Methodological Aspects of Gathering and Measuring Information

The information sources and the reliability of the data available on the practice of abortion are closely linked to its legal status. Various authors have pointed out the need to measure the occurrence of abortion in order to prove its consequences for the health and well-being of women and their families, as well as to determine its eventual impact on fertility and contraceptive demand. This accounting may help, among other things, to make politicians aware of the need to liberalize laws on the matter. Emphasis has also been placed on the difficulty of undertaking surveys on this issue, as well as on the limitations on the information available through hospital records, even in contexts where abortion is legally authorized (Huntington et al., 1995; Guillaume, 2004). The type of methodology used to gather information should be determined by the aims of the study and the quality of the available data sources. In the case of abortion, the approach and analytical strategy used will be different, depending on whether the aim is to estimate the frequency with which it occurs or to describe the circumstances in which it is performed. This is also true for attempts to define and evaluate intervention programs, determine decision-making processes, illustrate particular experiences, or determine the actors that participate in the various procedures for inducing abortions or dealing with the complications of the latter, among other aspects.

What Information and Data Sources Are Available?

Statistics from Health Systems and Hospital Records

In countries where abortion is legal, records from health systems provide more reliable, accurate information, although information from them is still incomplete and underestimated. For example, available records tend not to include abortions performed outside official facilities or abortions undergone by women from one country in other countries where there are more facilities and better conditions. They may also fail to include abortions registered under false names, which are therefore not taken into account. But in countries where abortion is illegal and performed in clandestine contexts, the situation is far worse. In these, the information obtained is partial, incomplete and fragmentary, while estimates on the number of abortions have serious problems of under-registration, misclassification, time lags and generally only referring to cases that end with complications in health services (Zamudio et al., 1999; Núñez Fernández, 2001; LLovet and Ramos, 1998). Under these circumstances, indirect methods tend to be used to obtain more accurate information on the phenomenon, although this alternative is not entirely free of problems.

Death Certificates Obtained from Civil Records

These sources of information, which in theory should provide data on the morbidity and mortality associated with and caused by abortions, have severe deficiencies and problems. On the one hand, the cause of these events is usually inaccurately or under-declared, particularly if the practice is illegal, since registering a death from abortion implies the intervention of the judicial authorities and may have consequences for the providers of this service. On the other hand, in Latin American countries, as in many other developing countries, information on causes of death, of the kind recorded in civil registers, is extremely deficient, particularly in rural and indigenous zones.

Quantitative Surveys and Specific Qualitative Approaches

The types of surveys applied to obtain information on abortion tend to be widely discussed in the literature related to the topic, both in countries where it is legal and where it is not. The issues to be considered in the use of these surveys are linked to the characteristics of the population to be interviewed, the concepts and terms used in these studies, and the design of the questions and the order in which they are given (Huntington et al., 1995). The intimate nature of abortion and its mortal, ethical, cultural, social and legal implications raises several questions to be considered in the conceptual design of these instruments, such as the following: should abortion be dealt with directly and/or using an indirect approach? Should unproductive pregnancies, false labors and menstrual regulations be taken into account? Should the issue be dealt with on the basis of questions on unwanted or unplanned pregnancies? In addition, surveys undertaken in different countries use different methodologies, techniques and information gathering strategies, depending on the objectives of the study and the population or population groups in which they are implemented. Some of them only include closed questions, others are complemented with open questions or systems of secret or self-administered answers, while still others restrict themselves to obtaining quantitative or qualitative information or combining both types of approach. There are also those aimed exclusively at sexually active women of childbearing age. Others apply to pregnant women, adolescents, students, health professionals, other social actors or opinion leaders, or persons related to the press or religious groups. Very few studies directly investigate males.

Some demographic surveys on fertility, reproductive health or health in general include questions on abortion or sections linked to the topic that incorporate aspects that allow an approximate measure of the phenomenon. Through them, it is also possible to construct indicators and variables to determine the characteristics of women that resort to abortion, the conditions in which they do so (duration of pregnancy, method used, place where it is performed, personnel involved, costs, etc.), the decision-making process surrounding it, the participation or influence of males or other actors, as well as the consequences of abortion performed in risky conditions and post-abortion follow-up. Although these survey data may still underestimate abortion, they provide more specific information on the characteristics and practices of those that resort to abortion, although they often fail to distinguish between miscarriages and induced abortions. The main problem of these surveys, however, is that they are not specifically designed to measure or record the wide range of aspects related to the practice of abortion. They usually only include a few questions on the subject, within a wide range of completely different aspects that need to be considered.

