CMMRI

The Chilean Maternal Mortality Research Initiative

Response to Guttmacher Institute criticisms by Koch et al. on the Impact of Abortion Restrictions on Maternal Mortality in Chile
Elard Koch, John Thorp, Miguel Bravo, and Sebastin Gatica, on behalf of the CMMRI

On May 23rd, the Alan Guttmacher Institute (GI) released an advisory comment entitled Review of a Study by Koch et al. on the Impact of Abortion Restrictions on Maternal Mortality in Chile [1]. Following a short background on maternal mortality, this document contains several criticisms that attempt to debunk some conclusions of our article recently published in PLoS ONE entitled Women's Education Level, Maternal Health Facilities, Abortion Legislation and Maternal Deaths: A Natural Experiment in Chile from 1957 to 2007 [2]. After carefully reviewing the document released by the GI [1], we think it contains erroneous and misleading information, which may influence public opinion into disregarding important findings revealed in our article. The following commentary presents a point-by-point rebuttal that will clarify several issues and will help readers to make their own conclusions. For simplicity purposes, we excluded the references within quotes of both our article and the document by the GI. GI: Assessing a new study from Chile A new study by Koch et al. asserts that the expansion of abortion restrictions in Chile in 1989 did not lead to an increase in the incidence of abortion-related mortality. The study concludes that making abortion illegal is not necessarily equivalent to promoting unsafe abortion. However, as detailed below, the study has several serious conceptual and methodological flaws that render some of its conclusions pertaining to abortion and maternal mortality invalid: [1] Authors reply: Which are exactly the serious methodological flaws of our study? The single conclusion that they challenge appears to relate only to the null effect of abortion ban on maternal mortality trend in Chile. More seriously, GI experts do not present any actual evidence in support of any serious

CMMRI
The Chilean Maternal Mortality Research Initiative
methodological flaw in our study.

Therefore, we realize more clarification is needed: our main goal was not assessing the consequences of abortion, as the GI affirms although we understand how delicate this matter is. Rather, we assessed the impact of different factors thought to influence maternal mortality, including womens educational level, better access to maternal health care facilities and professionals, complementary nutrition programs, expansion of sanitary services, among others. The legal status of abortion was no more than another factor subjected to analysis. Regarding the legal status of abortion, we conclude at the end of our article: Finally, prohibition of abortion in Chile did not influence the downward trend in the maternal mortality ratio. Thus, the legal status of abortion does not appear to be related to overall rates of maternal mortality. [2] GI: 1. Chiles pre-1989 abortion law was already highly restrictive, so no conclusions can be drawn about impact of a change from liberal to restrictive laws: [1] Authors reply: This is a major misinformation. In Appendix S1 of our article [2], we clearly describe that the Chilean law allowed abortion for therapeutic reasons with the respective notes. Abortion was legal in Chile from 1931 to 1967, after permission of three physicians or one physician and two witnesses. From 1967 to 1989, abortion was simplified and allowed based on the opinion of two physicians. It is relatively well documented that, until 1989, an undetermined number of elective abortions were conducted by several physicians utilizing the latter law in different regions of the country. In fact, the term therapeutic was loosely interpreted by Chilean physicians, allowing abortions to be performed for socioeconomic or mental health reasons, or simply on request. For instance, in 1973, six months immediately before the Chilean coup d'tat, over 3,000 abortions were conducted on request in Hospital Barros Luco Trudeau, located in the south sector of Santiago. This case is very well documented and was recently reported in the Chilean newspaper The Clinic, as a chronicle ironically entitled La va chilena hacia el aborto (literally in English The Chilean way to abortion) [3]. Flexible interpretation of the practice of therapeutic abortion was progressively restricted and prosecuted until the definitive derogation of all types of abortion in 1989. Thus, it is inaccurate to affirm that: Chiles pre-1989 abortion law was already highly restrictive [1]. Moreover, it is clear that abortion ban in Chile meant a major transition from a partially restrictive to a fully restrictive law in

CMMRI
The Chilean Maternal Mortality Research Initiative
practical terms. GI: Before 1989, abortion was legal in Chile only to save a womans life (the law was sometimes interpreted to allow abortion if a womans health was threatened). This means that on the spectrum of abortion restrictions, Chile was already among the countries where abortion is highly restricted. The further tightening of these already severe restrictions in 1989 then put Chile firmly in the group of countries with the most restrictive lawsthose where abortion is banned under any circumstances [1]

