and publication of all vital statistics, a time lag occurs between the collection of the data and their analysis and publication. Competing agency priorities, funding limitations, the need for administrative clearance and other fact ors tend to delay the release of information at the federal agency level. On the other hand, the difficulties inherent in raising private funds, repeatedly, for a massive information-gathering effort limit AGI's ability to go into the field with greater regularity.

Consequently, the most complete published data are for calendar year 1992. AGI's information for that year was published in May 1994, and the CDC report was issued in May 1996.

What Is Reported By Each Agency?
For 1992, the CDC reported the following information:

•Total number of abortions for the 50 states, the District of Columbia and New York City;

•Number of births and previous abortions among women having abortions;

•Selected socioeconomic characteristics of women having abortions—age, race, ethnicity and marital status;

•Proportion of abortions obtained in each state by residents of other states;

•Length of gestation for abortions obtained at 7-20 weeks of gestation, with no breakdown below seven weeks or beyond 20 weeks;

•Method of abortion, in broad categories that do not distinguish D&X from other late-term procedures; and

•Number of abortion-related deaths.

The most recent AGI survey, also for 1992, reported on the following:

• Total number of abortions for the 50 states and the District of Columbia;

• Minimum and maximum length of gestation at which abortions are performed in the reporting facility;

• Interval between the patient's first contact with the facility and the abortion;

• Whether abortions are performed upon first visit to the facility;

• Cost of abortion to the patient at 10, 16 and 20 weeks of gestation;

• Distance traveled by the patient to the provider;

• Type of provider (private physician, specialized clinic, hospital, etc.); and

• Forms of clinic or provider harassment (blockades, stalking, bomb threats, etc.).

At What Gestational Ages Are Abortions Performed?
As noted above, the only national data on the incidence of abortion by weeks of gestation come from the CDC reports, which are dependent on state-generated information that is often incomplete. States also vary in their methods of recording gestational age: Some use the number of weeks that have elapsed since the woman's last menstrual period (which overstates the length of gestation), and others record the physician's estimate of gestational age. In addition, individual states, over time, have changed their reporting format, making it difficult to observe trends and make comparisons.

The CDC reports group all abortions after 20 weeks of gestation into one category. After the CDC figures are adjusted for underreporting, approximately 16,450 procedures, or roughly 1% of all abortions in 1992, were estimated to have been performed beyond 20 weeks since the woman's last menstrual period (see Table 1).

Table 1. Induced Abortions, 1992
Gestational age Number
Total 1,528,930
<9 weeks 798,850
9-10 weeks 377,570
11-12 weeks 181,960
13-15 weeks 94,060
16-20 weeks 60,040
>20 weeks 16,450
21-22 weeks 10,340
23-24 weeks 4,940
25-26 weeks 850
>26 weeks 320
Note: Numbers are estimates by AGI based on AGI survey data, the CDC abortion surveillance reports and data compiled by the National Center for Health Statistics.

Extrapolating from unpublished data for 14 states compiled by the National Center for Health Statistics (NCHS), it is possible to estimate that two-thirds of these abortions were performed at 21-22 weeks. After 26 weeks, the number of abortion s nationwide is estimated at 320; given the uncertainty of the data, however, the number could be as high as 600.

Either way, these estimates must be viewed as tentative. One uncertainty stems from the limited number of states on which the estimates are based, since these states may not be representative of the nation as a whole and their reported data may be incompl ete. In addition, because the number of providers who perform late abortions is very small, they may have relatively large caseloads; this factor may bias the reporting, depending on whether states in which these providers are located are part of the NCHS sample. There may be errors by clinicians in their evaluation or recording of the gestational age. Finally, although all states report to the NCHS the number of natural fetal deaths beyond 20 weeks of gestation (see Table 2), some of these deaths may err oneously be classified as abortions if the removal of the fetus is accomplished by the same procedure as an induced abortion.=paragraph

Table 2. Fetal Deaths, 1992
Gestational age Number
20-23 weeks 8,152
24-27 weeks 4,567
28-31 weeks 3,635
32-35 weeks 4,107
Source: National Center for Health Statistics, Vital Statistics of the United States, 1992: Vol. II—Mortality, U.S. Government Printing Office, Washington, D.C., 1997 (in press).

Regarding methods of pregnancy termination, 86% of all late abortions appear to be performed by dilatation and evacuation (D&E), and most of the remainder by inducing labor. There is no information on the number performed by D&X, which is a type o f D&E.