Methodologies for Recording Information on Abortion

Various methodologies and techniques have been used to record information on abortion in order to measure the phenomenon and above all, to mitigate or reduce the difficulties involved in grasping and observing it. The conclusion that can be drawn from the analyses of its measurement is both conclusive and generalized: the estimates made are flawed by under-registration and therefore only represent base hypotheses for measuring the scope and approximate intensity of the phenomenon. In an article on the measurement of abortion, Rossier (2003) mentions the existence of eight methods for estimating its frequency, on the basis of different sources of information:

• Indirect estimates based on hospital statistics on abortion complications
• Estimates of deaths from abortion based on death records or statistics on maternal mortality.
• Indirect estimates based on surveys of service providers that perform illegal abortions.
• Estimates based on surveys in which women are asked directly about their abortions.
• Estimates based on prospective studies in demographic follow-up surveys.
• Indirect estimates through surveys using the method of “anonymous report by a third party.”
• Indirect estimates based on the residual method.
• Indirect estimates by experts based on different sources of information (through adjustments, weightings, corrections, etc.)

Singh (2005) provides a summary of the various methodologies for estimating the incidence of abortion, on the basis of different approaches: a) community surveys that may be cross-sectional or prospective or that can use specific techniques such as random answers, self-administered questionnaires, self-interviews with computers or a secret ballot box; b) interviews with abortion providers; c) data on hospital complications of abortion using the indirect estimate method and other methods based on hospitalized women; d) triangulation; e) the World Health Organization (WHO) method for global or regional estimates; f) the Bongaarts residual method; g) the method using an anonymous report by a third part and h) other methodologies such as histories of repeated abortion. Several authors have described the strengths and weaknesses of some of the methods mentioned (Juárez, 2005; Lara et al., 2004; Núñez, 2001; Zamudio, 1998; Sanseviero, 2003).

In this chapter, we shall only refer, for illustrative purposes, to some of the methods available to record empirical evidence on the practice of abortion, on the basis of interviews undertaken on various sectors of the population, primarily women. In this exercise, we will simply describe some of the main advantages and shortcomings of these methods for estimating this practice.

The Direct Observation Method

This method involves asking women about the number of abortions they have had over the course of their childbearing years or during a set period of the same. It usually also seeks to determine the characteristics of these women and the various circumstances in which abortions are performed. It involves both ad hoc surveys and follow-up surveys conducted in various communities or contexts. To this end, questionnaires are applied with varying degrees of breadth and depth, usually aimed at samples of the national population or types of settings such as urban or, rural or specific geographic and social contexts. Under-registration in this type of surveys varies and is associated with the perception, socialization, legitimization and stigmatization of abortion in society. Factors of this nature mean, for example, that some abortions are declared as “miscarriages” (Erviti, 2004). According to some authors, under-registration of abortion varies from 25% to 50% (Llovet, 1998).

Direct observation surveys are generally carried out through face-to-face interviews between the interviewees and interviewers. They are usually applied to women, who may also report on how their partners are involved in this event. In other cases, although infrequently, they involve surveys or interviews of males, who are asked about abortions related to pregnancies that resulted from intercourse they had with a particular woman. A determining factor in the quality of the data is the conditions in which the surveys are carried out, foremost among which are the place of the interview and the time after the abortion when it takes place. Another key factor is the interviewer’s ability to establish a climate of trust with the person to be interviewed, as well as many other factors in dealing with such a delicate issue. Some authors note that the response rate, in relation to the occurrence of induced abortion, appears to be higher when women are questioned at health service facilities (Guillaume, 2004), but others note a lower response rate because women in health facilities services are afraid to be denounced. Others point out that the responses vary in comparison with those obtained outside this atmosphere and after having been attended at hospitals (Núñez Fernández, 2001; Zamudio et al., 1999). Different ways of asking questions have also been tested to improve the quality of the data obtained (Anderson et al., 1998; Huntington et al., 1995). The face-to-face method provides low response rates and does not guarantee anonymity, but does allow analysis of various characteristics, consequences and other aspects concerning the process of abortion (Lara et al., 2001).

Other more sophisticated methods have been used or considered to explore issues as sensitive as abortion, such as sexual practices, violence against women, particularly in developed countries, as in the case of France where these women are interviewed by phone. This method is not viable in developing countries, since telephones are only available to certain sectors and geographical areas of the population. Nevertheless, it has been used in certain studies conducted in the region, as in the case of one conducted in Buenos Aires, in 2002, in which 607 women aged 15 to 69 were interviewed about their preventive practices in reproductive and post-reproductive health. It was decided to conduct telephone interviews, due to the growing lack of safety on the streets which made it difficult for interviewers to be admitted into interviewees’ homes (López and Finding, 2002, 2003; and 2005).