Authors reply: As pointed out above, this is an error or misinterpretation of the Chilean abortion law probably related to misinformation by the GI. But even if the argumentation by the GI experts stating that Chile was already among the countries where abortion is highly restricted [1] was correct, this does not invalidate one of the main conclusion of the Chilean natural experiment: abortion restrictive laws did not influence maternal mortality trends. In fact, maternal mortality ratios steadily decreased over the last fifty years. In other words, it is absolutely possible for developing countries to decrease maternal and abortion mortality without requiring any liberal law of abortion. This conclusion remains unaltered even after the GI criticism. For clarification purposes, the Chilean law (Article 19 of the Chilean Sanitary Code) was redacted in the following terms: Slo con fines teraputicos se podr interrumpir un embarazo. Para proceder a esta intervencin se requerir la opinin documentada de dos mdicos cirujanos (literally in English, Only for therapeutic purposes a pregnancy can be terminated. Carrying out this intervention will require the documented opinion of two physicians). To the letter of the law, the Spanish word teraputico is literally translated to therapeutic. As stated previously and because of the evident ambiguity of the word therapeutic, the interpretative use and abuse of this law, allowed conducting an undetermined number of elective induced abortions in Chile. In 1989, Article 19 was replaced by the text no podr ejecutarse ninguna accin cuyo fin sea provocar un aborto (literally in English no action can be performed whose purpose is to cause an abortion). In practical terms, current law does not prohibit the pre-viable delivery of a fetus to save the life of the mother, because it is interpreted as a medical ethics decision and abortion is not the primary purpose. In terms of Chilean medical practice, exceptional cases, when the life of the mother is at risk, are ethically solved by applying the principle of double effect and the concept of indirect abortion [4]. In other words, medical ethics is sufficient to settle the matter, and the interruption of a pregnancy intended to save the life of the mother is not a legal issue in Chile.

CMMRI
The Chilean Maternal Mortality Research Initiative

Finally, the Chilean experience shows that it is absolutely possible for developing countries to lower maternal mortality rates, without requiring the legalization of abortion. In other words, the fiftyyear natural experiment conducted in Chile strongly supports that abortion legalization is unnecessary to decrease maternal mortality and reach the millennium development goal 5 (MDG-5). It is a matter of scientific fact in our study. GI: Relatively few legal abortions are performed to save the life of the mother or to protect the mothers health. Thus, it is not surprising (and should in fact be expected) that after Chiles law was further tightened in 1989, the proportion of all abortions that were illegal did not increase substantially [1] Authors reply: We think this question has already been addressed in our previous rebuttal comments. All the argumentation carried out by the GI is flawed because of serious misinformation and misinterpretations regarding the Chilean abortion law. We appreciate that the GIs acknowledges: the proportion of all abortions that were illegal did not increase substantially [1]. In fact, when we considered the hospitalizations for any kind of abortion in Chile, we observed a continuous reduction from the 1970s. As we already discussed in our article, in 1960, when the leading cause of mortality was abortion, there were 287,063 live births and 57,368 hospitalizations from abortion (whether spontaneous or induced), representing a 5:1 ratio. In the last decade, the ratio between live births and hospitalizations from abortion has remained relatively stable at approximately 7:1 [2]. More research is required to accurately address the factors related to the parallel decrease in complications and hospitalizations from abortion in Chile. The decline of maternal mortality and abortion mortality in Chile occurred in parallel with the decrease of number of hospitalizations attributable to complications from clandestine abortions: while over 40% to 50% of all abortionrelated hospitalizations were attributable to complications from clandestine abortions during the 1960s decade, this proportion decreased rapidly in the following decades; indeed, only 12-19% of all hospitalization from abortion can be attributable to complications from clandestine abortions between 2001 and 2008. Nowadays, over 80% of all hospitalizations from abortion are related to miscarriages or spontaneous abortions. These data suggest that throughout time, restrictive laws may have a restraining effect on the practice of abortion similar to restrictions on tobacco smoking or alcohol consumption that are hypothesized to cause a dissuasive effect on the population when combined with adequate policies improving womens

CMMRI
The Chilean Maternal Mortality Research Initiative
educational level and access to maternal health care. In fact, Chile exhibits today one of the lowest abortion-related maternal deaths in the world, displaying 92.3% decrease from 1989 and 99.1% accumulated decrease over fifty years. More research to assess this important hypothesis emerging after the Chilean natural experiment is required.

GI: 2. The authors rely on a far too narrow, unreliable evidence base: The authors state that they are using empirical evidence as a basis for their claims regarding the incidence of abortion and abortion-related mortality in Chile. However, their exclusive reliance on Chiles vital registration system to assess the incidence and consequences of abortion in a setting where the procedure is highly restrictedand therefore largely clandestineis a critical methodological weakness. [1] Authors reply: On the contrary, this is a major strength in our study. The work was not exclusively limited to describe the trend on maternal deaths. In fact, ours is the first in-depth analysis of a large and parallel time series, year-by-year, of maternal deaths and the simultaneous assessment of their determinants at a country level, including years of education of women in reproductive age, per capita income, total fertility rate, birth order, clean water supply, sanitary sewer, childbirth delivery by skilled attendants, and different historical policies. In this sense, it is a unique natural experiment conducted in a developing country. To accurately assess the trend in maternal deaths causes, a homologation of four codes of International Classification of Diseases (ICDs) was needed; concretely we standardized the ICDs version 7, 8, 9 and 10 to analyze maternal death causes from the original ICD. Now, all this information has been included in our publication at PLoS ONE and it is already available to be utilized by the scientific and medical community interested in maternal health research utilizing long time series or different historical periods at least from 1950s decade. In addition, a rigorous statistical analysis controlled by multiple confounders was performed. To the best of our knowledge, this is the first study counting with sufficient data and time points to simultaneously conduct a multiple time-series regression, a segmented regression, and a pathway regression using a robust autoregressive procedure (ARIMA). It is not a matter of circumstantial or anecdotic evidence or indirect estimates, but rather a matter of scientific data representing actual vital events and every case representing a real woman dying from maternal-related or abortion-related causes whose methodology has been published for the first time in a independent and external peer-reviewed scientific journal. Furthermore, by definition maternal mortality ratio (MMR) is the main indicator to monitor the MDG-

