Estimation issues for the incidence of illegal or underreported abortion
by Pablo Mestrovic
Photo Credit Anna Nekrashevich with Pexels
Collecting accurate figures about the incidence of abortion is remarkably difficult. This is the case not only in countries where abortion is illegal, but also in those where the practice is legal but the healthcare system either can not or does not want to provide accurate figures. Generally speaking, no country with a low or medium Human Development Index (i.e. below 700) keeps accurate official registries of the incidence of abortion.
In order to remedy this shortcoming, several methods have been developed for estimating the incidence of abortion. The two most broadly used methods were developed by pro-choice organizations. The residues method (also known as proximate determinants of fertility framework) was developed in 1982 by Population Council’s researcher John Bongaarts and the Abortion Incidence Complications Method (AICM) was developed by the Guttmacher Institute in the early 1990s
Residues method yields very high figures.
The residues method, developed by John Bongaarts, involves using a complex equation including possible fertility (the total number of children a woman would be able to have along all her life should she not use any family planning method nor had any fertility issues), the contraceptive prevalence rate, and other data in order to get an abortion rate/ratio. The main point is that Bongaarts’ method is extremely likely to yield very high abortion rates/ratios. For example, the residues method gave a figure of around 500,000 abortions in Argentina for 2005.
AICM also produces alarmingly high estimates.
The Abortion Incidence Complications Method (AICM) is simpler, but the bias takes a different form. This method takes the figure of women treated for complications of induced abortions and then uses an estimated multiplier for the number of abortions corresponding to every hospitalization. The multiplier is calculated by making a survey of “public health experts.” This item does not include just obstetricians, gynecologists and doctors in general, but also people like “community activists” or “activists of women’s organizations,” which looks much like a code word for pro-choice activists. Moreover, given the current polarization regarding the abortion issue, the recruitment of a research team without a bias in one sense or the other is quite difficult.
The complications method has resulted in alarmingly high estimates. For example: around 800,000 abortions in 2006 and around 1,000,000 in 2009 for Mexico (44 abortions/100 births). For Pakistan, an estimation of 2.2 millions in 2012 (41 abortions/100 births) was given. For Colombia, the estimated figure was 400.000 (56 abortions/100 births) in 2008. The same exaggerated figures are given for almost every country where abortion is illegal, regardless of their social, economic or cultural features.
Actual figures turn out to be vastly lower than Residues and AICM estimates.
Very often, when abortion is legalized, it emerges that the real figures for the incidence of abortion were less than a third or even less than a seventh of the previous estimations.
For example, in 2006 the AICM estimated 165,000 abortion in Mexico City. After abortion was made legal in Mexico City in 2007, the number of abortions performed in government-run hospitals was less than 20,000 a year (data about abortions from private healthcare providers are not available, but given the level of widespread reliance on government-run healthcare in Mexico, the figure should be lower).
Similarly, in 2005 the Residues method estimated 500,000 abortions in Argentina. Abortion was legalized in Argentina in 2020, and in 2021 and 2022, the first two years in which the practice was legal, actual figures were 73,487 and 96,664, for each year.
Other methods of estimating
Other methods which are less used involve interviewing people close to post-abortive women, using probabilistic samples, the so-called “confidante method.” The reliability of this method is compromised by the controversial nature of the issue and the influence which the attitude towards abortion or experiences with it could have.
Abortion providers surveys are used in countries where abortion is legal, yet official figures are not reliable. This method has been used in India with somewhat good results.
The “Bearak” Method
More recently, a team of researchers linked to the medical magazine The Lancet and Guttmacher Institute, and led by sociologist Jonathan Bearak developed what they call a new model for estimating the incidence of unintended pregnancy and abortion. The model is based on an equation involving factors like the rate of unmet need for contraception, the contraceptive failure rate, and nuptiality, among others. The most recent publications of the Guttmacher Institute regarding the incidence of unintended pregnancy and abortion are based on this research.
