World Abortion Estimates: An Audit
By Laura Antkowiak and Randall K. O'Bannon, Ph.D.
Part IIEstimating Abortion in the Developing World: Assumptions and "Adjustments"
In Part I of this series, we presented estimates of worldwide abortions, unsafe abortions, and abortion-related deaths from the World Health Organization (WHO) and the Alan Guttmacher Institute (AGI). We also noted that regions with the highest totals of unsafe abortions and abortion-related deaths tend to be the most lacking in reliable data.
This issue will explain how WHO builds its "unsafe abortion database" and begin a discussion of the specific types of studies in the database and the methods used by WHO to estimate abortions with these data. Additionally, this article reveals two crucial assumptions that drive the high reported estimates of unsafe abortions and abortion-related deaths: (1) the presumption that nearly every data source "under-reports" abortions; and (2) counting every illegal abortion as an unsafe abortion, even if "technically safe."
WHO's "Unsafe Abortion Database"
To construct estimates in the absence of any extensive national abortion data, WHO researchers appeal to the judgment of "experts," rely on a hodgepodge of community-based or individual hospital studies, and do special extrapolations based on theoretical models. This automatically opens the door to possible bias, non-representative samples, and magnified miscalculations, any of which could severely skew any estimation.
The World Health Organization's own description of how and where it obtains the data it uses in its "unsafe abortion" database raises a lot of red flags. WHO says, "Reports included in the data listing were identified through a search of library databases and by tracing references. Not all articles identified could be traced and the compilation is based on sources available at the headquarters of the World Health Organization, in Swiss libraries and in other reference centers. In addition, data from WHO supported country studies, studies supported by other United Nations agencies and NGOs, papers presented at meetings, unpublished reports as well as information provided by national authorities, other agencies and colleagues around the world are included. [emphasis added].
One can only imagine who these "NGOs" (non-governmental organizations, such as International Planned Parenthood, British-based abortion and family planning agency Marie Stopes International, etc.), "national authorities," "other agencies," and "colleagues" of the pro-abortion World Health Organization might include. WHO, in fact, invites contributions to its database, apparently indiscriminately:
"Epidemiological information on the extent of the problem of unsafe abortion worldwide is needed and readers are encouraged to share their data and reports with us." No standards of data collection are mentioned.
WHO's report contains a listing with some basic information about the studies in its database. We draw from this, in addition to the text, to describe WHO's data and its sources.
National Statistics: In Short Supply
Researchers' descriptions of the type of data they have and use is itself instructive. WHO uses national statistics "in a few cases," but these tend to be available only in the wealthier, more developed countries. Further, WHO believes that "there are no feasible data collection methods that can reliably be used to measure the overall burden of unsafe abortion in the population" in these countries, so its researchers look for other indicators. In the absence of official records, WHO's main sources of data are hospital studies, community surveys, mortality studies, estimates, public sources, and abortionist surveys.
If each kind of data were available for every country, one might be able to cross-check figures and methods and arrive at a fairly accurate estimate. But the totals reported for individual countries or regions, as shown in part I of this series, are often based on only the scantiest of data.
Hospital Studies: The Problems of Miscarriage and Medicine
One method WHO uses to derive estimates of "unsafe abortions" and deaths due to "unsafe abortions" is extrapolating data from hospital admissions for abortion complications and hospital records of those women's deaths. These studies make up the bulk of the data listings for Africa, and as well as a large part of the listings for Latin America and Asia, but are rarely listed for Europe, North America, and Oceania.
While some of the data it bases these estimates on may be authentic, many of the assumptions who bases its larger extrapolations upon are highly questionable and appear to introduce such a large margin of error that they may be totally unreliable.
There are several reasons why the numbers of abortions and abortion-related deaths that WHO reports from hospital studies might be substantially higher than the number that researchers actually count. First, relying on references from 1965 and 1979, WHO statisticians state that between 10% and 50% of unsafe abortions require medical care. This is a huge range of error, and there is no modern documentation verifying the original assumption.
Secondly, WHO says that not all women in that assumed 10% to 50% actually seek that needed health care, depending upon the legal status of abortion and other considerations. Assumptions such as these can produce a very wide range of estimates for abortions, depending on what statisticians assume regarding the prevalence of abortion complications and health care use.
Suppose, for example, that one was able to study admissions for abortion complications at a large hospital in a country where the unborn are protected by law. Imagine that researchers count 100 women as being admitted for abortion-related complications in a given year.
If 50% of "unsafe" abortions require medical care, at one end of the range WHO proposes, the formula would project that 200 "unsafe" abortions took place in the area served by that hospital. If only 10% of "unsafe" abortions require medical care, the other end of the proposed range, however, WHO's formula would project a figure of 1,000 abortions for the communities served by the hospital. WHO researchers do not disclose which percent they considered in their calculation or what allowance they made for abortions that were not treated but should have been, but one can see from the above examples what a wide variation of estimates these formulas provide.
Additionally, one might question the representativeness of such hospital studies. Though some studies are based on a large number of hospitals across a country, more often these studies appear to involve only one hospital. How WHO statisticians were able to adjust for any bias that would result from a non-representative sample of hospitals, such as a study of a major hospital that might attract a disproportionate number of abortion cases, is never really addressed.
WHO must also overcome the fact that most hospital records lump together both induced abortions and spontaneous abortions (miscarriages) under the general category of abortion. WHO does not appear to be particularly concerned by this, however.
To address the issue, WHO simply subtracts out a small fraction of the total that it believes to be actual miscarriages of unborn children between the ages of 13 and 22 weeks. It then assumes that earlier miscarriages are not likely to lead to dangerous and deadly complications, and that mothers suffering these early miscarriages do not go to hospitals.
How WHO handles miscarriages of 23 weeks or more and whether any of its assumptions for miscarriages in general hold for women facing the sort of sanitation issues, nutrition deficiencies, and tropical conditions often found in developing countries is not clear. Any one of these factors could radically change the base number and the subsequent extrapolation.
WHO also assumes that many induced abortions are actually reported as miscarriages, and so the researchers adjust their figures further for this assumed under-reporting, using a "WHO protocol" that they do not describe within the text. While WHO presumes that each of these "corrections" and "adjustments" bring them closer to an accurate number, the truth is that these adjustments may be taking them farther and farther away from the hard data that would serve as a crucial corrective to their own narrow ideological outlook and limited exposure.
Editor's note. Next month we will pick up community surveys. When this series is complete it will appear on the NRLC web page--www.nrlc.org.
Laura Antkowiak is NRL-ETF special research assistant. Randall K. O'Bannon, Ph.D., is NRL-ETF director of education and research.