CMMRI
The Chilean Maternal Mortality Research Initiative

5. Thus, we precisely analyzed the impact of the each predictor on MMR over a fiftyyear period, controlling for each other variable including changes or breaks in the observed trends. Finally, in strict scientific rigor, when you are looking to compare different studies, you need to compare the reliability of the data. In this context, studies based on indirect estimates are substantially more exposed to bias than studies based on actual data. Practically the totality of the studies conducted by the GI over the last two decades is based on indirect estimates. Moreover, they are largely based on opinion surveys whose scientific validity is unknown. In contrast, the Chilean study is very transparent and all the information was made available at PLoS ONE. In fact, because our extremely controversial finding related to the null impact of the illegal status of abortion, the manuscript was intensively peerreviewed by independent scientists. Furthermore, we chose PLoS ONE after editors of The Lancet rejected even conducting a peer-review of our article. This is not surprising, especially when we consider that the Chilean natural experiment seriously challenges, for the first time, several reports most of them based on indirect estimates of induced abortions in developing countries by researchers from the GI published in this prestigious medical journal over the last decade. GI: To properly understand the impact of a clandestine practice, it is necessary to probe much further. A body of research using data sources such as surveys of women and surveys of health professionals has been developed, peer reviewed and published in scientific journals in recent decades specifically to address the severe limitations of registration systems in measuring the incidence and consequences of unsafe abortion. [1] Authors reply: Surveys of women and surveys of health professionals have been long used by researchers of the GI, almost to the point of abuse. Although we grant that some anecdotic and qualitative value can be drawn from such instruments, from an epidemiological viewpoint they are flawed to accurately quantify the number of induced abortions: these are purely indirect approaches that may lead to under- or overestimations, and due to the subjective nature of opinion surveys, they can be extremely biased. Moreover, as we pointed out, the scientific validity of these methodologies is unknown. We explored the serious methodological flaws by the GI in a recent independent peer-reviewed publication [5]. We found that the GI grossly overestimated abortion rates in all analyzed countries. Overestimation by this methodology can also be observed by contrasting actual figures of abortion from countries with liberal abortion laws, such as Mexico, with figures estimated with the methodology developed by the GI: employing this methodology, in 2006,

CMMRI
The Chilean Maternal Mortality Research Initiative

researchers from the GI estimated between 700,000 and 1,000,000 clandestine abortions per year in Mexico and published this finding in their own in-house, not independently peer-reviewed journal [6] and therefore we consider that publication bias cannot be ruled out. Obviously, this large black figure of induced abortion triggered the alarm in the government and the public opinion. In fact, this study was pivotal for the legalization of abortion in Mexico FD on April 24th, 2007. But, what has really happened in Mexico FD since then? The case of this country provides a natural laboratory to assess what occur after abortion legalization. The number of elective induced abortions registered since abortion legalization until 2011 inclusive totals 71,937, according to the GIRE (Grupo de Informacin en Reproduccin Elegida, in English, Group of Information in Elective Reproduction) [7]. Moreover, after abortion legalization, elective abortions have shown a clear upward trend year-by-year, from 13,404 in 2008 to 20,314 in 2011 [7]. The case of Mexico allows making four important conclusions: First, even though some degree of under-report in the number of elective abortions may occur in Mexico at the present, the methodology based on surveys conducted by the GI leads to at least a 30-fold overestimation in the number of induced abortions conducted before abortion legalization in 2007 (i.e. 700,000 according to the GI versus the actual figure of 20,314 in 2011). Second, abortion registry in Mexico suggests that abortion legalization is promoting a continuous increase in the number of induced abortions. This is not surprising and has been well documented in developed countries that have legalized abortion. For instance, the Ministry of Health of Spain keeps a precise registry of elective abortion since 1987, a year after its legalization in this country [8]. In fact, elective induced abortion figures from Spain have steadily increased between 1987 and 2008, from 16,766 to 115,812. These epidemiological observations based on actual data directly contradict one of the main conclusions by the GI researchers over the last decades: liberal laws do not control the problem of induced abortion; they just increase its incidence, such as occur in a modern epidemic. Third, since the methodology employed by the GI is indirect and mainly based on opinion surveys, it can lead to overestimations that reach beyond what is empirically plausible, calculating figures of abortions comparable to the number of live births. For instance, in Colombia, 715,453 live births were observed during 2008, and researchers from the GI estimated 400,400 abortions in another in-house journal without independent peer-review, and therefore, again publication bias cannot be ruled out [9]. This leads to a figure of pregnancies beyond the empirically possible biological reproductive rate for that country. In fact, using valid

CMMRI
The Chilean Maternal Mortality Research Initiative
epidemiological methods of estimation, we observe that the opinion-based methodology by the GI researchers overestimated by a factor of 18 the possible number of induced abortions (400,400 versus 21,978) in Colombia [5].