Likewise, surveys have been conducted using questionnaires that are “self-administered” by women (Peláez Mendoza, 1999). In Colombia, Mexico and Brazil, this method has been tested with women who deposit their questionnaires anonymously (LLovet and Ramos, 1998; Zamudio et al., 1999; Núñez Fernández, 2001; Lara et al, 2004; Olinto, 2004), but it also has its limitations. These include the fact that it uses very simple, brief questionnaires, with closed questions aimed at people with higher educational attainment and therefore capable of answering in written form. Another way of using self-administered surveys is through the use of the computerized questionnaire “Audio Computer-Assisted Self-Interview” (ACASI), in which women have to record their answers themselves. This method has been tested in Mexico (Lara et al., 2001; and 2004; Olinto and Moreira Filho, 2004). This instrument has similar limitations and problems to those described earlier. Other features are also required for its use, such as computer equipment, people familiar with this type of tool, as well as electricity, which may not be available in the places where the research is carried out. Despite requiring more resources, this type of technique also does not guarantee more accuracy in the data obtained (Rossier, 2003).

The “random response technique” has also been used to measure behavior that has been underestimated in surveys. This technique involves providing the people interviewed with a series of cards with questions and asking them to say “yes” or “no” to two questions on the cards selected; one dealing with a sensitive topic (abortion) and another dealing with a non-sensitive topic, and in which the probability of choosing between the two cards is already known. A device randomly selects the question to be answered by the interviewee, and according to the answers, estimates are made of the number of abortions, on the basis of the probabilities assigned to both questions. The main limitations of this technique are the woman’s lack of understanding of the procedure to be followed and needing larger samples than those of other surveys. Likewise, its results cannot always be related to the individual characteristics of the people interviewed, or to detailed information on the practice of abortion, unless a different instrument is designed for this purpose (Lara et al., 2005). Nevertheless, according to the studies conducted, this technique appears to permit more declarations of abortions than any other method (Lara et al., 2001; and 2004; Rossier, 2003; Núñez Fernández, 2001; De Souza y Silva, 1998).

As Lara et al. (2001) conclude in their comparison of the measurement of abortion in Mexico using four different survey techniques, the best choice for obtaining optimal results consists of combining the various research techniques: some permit a better measurement of the prevalence of abortion, while others provide more information on the experiences of women that resort to this practice.

The “Anonymous Report by a Third Party” Method

This method, used primarily through qualitative interviews applied to a well-defined population, consists of identifying and locating the social networks in which information on abortion circulates. It involves interviewing a random sample of women or key informants so that they will report on the cases of abortion they have heard about in certain social circles. This method seems more suitable in countries where abortion is illegal and preferably where there are structured social networks. The selection of informants is a key element in determining the quality of the data obtained (Rossier, 2003). This technique is recommendable in small geographical zones or restricted sectors of the population. Some surveys also include questions on the recurrence of abortion among third parties (sisters, friends, etc.). Research conducted using this method encounters problems where estimates are made of the level of abortion practice, due to the risk of counting abortions twice, which would lead to over-estimation. This method seems to have made it possible to make acceptable measurements of the number of cases of abortion in Burkina Faso (Rossier et al., 2006).

The Complications Method

Hospital data are used in the various types of estimates that can be made using this method. It consists of estimating the incidence of abortions on the basis of the data on women hospitalized to be treated for complications derived from either miscarriage or induced abortion. Data obtained in this way must be adjusted to correct for the under-registration due to the non- or mis-reporting of cases and to distinguish miscarriages from induced abortions. In countries where abortion is illegal, hospitals tend to classify provoked abortions as spontaneous or “non-specific.” In order to solve this problem, WHO has proposed a classification that enables the various types of abortions to be distinguished: induced abortions, within which it distinguishes between those that have obviously been provoked from those that may have been provoked and miscarriages. This information is used to make estimates by applying coefficients to obtain the number of provoked or spontaneous abortions, and the data are then adjusted to calculate the total number of abortions among the general population. The adjustments are made according to the estimates of experts in the health field, in which a coefficient from 3 to 7 is usually applied, according to the countries and the health service coverage of the latter. This is used to estimate the number of women that underwent an abortion and did not need or have access to hospital treatment (Singh and Deirdre, 1990; 1991; and 1994). A variation of the complications method is based on maternal mortality statistics, which counts the number of deaths due to induced abortions (Rossier, 2003).

