World Abortion Estimates: An Audit

Part IV: Piling Error upon Error

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

In our previous sections, we have seen how biases and inherent weaknesses can corrupt the collection and projection of world abortion data, especially in developing countries lacking national health data collection systems. Hospital studies, we explained, may be limited and non-representative. Community surveys and mortality studies extrapolate assumptions that may be invalid. And so-called "expert studies" may be the product of individuals or groups with pro-abortion agendas whose experiences are atypical.

What happens, however, when these mistakes and false assumptions are piled on top of one another and then inflated yet again? This month's excerpt will look at how statisticians at the World Health Organization (WHO) and the Alan Guttmacher Institute (AGI) translate these limited data into the questionable national or regional estimates of abortions, "unsafe abortions," and abortion-related maternal deaths that are so widely publicized in the press and government reports.

Extrapolating Poor Data Inflates Number of Abortions

To translate limited data into an estimate of the total number of abortions in a given nation, WHO and AGI analysts go through a couple of steps. First, statisticians try to ascertain how many births there have been in the area covered by the abortion study, using birth rates and population estimates - - statistics which may themselves be flawed.

Then, using the data from the available hospital studies, community surveys, mortality studies, or expert estimates, they estimate the number of abortions in a given area. (Crucially, by definition all abortions are considered "unsafe" in countries where they are not legal.) By dividing the estimated number of abortions by the estimated number of live births in the area, they obtain a ratio of "unsafe abortions" to live births. When this is extrapolated to the country as a whole, it yields a national ["unsafe"] abortion estimate.

All might be well and good if these statisticians started with actual abortion data from the original studies. But, as mentioned in earlier parts of this series, it is not from that original raw data that these international agencies extrapolate.

It comes from figures they have adjusted for the presumed "underreporting" of abortions. The statisticians begin by assuming that the raw numbers they have are too low and adjust their numbers to the level they think they ought to be.

To take one example, WHO researchers say, "Adjustments took into account the existing abortion law (de jure) and its application (de facto), information on the providers of unsafe abortions, and cultural and rural/ urban differences. Careful consideration was also given to fertility rates, and reported contraceptive prevalence in countries."

When data were not available for a country, researchers substituted their own data based on estimates of countries with similar abortion laws, cultures, fertility rates, contraceptive usage, and urban populations. Of course, the error is only compounded if data from the original countries are suspect.

It is after they have "adjusted" these initial study figures to suit their assumptions that WHO statisticians create their estimates of the "unsafe abortion to live birth ratio" and the percent of "maternal deaths due to abortion." What comes next?

They apply the ratio to the number of births that the United Nations Population Division estimated for each country for 1996-2000. (Estimates for countries sharing the political and demographic characteristics mentioned above were substituted when no birth estimates were available for a given nation.)

The use of these birth and population figures from the UN Population Division opens the calculations up to further error. Because the WHO report was published in 1998, using 1997 population figures, most if not all of the birth estimates for 1996-2000 had to be projections. Such forecasts are notoriously unreliable, and some evidence already suggests that the UN population estimates WHO was using may have been too high.

For instance, in its 1998 Revision of World Population Prospects, the United Nations Population Division adjusted its Long-Range World Population Projections for 1995-2150 downward. The agency discovered that fertility had dropped more than it expected.

By 1995, only 17 developing countries, representing less than 4 percent of the world's population, had not reduced their fertility. In a 2001 UN report on population, there was a difference of 54 million people between the UN's low and high total population projections for 2000.

Why is that significant? If birth estimates are overestimated by several million, abortions could be overestimated by millions as well.

As alluded to earlier, whenever these abortions are attributed to countries where abortion is not legal, they are automatically classified as "unsafe abortions." This is how these statisticians arrive at frightening numbers as large as 20 million worldwide "unsafe abortions" a year.

Even while putting forward this claim, WHO concedes that "the legality or illegality of the services may not be the defining factor of their safety." Even AGI points out that the 20 million "unsafe" abortions estimate it puts forward "includes some procedures performed by trained physicians, even though most of these are technically safe."

The truth be told, the danger or "safety" of abortion in these countries has very little to do with its legality. We discuss this dubious assumption linking legality and safety again in upcoming installments of the series.

Uncertainty Remains

Despite the authoritative assertions of the world abortion lobby, it is telling how tentative analysts for AGI and WHO are in their estimates of "unsafe" abortions and abortion-related deaths. AGI calls the estimates for the developing world "not definitive." WHO says its unsafe abortion-related mortality estimates "must be considered rough" and "necessarily have a high degree of uncertainty."

Christopher Murray of Harvard and Alan Lopez of WHO, two of the most cited experts in the field, say of their own cause-of-death estimates, "we do not know how reliable these estimates are," and note special uncertainty in sub-Saharan Africa. Yet according to WHO estimates, this region is responsible for at least 22 percent of all "unsafe" abortions and 42 percent of deaths due to "unsafe" abortion.

Closely examined, this impressive assemblage of statistics begins to look more like a house of cards, one flimsy assumption teetering upon another with very little hard, reliable data at the bottom. But then again, was it ever really a search for the truth that was driving this "research"?

Next: False Assumptions Equating Abortion Legality and Safety.

Laura Antkowiak is NRL-ETF special research assistant. Randall K. O'Bannon, Ph.D., is NRL-ETF director of education and research.