Finally, similar studies carried out by researchers from the GI have been conducted in Argentina [10], Guatemala [11], Brazil, Chile, Peru, and Dominican Republic [12], reporting elevated and overestimated in a similar way to the case of Mexico and Colombia [5] numbers of clandestine abortions. In contrast, trends of maternal mortality observed in most of these countries show significant progress over the last two decades [13]. Our recently published article [5] supports these assumptions. In this scenario, conclusions by analysis purely based on opinion surveys appear inconsistent and biased: in strict scientific rigor, the body of research, even though published in scientific journals in recent decades, fails to reflect reality. GI: Abortion as a cause of death is often misreported or underreported in countries where the procedure is illegal under all or most circumstances. In Chile, women who suffer complications after undergoing unsafe abortions are highly unlikely to admit to these actions given possible criminal sanctions (including prison sentences for having obtained abortions). For the same reason, many women may not seek medical help for abortion-related complications. Likewise, physicians treating women for postabortion complications may misreport (or not report at all) deaths and injuries from unsafe abortion to protect their patients from criminal sanctions. [1] Authors reply: This speculative assumption by GI experts is not likely to be a relevant factor influencing the results in the Chilean case. In fact, this subject has been already discussed in the Limitations section of our article and was intensively and critically evaluated during the external and independent peer review process at PLoS ONE [2]. In Chile, integrity and medical certification of global deaths is practically 100% and misclassification is between 2% and 4% considering the general registry of all death causes [14]. In Appendix 2 of the WHO report [15], Chile appears along with 63 countries with civil registration data characterized as complete with good attribution of cause of death. Chile is classified in the A list by the United Nations regarding vital statistics information from a long time ago. The current registry of maternal death causes has 100% integrity and practically null misclassification due to active epidemiological surveillance and mandatory notification and constant audit of maternal death causes under strict rules of confidentiality [16]. Live births registry is corrected by the method of delayed registry (1 to 7

CMMRI
The Chilean Maternal Mortality Research Initiative
years). Nowadays, delayed registry is between 0.5 and 1%. Thus, this registry is also characterized as complete (Appendix S1 [2] details this information). This is not surprising since currently, 99.8% of live births occur in maternities at the Chilean hospitals. The correction method is carried out by the National Institute of Statistics (INE) and released as an official year-by-year publication. Currently, Chilean INE is the technical referent entity for the Organization for Economic Cooperation and Development (OECD) in Latin America. Therefore, given the strengths and integrity of the Chilean registry of maternal deaths and live births, as well as the widely acknowledged credentials of the Chilean INE, it is highly unlikely that maternal deaths of any kind (including those caused by complications of any kind of abortion) are misrepresented or under-reported. The report of the high number of maternal deaths due to abortion in the early 1960s is further evidence of the high quality of the Chilean registry [17].

GI: The argument that restrictive abortion laws do not have a negative impact on womens health is not supported by the existing body of evidence: In addition to their own study, the authors cite low maternal mortality ratios in Ireland, Malta and Poland as evidence that restrictive abortion laws are safe for women. But women in these countries are known to travel to nearby countries with liberal abortion laws to terminate pregnancies or seek postabortion care. Moreover, these countries are exceptions to the rule. [1] Authors reply: The first problem with the GI statement is that it doesnt appear to be a rule regarding the legal status of abortion. In fact, the extent of abortion restriction is variable among different countries. Moreover, if the causal assumption by GI were correct, you would expect to find abortion mortality at least plateauing in Chile, if not going up. In fact, it went down. We grant that it is possible that travelling to nearby countries may be easier for women from Malta, Ireland, and Poland, since most European countries allow elective abortion. This may be acting as a confounder difficult to control. Abortion prohibition in most Latin America, however, makes it difficult for Chilean women to seek and perform abortions abroad. Thus, it is unlikely that the decrease in abortion mortality in Chile is explained by safe abortion procedures in more liberal countries. While it is true that Malta, Ireland, and Poland currently exhibit very low MMR, in some of them (e.g. Poland) these ratios were already low at the time of passing restrictive abortion laws, possibly due to public policies similar to those promoting the decrease of maternal mortality in Chile. To test this hypothesis, an analysis of

CMMRI
The Chilean Maternal Mortality Research Initiative

10

maternal mortality data from this country, similar to that made with Chilean data, is required. Finally, the evolution of maternal mortality in Poland, Malta, and Ireland is yet to be analyzed in depth in the formal biomedical literature. In fact, such analysis was also lacking for Chile before our publication. GI: Almost all of the countries classified as having the lowest maternal mortality rates in the world allow legal abortion on broad grounds, and almost all of the countries with the highest maternal mortality rates have highly restrictive abortion laws. [1] Authors reply: This is a harmful and risky generalization. As scientists, we are concerned about making causal associations that are not firmly founded on empirical data. This is why we do not make a causal association between prohibition of abortion in Chile in 1989 and the following decrease in maternal mortality. We simply state that maternal mortality trends are uninfluenced by the illegal status of abortion. Likewise, we call the same rigor in causal associations from other scientists. So far, we are unaware of empirical data demonstrating a causal link between prohibiting abortion and an increase in maternal mortality. Rather, there is a collection of anecdotic information that has yet to be analyzed in-depth to test that postulate. Our experiment in Chile shows that both maternal mortality trend and illegal status of abortion are not causally related in one way or another. We think the statement that almost all of the countries with the highest maternal mortality rates have highly restrictive abortion laws is essentially mistaken given the cases of Chile, Ireland, Malta, and Poland and the decreasing trends on maternal mortality observed in most of Latin American countries with restrictive abortion laws. There are, however, developing countries with restrictive abortion laws that still exhibit high MMR, such El Salvador and Nicaragua. Nevertheless, these countries are progressing rapidly to improve maternal health, according to the latest independent global reports on maternal deaths by Hogan et al. [13] and may benefit from the Chilean example of public policies leading to better access to education and maternal health facilities. On the other hand, developed countries such as Spain, Canada and the US, show an increase in maternal deaths over the last decade. Moreover, Guyana legalized abortion in the early 1990s showing no sign of decreasing MMR. Therefore, the statement Almost all of the countries classified as having the lowest maternal mortality rates in the world allow legal abortion on broad grounds is another generalization without a real causal link. It is just a speculative assumption