Estimating abortions
For Argentina, an estimation was made of 368,000 abortions a year for the period 2015-2019. As mentioned before, figures after legalization ranged between only 20% and 30% of that estimation. This team of researchers has tried to make estimations for as many countries as possible, including those where abortion is legal, in contrast with previous estimations which focused only on some countries, which were regarded as representative of a given region.
This is not the only case in which Bearak and his team get outcomes of dubious credibility. The estimation for Chile reproduced on Guttmacher’s website was a yearly average of 170,000 abortions for 2015-2019. In the same period, the number of births in that South American country went down from 245,000 to 210,000 a year. That means that Chile would have one of the world’s highest abortion ratios (69 abortions/100 births for 2015 and 80 abortions/100 births for 2019).
Most countries defined by the World Bank as “high income economies” for 2019 have liberal abortion laws, and the only demographically relevant exceptions are Chile and Poland. Let’s look at Poland’s figures. According to Bearak and his team’s estimations, the Eastern European country had an average of 255,000 unintended pregnancies and 93,000 abortions between 2015 and 2019. On the one hand, Poland’s proportion of unintended pregnancies ending in abortion is lower than the average for high income countries with legal abortion (37 vs 43%). But, surprisingly, Poland’s incidence of unintended pregnancy and abortion went down 58% and 76% each since 1990. Abortion was formally banned in Poland in 1997, after being severely restricted in practice since 1990. Yet, Bearak’s research implausibly concludes that “We found no evidence that abortion rates were lower in settings where abortion was restricted” (page 8 of this PDF).
Bearak’s data undermined his own pro-choice conclusion. We mention this case just to show the theoretical and methodological weakness this new estimation model displays. After having given absurdly high estimations for abortion in all pro-life countries, the Guttmacher Institute claims that the incidence of abortion declined after a strongly pro-life policy was adopted.
Bearak’s research revises up previous estimations of the Guttmacher Institute itself, which gave worldwide abortion figures in the range of between 40 and 50 million, to an average yearly figure of 73 million abortions for the 2015-2019 period. For the same period, the total number of births worldwide went from 142 million to 137 million. So the estimation for the average abortion ratio between 2015-2019 was more than 50 abortions/100 births.
The “Koch” Methods
Finally, we must mention the main estimation methods used until now by pro-life researchers: specifically the ones developed by Chilean doctor Elard Koch. Koch developed two methods.
The first is similar to the AICM, but the difference is that Koch estimates the incidence of spontaneous abortion (miscarriage) at a higher level than pro-choice researchers had estimated. Under the AICM, women hospitalized for complications of miscarriage were often counted as having induced abortions rather than miscarriages. Under Koch’s method women hospitalized for complications of miscarriages were more likely to be counted as having actually had miscarriages, so the estimation of abortions decreases considerably.
The second Koch method involves taking the abortion figures for Spain in 1987 and extrapolating it to other countries. Koch takes for granted that 1987 was the first year in which official, reliable abortion figures were available for Spain.
Both methods give strikingly low figures (abortion ratios between 5 and 10 abortions/100 births). For the first method, Koch doesn’t take the official figures for causes of hospitalizations, replacing them with his own estimation of the probability of spontaneous abortion. In the second method, Koch doesn’t take into account that the abortion law in force in Spain was vaguely drafted regarding the grounds for allowing abortion. Up to the 2000s, many private healthcare providers failed to fully report the abortions they performed out of uncertainty over the legality of these procedures. The shocking increase in abortion figures between 1985 and 2010 can be explained, at least in part, not by an increase in the incidence of abortion but by reporting issues.
However, the deterioration in the levels of scientific integrity and the aggravation of ideological distortions in pro-choice investigation as shown in Bearak’s “new model” make Koch’s estimations a worthy, yet insufficient, effort.
In summary
The near-total impossibility of collecting accurate figures on the incidence of abortion in countries or jurisdictions where it is illegal or the healthcare system does not provide them allows for ideologically biased estimations to be made. Perhaps in a context of lower political polarization regarding the issue of abortion, it would be possible to develop and use more reliable and effective estimation methods.