These different methods have a series of limitations that have consequences for the indicators obtained. The reliability of the data has been questioned, since very few women declare their abortions, and because health systems do not record all cases, particularly when interruptions of pregnancy are illegal. Under-registration is compounded by the problem of lack of representativeness of the data obtained and the extrapolation of results for the population as a whole, when surveys only refer to set groups of the latter. It is also difficult to locate the population of reference to calculate the indices at a general level, for example, to measure the mortality rates due to abortions, when only the deaths that have occurred in hospitals are counted. The estimates of these studies in certain countries lead to low estimates of the incidence of abortion, among other things, because women treated in hospital only account for a small portion of those that interrupt their pregnancy or because they are only the ones that often have serious complications. Moreover, these estimates usually represent women from low socio-economic strata, since women that are better off are able to obtain an abortion in suitable health conditions and therefore run far less of a risk of later requiring hospitalization. Likewise, in many cases these interventions are not declared as abortions (Strickler, 2001; Llovet, 1998).

Other Methods of Estimation

The residual method, based on Bongaarts model, is used to estimate the weight of the various determinants of fertility, such as age at first union, contraceptive practice, post-partum infertility, sterilization and abortion (Johnston and Hill, 1996; Llovet, 1998). This method can obtain an estimate of abortion rates. However, the parameters to be considered are not always taken into account in surveys, particularly demographic and health surveys conducted in Latin America. They also have the limitations pointed out earlier in capturing data.

The expert method, which consists of asking a set number of local experts (doctors, nurses, reproductive health and family planning program directors) in a set geographical space about the number of women hospitalized for abortion complications, levels and trends in their practice, the abortion methods used and other aspects. The information obtained can be compiled to make estimates. This is the case, for example, of estimates conducted by certain international organizations (Henshaw et al., 1999; Singh and Ratnam, 1998). In Uruguay, Sanseviero (2003) used various sources to estimate the incidence of abortion: live birth certificates, hospital discharges from complications from abortion and data from clandestine clinics. In his study, he also used qualitative approaches to explore other aspects related to the experience of abortion, both on the part of women and health professionals and other social agents.

Surveys aimed at illegal abortion “providers”: very few surveys have been conducted on the personnel that perform abortions in the formal or informal health system. In some countries in the region, surveys have been conducted on health providers that include questions specifically related to abortion to obtain information on the status of women that resort to this recourse and some of their characteristics (Abernathy et al., 1994; Pick et al., 1999; López and Findling, 2003).

The diversity of sources and methods proposed to analyze and estimate the incidence of the practice of abortion illustrates the complexity of studies on the subject, as well as the difficulties involved in obtaining reliable data. The joint use of various sources of information and various methods in a single country may contribute both to the measurement of the practice of abortion and to having a more complete overview of its characteristics. Finally, it is worth recalling that using all the sources mentioned raises a central ethical question, particularly in countries where abortion is illegal: respecting the anonymity of those that provide information on the subject.

Surveys and Interviews with Other Objectives

As shown in the previous chapter, several direct surveys have also been conducted, aimed at different sectors of the population, such as health professionals, to determine their medical practices in the reproductive sphere in general and in relation to abortion in particular (Gogna et al., 2002; García and Becker, 2001; IPAS México, 2002; Núñes and Delph, 1995; Católicas por el Derecho a Decidir, 2004). In addition to including health professionals, the surveys may target other specific sectors of the population (such as university students, practitioners of a particular religion, such as Catholics, leaders and other social actors, and people in general) to explore the knowledge and opinions of the interviewees regarding certain aspects of abortion, particularly the legal status of this practice in each country (Ramos et al., 2001; Faúndes et al., 2004; García et al., 2003; Silva Acuña, 2005, Petracci, 2005; Rodríguez, 2005; Ramos, 2005; García, 2005; Yam, 2005).

Finally, there are the qualitative methodological approaches, implemented through individual, semi-structured or in-depth interviews. They can also be conducted through focus groups involving interviews with women and occasionally health service providers or other key informants. They are studies conducted with small samples of women, males or groups of persons with certain characteristics, for which it is essential to establish close, trusting relationships with the persons being interviewed. The duration of these interviews, both as regards the capturing of information and its analysis, is relatively longer. The information obtained through these studies provides in-depth knowledge of various aspects of specific experiences in the practice of abortion, primarily those related to attitudes, assessments and perceptions as well as conditions of access to and quality of services, the influence of various actors, etc. (Rivas and Amuchástegui, 1996; Ramos and Viladrich, 1994; Cardich and Carrasco, 1993; Sanseviero, 2003; Azize, 2005. See other references in Chapters 4 and 9). One important task that remains to be done is to systematize the results found in these types of studies on abortion.

Despite the diversity of methodologies and techniques used, it is difficult to obtain comparable, basic indicators on abortion levels: the joint use of different methods would enable these estimates to be validated. Although there is a tendency to regard survey methods registering a higher incidence of abortion as better, this must be taken with precaution and has yet to be proved.

^ Top of page