CMMRI
The Chilean Maternal Mortality Research Initiative
based on purely descriptive data or indirect estimates.

11

As our study [2] shows, there are several factors influencing MMR in a country, and all of them should be taken into account when analyzing their effect on maternal mortality trends. In light of our data, it is likely that an improvement in womens education level, maternal health facilities, nutrition, sanitary services (clean water and sanitary sewer) may help developing countries to reach MDG-5 without having to resort to legalization of abortion. GI: In countries that have liberalized their abortion laws over the past two decades, such as Ethiopia, Nepal and South Africa, the evidence is beginning to demonstrate that abortion law reforms are associated with improved health outcomes for women. [1] Authors reply: Our study [2] shows that womens educational level is the main factor influencing the decrease in Chilean maternal mortality rate: for every additional year of maternal education there was a corresponding decrease in the MMR of 29.3/100,000 live births. Furthermore, education influenced the effect of all other factors analyzed in our study, likely by promoting behavioral changes in and the utilization of maternal health facilities by the female population. Thus, womens educational level should be considered as a major confounder when studying the effect of abortion policies. Unfortunately, to the best of our knowledge, no other study has taken into account such confounder when analyzing trends in maternal mortality. Recent studies on maternal mortality conducted in Nepal [18] and Ethiopia [19] were not controlled by the change in womens educational level, the improvement in maternal health facilities and other significant confounders that may be influencing the decrease in maternal mortality, as our study shows. For example, according to data from the World Bank [20], Nepal shows an important improvement in the educational level of women of reproductive age: an increase of the average schooling years from approximately 1.5 to 4.4, between 1990 and 2010 (just between 2000 and 2010 this figure increased from 2.5 to 4.4); in addition, the literacy rate of women over 15 years-old has steadily increased, exhibiting figures of 9.2%, 17.4%, 34.9%, and 46.9% for the years 1981, 1991, 2001, and 2009, respectively. Therefore, it can be argued that the decrease in maternal mortality may be accounted for by the improvement in the educational level of women in Nepal, and not by legalization of abortion. On the other hand, the 2010 report of the World Health Organization for South Africa suggests an increase in maternal mortality over the last two decades: in 1990, there were 121 maternal deaths per 100,000 live births while in 2008 this

CMMRI
The Chilean Maternal Mortality Research Initiative

12

figure has increased to 237 [15]. Caution, however, is necessary when making conclusions from indirect estimates for maternal mortality or when they are based on hospital records. It is necessary to note that quality and reliability of hospital records and vital statistic figures from these countries are questionable according to the World Health Organization [15]. Since these studies are likely to be conducted on the basis of partial and incomplete information, it is very difficult to reach any strong conclusion regarding the progress in maternal health of these countries, or to causally associate them to the legal status of abortion. GI: The authors underestimate the incidence of hospitalization for complications from unsafe abortion in Chile: Though it does not affect the estimated trends in abortion mortality, it is worth noting that the authors seem to have underestimated the incidence of induced abortion complications treated in hospitals, by overestimating the proportion of all abortion cases that are the result of spontaneous abortions (miscarriages). Authors reply: We think the experts from the GI misunderstood the methodology employed to estimate the number of Expected clinical spontaneous abortions, within Table S8 of our article in PLoS ONE [2]. The detailed methodology is already published in an independently peer reviewed journal [5], but the following summarizes the method: we used biological probabilities to calculate the expected number of total viable conceptions on the basis of the number of live births observed in a particular year. According to Wilcox et al. [21], and independently corroborated by Wang et al. [22], approximately 66% of the viable conceptions end in a live birth, 25% of the viable conceptions end as early pregnancy loss, and 8% of the viable conceptions end as clinical spontaneous abortions. Biological probabilities are widely acknowledged, to the point of being repeatedly used in clinical trials to assess the effectiveness of hormonal contraceptives. Thus, from the actual number of live births of a particular year, it is possible to calculate the number of expected viable conceptions for that same year. Then, we calculated the number of expected clinical spontaneous abortions as 8% of the total number of expected viable conceptions. Therefore, early pregnancy loss was excluded from calculations of the expected number of clinical spontaneous abortions, making an overestimation very unlikely. GI: To estimate the numbers of cases that are due to spontaneous abortions, they rely on an unpublished methodology developed by the first author of the current paper. Using this approach, they estimated that 88% of all abortions were spontaneous abortions. By contrast, according to an approach that has been peer reviewed numerous times and published in a range of scientific journals, only 28%

