World Abortion Estimates: An Audit

Part III: More Muddled Methods

By Laura Antkowiak and Randall K. O'Bannon, Ph.D.

While American and European countries usually have in place national systems for reporting health statistics, such systems are rare in most developing nations. To develop estimates of abortion and abortion-related maternal mortality in such countries, researchers rely on several different methods, some of them questionable. In our last report, we looked at national statistics and hospital studies. In this section, we look at community surveys, mortality studies, abortionist surveys, and "expert" estimates.

Community Surveys: Assuming "Under-reporting"

Many times, the World Health Organization (WHO) relies on community surveys which involve interviews of what is supposed to be a random sample of women of childbearing age. "The success of this approach," WHO says, "depends on factors such as the legal status of abortion and on the skill of the interviewers."

WHO assumes that women in these studies under-report their abortions and over-report their miscarriages. Additionally, WHO adjusts these numbers based on factors discussed later in this article and on other generally undefended, unexplained "indirect methods." Only one of these methods is briefly identified by name. While the copious literature on abortion "under-reporting" highlights the illegal or immoral status of abortion as the main reason for under-reporting, some of this literature attempts to suggest that abortion numbers could also be higher because some women supposedly forget having had abortions (Jones and Forrest, Demography, February 1992; Okonofua, Studies in Family Planning, 1999; Senlet, Studies in Family Planning, March 2001).

WHO does not divulge the magnitude of its under-reporting adjustments, but its report and its references offer some clues. WHO says that studies on "unsafe abortion" usually report somewhere between 25 and 75 percent of the abortions that actually occur. The ability of its references to support this point, however, are somewhat questionable.

First, many of its sources are actually older studies of U.S. women. At the very least, we can question their applicability to the rest of the world, not only because of the age of such studies, but because of the different attitudes that Americans take toward abortion and health care in general. A study of four foreign countries included in WHO's endnotes also admits to problems with the completeness of its data. Given this example, one can only speculate on the degree of abortion "under-reporting" that WHO and the Alan Guttmacher Institute and their "local experts" suspect in countries where abortion is illegal.

Again, what under-reporting estimates researchers make will dramatically affect their estimates of total abortions and total abortion-related deaths. If one assumes that only 25 percent of abortions in a given country are reported, any raw study data is automatically quadrupled. A community survey showing 5,000 abortions is quickly translated into a claim of 20,000 abortions, and the numbers reported for the country or region soar.

Mortality Studies and Abortionist Surveys:
Too Low and "Just Right"

Mortality studies are supposed to estimate deaths in developing countries where records are poorly kept or not kept at all. These surveys are conducted by a number of different methods, such as by interviewing women about how their sisters died. Family member interviews constitute part of the base of the limited vital registration systems in China and India, two areas reporting a large number of abortions and abortion-related deaths.

Even WHO admits, however, in a 1997 report on the "sisterhood method," that maternal mortality ratios estimated in this manner cannot be considered precise because they "have wide margins of error." If so, this would again introduce a wide range of possibilities for what the abortion-related death rate actually could be. A June 1997 article in the Population Council's Studies in Family Planning journal also finds that the widely used "sisterhood method" produces estimates of maternal mortality that "could fall as far as 33 percent from the true values" and that "tended to be systematically higher than the direct estimates."

Murray and Lopez comment that "verbal autopsy" methods for adults are not particularly refined, and are only just being tested for validity. For these reasons, they use such surveys "sparingly."

Nevertheless, WHO considers mortality studies as legitimate data, though it once again sounds the theme of "under-reporting." WHO concludes that data from these studies also need to be inflated because it believes even these methods underestimate abortion deaths "to protect the family" from shame or embarrassment. WHO never considers or even mentions any reasons individuals or organizations may have for overstating abortions.

The WHO database includes a small number of "abortion provider surveys," which generally produce the highest numbers and which WHO believes to be the most accurate. Where abortion is legal and every abortionist is surveyed, this might be the case. But in conditions where abortionists are also activists arguing for legalization, there is both opportunity and motive to exaggerate demand and dire consequences.

The "Experts'" Estimates: Upward Bias

WHO's database also includes figures identified as coming not from studies, but from "estimates." These sources should give us particular concern about the chance of abortion figures being inflated due to possible bias of the person doing the estimation.

Murray and Lopez caution that estimates of death by epidemiologists, or experts on a particular disease, "tend to over-estimate the amount of mortality from of [sic] a particular disease or condition." They found that they had to adjust downward many of their figures derived from such estimates, because they added up to a total that exceeded the number of estimated total deaths for a particular age-sex group.

Murray and Lopez specifically list abortion-related deaths as one of the categories of death that appear to have been overestimated. Neither the WHO report, nor the report of the Alan Guttmacher Institute (AGI), which also recently published an often-referenced report on world abortions, discloses these issues. Rather than adjusting their estimates downward to correct for these obvious overestimations, WHO and AGI use estimates from "experts" to adjust their figures upward.

What kinds of "estimates" from these "experts" do the data listing contain? For Brazil, every national comparison of the number of abortions to the number of live births comes from an "estimate." These estimates of "unsafe abortions," offered as a percentage of the total number of live births, range from a low of 21% to a high of 95%. This would mean as few as 21 abortions or as many as 95 abortions for every 100 live births.

Similarly, the number of abortion estimates in Chile's data listing range from a low of 33 percent of live births to a high of 74 percent of live births. When abortion ratios are calculated from data sources other than "estimates," they do not exceed 38.3 percent. Almost always the ratio is lower than 20 percent.

For Colombia, three national hospital studies place abortions somewhere between 8 and 18 percent of the number of live births. Several "estimates," however, peg the total much higher, between 21 and 39 percent of live births. These countries are located in South America, where WHO claims that a staggering three million "unsafe abortions" are performed.

We suspect that this same upward bias may also be present in what WHO refers to as "public source data." These data also apparently come from "experts," as our search through the database listing shows that the "public source" is typically journal articles and reports or unpublished documents from organizations with established abortion agendas.

This type of "data" is most plentiful in the listings for South America, to which WHO attributes the highest and second-highest rates of abortion-related deaths and unsafe abortions, and there is not always other sources of data to correct or adjust these estimates. Data from Uruguay, for example, is comprised only of "public sources."

These reports cover only maternal and abortion deaths, and their estimates for abortion mortality as a percent of maternal deaths range from 5 to 36 percent. (Even 5 percent would be extraordinarily high in Europe or the U.S.) The organizations responsible for these reports? The World Health Organization, Pan-American Health Organization, the Population Council, and the International Planned Parenthood Federation!

We should therefore be wary when the Alan Guttmacher Institute, a special research affiliate of American abortion giant Planned Parenthood, tells us that its 1999 report on "unsafe abortions" is based on sources such as these: a literature review, WHO estimates, and surveys of health professionals in Latin America, Nigeria, and Southern and Southeast Asia that asked respondents about "their perceptions of the characteristics of abortion provision, women's likelihood of suffering complications and being hospitalized and how the situation differs depending on whether women are poor and whether they live in urban or rural areas" (emphasis added).

When we examine the methods used by these researchers to develop these estimates of abortion and abortion-related maternal deaths, it quickly becomes apparent that there may be factors other than science driving the statistics. In the next issue of NRL News, we'll look at some of the political assumptions driving this "scientific" data.