What Are the Reasons Women Have an Abortion?
In 1987 and 1995, AGI collected information nationally on the socioeconomic characteristics of approximately 10,000 women obtaining abortions. The results of the 1995 survey show that the women who are most likely to obtain an abortion have an annual inco me of less than $15,000, are enrolled in Medicaid, are aged 18-24, are nonwhite or Hispanic, are separated or never-married, live with a partner outside marriage and have no religious affiliation. Catholics are as likely as the general population of women to terminate a pregnancy, Protestants are less likely to do so, and Evangelical Christians are the least likely to do so.

The only comprehensive source of information on the reasons women give for their abortion decision is from a 1987 AGI survey of 1,900 abortion patients nationwide. The survey deliberately oversampled women having abortions beyond 15 weeks of gestation, although the number was still relatively small (420).

The vast majority of respondents cited a variety of socioeconomic and family considerations as their main reasons for seeking an abortion. Most of the women reported that more than one factor contributed to their decision, with the average number of reaso ns being four. However, 3% of respondents said that the "most important reason" for their decision was concern for their own health, and another 3% cited concern that the fetus had a health problem.

The women having abortions after 15 weeks attributed their lateness in obtaining the procedure to not having realized earlier that they were pregnant (or how long they had been pregnant), having had difficulty in arranging the abortion and (in the case of teenagers) having been afraid to tell their parents they were pregnant.

Is More Accurate Reporting Likely?
The collection of health and vital statistics in a country as vast and decentralized as the United States is a massive undertaking, fraught with problems of definition, compilation and verification. State reports, which form the basis of the collection ef fort in this instance, are often inadequate. Furthermore, they are unlikely to improve in the years to come. In fact, the new welfare reform law, the Personal Responsibility and Work Opportunity Act of 1996, almost guarantees the status quo.

One of the many features of the law is the provision of up to five "bonuses" of $20 million each to states that can show a reduction in the number of out-of-wedlock births in the 1998 fiscal year, along with a lowering of the state's abortion rate (with fiscal year 1995 as the comparison year). The law attempts to guard against the potential manipulation of data by states seeking to qualify for a bonus. A section entitled "Disregard of Changes in Data Due to Changes in Reporting Methods" states: "In making the determination required by subclause (I), the Secretary shall disregard...(II) (bb) any differences between the rate of induced pregnancy terminations in a state for a fiscal year and such rate for Fiscal Year 1995 which is attributable to a change in State methods of reporting data used to calculate such rate."

Given the chronic underreporting (or, in some cases, nonreporting) of abortion statistics by the states, the legislation—in essence—provides them with an incentive to be even less diligent in the future in producing accurate and complete data on abortions obtained by residents of their state.

Data have also become more difficult to collect over time as harassment has escalated at abortion clinics or at the offices of physicians for whom abortion is a large part of their medical practice. The violence and fear engendered by some protests probably increases the reluctance of providers to report abortions to the state health authorities or even to respond to private inquiries from AGI.

In addition, the most difficult data to obtain are from private practitioners who perform a small number of abortions. Should the nonsurgical methods that are currently being introduced on a trial basis gain widespread acceptance in physicians' offices, the completeness of abortion data would be likely to decrease further.

Because information from providers is not likely to improve and, indeed, may erode further, perhaps inquiries regarding abortion should be addressed to the women themselves. However, women are often reluctant to report having had an abortion. In fact, und erreporting of abortion is a common problem in surveys around the world, a factor that then hinders the gathering of information on the number of pregnancies that women experience, especially unintended pregnancies.

The government-sponsored National Survey of Family Growth, a periodic study that is the main source of information about pregnancy and contraceptive use in the United States, seriously undercounts abortion. In the 1988 survey, for example, women reported only 35% of all abortions known to have been performed during the period covered by the survey. Strenuous efforts were made in the survey fielded in 1995 to assure women of confidentiality and thus improve reporting, but preliminary indications are that a sizable proportion of procedures still went unreported.

Data Sources

The results of the CDC's enumeration of abortion procedures are published as surveillance summaries that appear in Morbidity and Mortality Weekly Reports. The AGI surveys and analyses are published in its peer-reviewed, bimonthly journal, Family Planning Perspectives.

back to top

statement of accuracy © copyright 2004, The Alan Guttmacher Institute.