CMMRI
The Chilean Maternal Mortality Research Initiative
of hospitalized abortion cases in Chile in 1990 were spontaneous abortions. [1]

13

Authors reply: The article describing our methodology was in press at the time of publication in PLoS ONE and it has just been published on Friday, May 18th [5]. Again, the experts from the GI seem confused in terms of the methodology used for calculating the expected number of spontaneous clinical abortions. This number corresponds to approximately 8% of total viable conceptions, as already pointed above. This is entirely different from stating, 88% of all abortions were spontaneous abortions. As explained above, the calculation relies on widely acknowledged studies by Wilcox et al. [21] and Wang et al. [22], which observe that early pregnancy loss and clinical spontaneous abortion represent approximately 25% and 8% of all viable conceptions, respectively. GI experts are probably confused by the fact that 80.9% to 87.8% of Observed abortion hospitalizations were explained by Expected clinical spontaneous abortions from 2001 to 2008 [2]. Nevertheless, the latter figure was not calculated on the basis of the number of observed abortion hospitalizations, but from the actual number of live births. It just turns out to be 80.9% to 87.8% when compared to the actual number of observed abortion hospitalizations, and not 28% as estimated by experts from the GI using opinion surveys in 1990 [23]. With regard to the approach that has been peer reviewed numerous times and published in a range of scientific journals [1], the experts of the GI cite only 2 references, which correspond to a review published in-house by the GI [24] and an in-house journal [23], both without independent external peer-review and once more publication bias cannot be ruled out. The methodology employed by GI experts is likely to be overestimating the number of clandestine abortions in Chile and other Latin American countries, as we have reported in our recent article [5]. Not surprisingly, there are gross differences between our expected figure of clinical spontaneous abortions out of observed abortion hospitalizations in terms of percentage (80.9% to 87.8%) and theirs (28%), which is unlikely to be explained by the difference in the years cited: 2001 to 2008 in our case and 1990 in theirs. As we pointed out above, the difference is most likely due to gross overestimations by the GI experts since they use a flawed methodology based on opinion surveys. GI: A body of research, largely published in peer-reviewed journals, makes clear that the decline in maternal morbidity and mortality from unsafe abortion in Chile in the past decades coincides with greater access to and use of contraceptives, as well as the use of less dangerous clandestine abortion methods. Misoprostol, a drug that can be used to induce nonsurgical abortions, is legally and widely available in

CMMRI
The Chilean Maternal Mortality Research Initiative

14

Chile; womens groups have helped make misoprostol available to women seeking abortions, and abortion providers and women themselves have been using the drug to terminate pregnancies since the 1990s. The use of misoprostol as an abortifacient is associated with a lower risk of severe health consequences than the use of illegal surgical procedures, and is considered an important explanatory factor in the decline in abortion-related deaths in the past two decades. [1] Authors reply: Explaining the decrease of maternal mortality ratio in Chile as a result of using drugs such as misoprostol is speculation unsupported by our epidemiological data. Clearly, no study currently exists to date, that seriously supports a decline in maternal mortality associated with the use of abortifacient drugs such as misoprostol in Chile. The two quoted articles by GI experts [25, 26] correspond to mere opinions without any epidemiological evidence or quantitative data supporting such claims. Therefore, this is just a speculative assumption. Indeed, our study shows that global maternal mortality ratio as well as mortality by abortion steadily decreased from 1965-1967. This was well before the development and commercialization of misoprostol. In fact, misoprostol was introduced in the Chilean black market in the late 1990s, making it extremely unlikely that its introduction had any important influence on overall rates of maternal mortality, which were already significantly reduced at that time. Moreover, misoprostol is not sold to the public as an over-thecounter drug in Chile. Its sale is conditioned to a prescription by a physician and solely for the treatment of gastroenterological conditions. The prescription is retained by the pharmacist and audited month-by-month by the Chilean Institute of Public Health (ISP). This extensively limits its prescription for purposes other than its only indication approved by the ISP, i.e. duodenal ulcer disease or prophylaxis to avoid gastric ulcer provoked by NSAIDs. Thus, legally distributed misoprostol in Chile is strictly regulated and controlled by Chilean sanitary authorities. Any other conclusion regarding the legal distribution of misoprostol in Chile is a harmful misleading piece of information by experts of the GI. GI: By helping to reduce unintended pregnancy, family planning programs also help to reduce recourse to unsafe abortion. Contraceptive use has increased substantially in Chile since the 1960s. The initial rise in contraceptive use ushered in an extended decline in the incidence of unsafe abortion and abortion-related hospitalizations. [1] Authors reply: Regarding the role of contraception we stated in our Article: It is well documented that the Chilean program providing contraceptive methods after clandestine abortion was effective in decreasing abortion rates. In addition, the methods used to conduct clandestine abortions at present may have lower rates of

CMMRI
The Chilean Maternal Mortality Research Initiative

15

severe complications than the methods used in the 1960s, mainly based on highly invasive self-conducted procedures. Therefore, the practically null abortion mortality observed in Chile nowadays can be explained by both a reduced number of clandestine abortions and a lower rate of severe abortionrelated complications. This phenomenon also seems to be related to joint-effects between increasing educational level and changes on the reproductive behaviour of Chilean women, an observation that requires further research. [2] In addition, we analyzed the impact of total fertility rate (TFR) as an indicator of the reproductive behavior of Chilean women. We found a significant decrease in Chilean TFR from 5.0 to 1.9 between 1957 and 2007, together with an increase in the total percentage of primiparous women, and especially in women older than 29 years of age [2]. Our findings suggest that an important shift in the reproductive behavior of women is likely taking place. Health policies implemented during the analyzed period in Chile have gradually increased the access to a variety of contraceptive methods through the primary care system. Noteworthy, the actual use rate of hormonal contraceptives and intrauterine devices in Chile reaches approximately only one-third of women at reproductive age (36%) and is lower than developed nations [2, 16]. These results suggest that the fertility reduction is not limited to the use of artificial contraceptive methods, and other factors may be intervening such as the dramatic increase of the educational level of women from 1965 [2]. The decrease in TFR strongly correlated with the decrease in maternal mortality. Nevertheless, as we pointed out in our article, when the number of years of education of the female population is included in the explanatory model, the strong correlation between TFR and maternal mortality reduction is substantially attenuated. [2] Therefore, a conclusion of our study is that womens educational level is influencing both MMR and TFR in Chile. A likely explanation for this phenomenon is that increasing education levels would favor utilization of available maternal health facilities, as well as higher autonomy to women, allowing them to take the control of their own fertility using the method for fertility regulation of their preference. [2] We additionally found that For every 1% increment in primiparous women giving birth older than 29 years of age, an increase of 30 maternal deaths per 100,000 live births was estimated. [2] In other words, the delayed motherhood trend in Chile is accompanied by a deleterious effect on maternal health, hindering a further decrease in MMR. This may be a consequence of the increased obstetric risk associated with childbearing at advanced ages. We termed this phenomenon as a fertility paradox, possibly linked to the increasing levels of education that

CMMRI
The Chilean Maternal Mortality Research Initiative

16

produce a change on reproductive behavior, decreasing fertility and delaying motherhood. This remains for further research to fully understand its implications, especially in developed nations where maternal mortality is increasing over the last decade. GI: The evidence on abortion laws, unsafe abortion and maternal health indicates that further reductions in Chiles maternal mortality and morbidity could be achieved by such strategies as liberalizing the countrys abortion law and giving women meaningful access to safe and legal abortion services. Authors reply: As stated in our article in PLoS ONE [2], According to the most recent report published by INE, the MMR for 2009 was 16.9 per 100,000 live births (43 deaths) and the figures of indirect causes (codes O99, O98), gestational hypertension and eclampsia (codes O14, O15), abortion (code O06), and other direct obstetric causes were 18 (41.9%), 11 (25.6%), 1 (2.3%) and 13 (30.2%) respectively. This profile of maternal mortality causes suggests the apparition of a more complex residual pattern of maternal morbidity, unrelated to abortion complications. Accordingly, we cannot understand how the GI experts intend to further decrease MMR in Chile by legalizing abortion: this conclusion does not make any sense. Finally, it is imperative to remark that, because abortion restrictive law in Chile is unrelated to maternal mortality and this country reached one of the lowest rates of maternal-related and abortion-related deaths of the world (at the present, 16.9 and 0.39 per 100,000 live births, respectively) without legalizing abortion, this fifty-year natural experiment provides strong evidence, for the first time, that a liberal law of abortion is unnecessary to improve maternal health: it is a matter of scientific fact in our study. We think this should be recognized by a scientific community guided by principles of honesty and objectivity in science, no matter how controversial the finding might be.

REFERENCES 1. Guttmacher Advisory, May 2012. Review of a Study by Koch et al. on the Impact of Abortion Restrictions on Maternal Mortality in Chile. Retrieved from:

CMMRI
The Chilean Maternal Mortality Research Initiative

17

http://www.guttmacher.org/media/evidencecheck/2012/05/23/GuttmacherAdvisory.2012.05.23.pdf 2. Koch E, Thorp J, Bravo M, Gatica S, Romero CX, Aguilera H, Ahlers I: Women's Education Level, Maternal Health Facilities, Abortion Legislation and Maternal Deaths: A Natural Experiment in Chile from 1957 to 2007. PLoS One 2012; 7(5):e36613. Retrieved from: http://dx.plos.org/10.1371/journal.pone.0036613 3. The Clinic Article: La va chilena hacia el aborto [The Chilean way to abortion]. Retrieved from: http://www.theclinic.cl/2012/03/13/la-via-chilena/ and http://es.scribd.com/doc/94160562/The-Clinic-La-via-Chilena-Hacia-El-Aborto 4. Valenzuela CY. Scientific ethics of therapeutic abortion. Rev Med Chil 2003;131:562-8. 5. Koch E, Bravo M, Gatica S, Stecher J, Aracena P, Valenzuela S, Ahlers I: Sobrestimacin del aborto inducido en Colombia y otros pases latinoamericanos [Overestimation of induced abortion in Colombia and other Latin American countries]. Ginecol Obstet Mex 2012;80(5):360-372. Retrieved from: http://www.nietoeditores.com.mx/download/gineco/2012/MAYO/Femego %20MAYO%2012%20Sobrestimacion.pdf 6. Juarez F, Singh S, Garcia S, Diaz Olavarrieta C: Estimates of Induced Abortion in Mexico: What's Changed Between 1990 and 2006? Int Fam Plan Perspect. 2008;34(4):158-68. Retrieved from: http://www.guttmacher.org/pubs/journals/3415808.html 7. GIRE Group of Information in Elective Reproduction: Cifras de Interrupcin Legal del Embarazo [Figures of Legal Interruption of Pregnancy]. Retrieved from: http://www.gire.org.mx/index.php? option=com_content&view=article&id=504%3Acifrasile&catid=166%3Ainformacon-relevante&Itemid=1397&lang=es and http://es.scribd.com/doc/94847430 8. Spain Government, Ministry of Health, Social Services, and Equality (Accessed on May 23rd, 2012) Interrupciones Voluntarias del Embarazo [Voluntary Interruptions of Pregnancy]. Retrieved from: http://www.msc.es/profesionales/saludPublica/prevPromocion/embarazo/hom e.htm#publicacion 9. Prada E, Biddlecom A, Singh S. Induced Abortion in Colombia: New Estimates and Change Between 1989 and 2008. Int Perspect Sex Reprod Health 2011;37(3):114124. Retrieved from: http://www.guttmacher.org/pubs/journals/3711411.html 10. Mario and Pantelides (2009) Estimacin de la magnitud del aborto inducido en la Argentina. [Estimation of the magnitude of induced abortion in Argentina]. Notas de Poblacion N87:95-120 . Retrieved from: http://www.cepal.org/publicaciones/xml/1/36501/lcg2405-P_4.pdf 11. Singh S, Prada E, Kestler E. Induced Abortion and Unintended

CMMRI
The Chilean Maternal Mortality Research Initiative

18

Pregnancy in Guatemala. Int Fam Plan Perspect 2006;32(3):136145. Retrieved from: http://www.guttmacher.org/pubs/journals/3213606.html 12. Guttmacher Institute (2006) Issues in brief: An Overview of Clandestine Abortion in Latin America. Retrieved from: http://www.guttmacher.org/pubs/ib12.html 13. Hogan MC, Foreman KJ, Naghavi M, Ahn SY, Wang M, Makela SM, Lopez AD, Lozano R, Murray CJ: Maternal mortality for 181 countries, 1980-2008: a systematic analysis of progress towards Millennium Development Goal 5. Lancet 2010, 375(9726):1609-1623. 14. Nuez FM, Icaza NM: Quality of Mortality statistics in Chile, 1997-2003. Rev Med Chil 2006, 134(9):1191-1196. 15. World Health Organization: Trends in maternal mortality: 1990 to 2008. Retrieved from: http://whqlibdoc.who.int/publications/2010/9789241500265_eng.pdf 16. Donoso-Sia E: Unsafe abortion en Chile? Rev Chil Obstet Ginecol 2008, 73(6):359-361. 17. Government of Chile Ministry of Health (2007) Normas Nacionales sobre la Regulacin de la Fertilidad [National Normative on the Regulation of Fertility]. Retrieved from: http://www.redsalud.gov.cl/portal/url/item/795c63caff4ede9fe04001011f014b f2.pdf 18. Samandari et al. (2012) Implementation of legal abortion in Nepal: a model for rapid scale-up of high-quality care. Reproductive Health 9:7. Retrieved from: http://www.reproductive-healthjournal.com/content/9/1/7/abstract 19. Abdella (2010) Maternal Mortality Trend in Ethiopia. Ethiopian Journal of Health Development 24(S1):115-122. Retrieved from: http://ejhd.uib.no/ejhdv24-sn1/115%20Maternal%20Mortality%20Trend%20in%20Ethiopia.pdf 20. The World Bank (Accessed on May 23rd) World Data Bank (BETA) Education Statistics. Retrieved from: http://databank.worldbank.org/Data/Views/VariableSelection/SelectVariables.a spx?source=Education%20Statistics# 21. Wilcox AJ, Weinberg CR, Baird DD: Timing of sexual intercourse in relation to ovulation. Effects on the probability of conception, survival of the pregnancy, and sex of the baby. N Engl J Med 1995, 333(23):1517-1521. 22. Wang X, Chen C, Wang L, Chen D, Guang W, French J: Conception, early pregnancy loss, and time to clinical pregnancy: a population-based prospective study. Fertil Steril 2003, 79(3):577-584. 23. Singh S and Wulf D, Estimated levels of induced abortion in six Latin America countries. Int Fam Plan Perspect 1994;20(1):4-13. 24. Singh S, Prada E, Juarez F: The abortion incidence complications method: a quantitative technique. In: Methodologies for Estimating Abortion

CMMRI
The Chilean Maternal Mortality Research Initiative

19

Incidence and Abortion-Related Morbidity: A Review. Edited by Singh S, Remez L, Tartaglione A. New York: Guttmacher Institute; 2010: 71-98. 25. Donoso-Sia E: The reduction in maternal mortality in Chile, 1990-2000. Rev Panam Salud Publica 2004, 15(5):326-330. Retrieved from: http://www.scielosp.org/pdf/rpsp/v15n5/22005.pdf 26. Shepard BL, Casas Becerra L: Abortion policies and practices in Chile: ambiguities and dilemmas. Reprod Health Matters 2007, 15(30):202-210.