Print version ISSN 1415-790X
Rev. bras. epidemiol. vol.15 n.1 São Paulo Mar. 2012
Tássia Ferreira Santos; Solange Andreoni; Rebeca de Souza e Silva
Department of Preventive Medicine, School of Medicine, Universidade Federal de São Paulo UNIFESP, State of São Paulo, Brazil
In Brazil, abortion is among the leading causes of maternal mortality. Research has shown that abortion is practiced clandestinely by women of all social classes, but has unequal consequences depending on social inclusion, producing risks to poor women. Although the issue has been widely explored in the past 20 years, there is a lack of data about low-income women. Thus, the present study aims to estimate the prevalence of women with induced abortion. Women from a population-based household survey in low-income sectors of São Vicente, São Paulo were recruited. Women of childbearing age from 15 to 49 years were eligible. The evaluation of the prevalence ratios for women with induced abortion was performed by using generalized linear models, with Poisson log-link function and robust variance to approximate the binomial. The most frequent variables that influenced reporting of abortion were: "always accept this practice" (95% CI 2.98 - 11.02), followed by "not having a child born alive" (95% CI 1.35 - 19.78), having "two to five live births" (95% CI 1.42 - 14.40 ), "having 'six or more live births" (95% CI 1.35 - 19.78), "age at interview" (95% CI 1.01 - 1.07) and "income" < R$ 484.97' (95% CI 1.04 - 2.96). A widespread campaign about the practice of abortion, which can raise awareness among women in favor of the cause, especially among those in low-income strata is necessary to prevent unnecessary deaths.
Keywords: Induced abortion. Unsafe abortion. Epidemiology. Poisson Regression. Reproductive health. Poverty.
Every year, all over the world, 20 million women risk their lives by undergoing unsafe abortion procedures, 25% of whom suffer serious health complications, including approximately 66,500 women who die from induced abortion without adequate medical care.1 The legislation on abortion varies according to the country. Currently, only 26% of the world population lives in countries where abortion is prohibited by law, i.e. the great majority of countries have laws that allow abortion.
The World Health Organization published a document this year that emphasizes among its conclusions that unsafe abortion and the deaths caused by unsafe abortion complications continue to affect the lives of many women, especially in developing countries. Unsafe abortion is the cause of serious health complications and disability for millions of women each year and an important cause of maternal death. Despite the efforts made to meet the 5th Millennium Development Goal - to reduce by three-quarters the maternal mortality ratio between 1990 and 2015 - the percentage of maternal deaths from unsafe abortion remains at 13%. The number of unsafe abortions has increased with the rise in the number of women of childbearing age. This trend will continue unless there is an improvement in women's access to safe abortion and adequate contraceptive methods.
In the particular case of Brazil, induced abortion is considered to be a crime against life in the penal code, including severe punishments against women undergoing the procedure and the individuals who perform it. Only in cases of imminent risk of death and rape is the practice of abortion legally permitted.2
Consequently, unsafe abortion is the fourth cause of maternal mortality in Brazil3, totaling 11% of all deaths.4 The risk of death by unsafe abortion is three times higher among black women than white women5, probably due to low access to family planning policies aimed at unwanted pregnancy prevention5,7, but especially due to fewer resources to go to illegal abortion clinics, putting their lives in the hands of uncertified midwives or or even self inducing abortion.
At the same time, according to the 2006 Pesquisa Nacional de Demografia e Saúde (PNDS - National Demographic and Health Survey), nearly 18.2% of pregnancies were unwanted by Brazilian women. Higher prevalences of unwanted births were found among women who were poorer, black, older and not married/cohabitating, and who had a lower level of education and many children.6
In short, there are more unwanted pregnancies among poorer women. Although some women choose to stop their pregnancy at any cost, there are others who end up assuming the responsibility for the birth of their child at an inappropriate time in their lives.
Despite the fact that the issue of abortion has been widely discussed in the last 20 years, there are still few population data, particularly on low-income population. In this sense, the present study aimed to estimate and characterize the prevalence of women aged between 15 and 49 years with induced abortion living in a slum situated in the city of São Vicente, state of São Paulo.
A cross-sectional study was conducted, based on secondary data derived from a sub-population of women participating in the project entitled "Aborto provocado, fecundidade e contracepção: imbricações com a integralidade em saúde, relações de gênero e exclusão social - Cidade de São Paulo e na Favela México 70" (Induced Abortion, Fertility and Contraception: Associations with Health Comprehensiveness, Gender Relations and Social Exclusion - City of São Paulo and Favela Mexico 70 Community), in late 2008, approved by the research ethics committee of the Universidade Federal de São Paulo (UNIFESP) under number 0196/08.
The Mexico 70 slum is situated in the southwestern-most part of the island of São Vicente and it is characterized by an illegal settlement of the Navy lands in an area of approximately 30 hectares of flood-prone lowlands, located between the Barreiros and Mar Pequeno bridges in this city. This slum was divided into three areas at the moment of data collection: buildings constructed and administered by the Companhia de Desenvolvimento Habitacional e Urbano do Estado de São Paulo (CDHU - State of São Paulo Housing and Urban Development Company); settlements; and stilt houses present in areas without infrastructure, especially basic sanitation, water services and sewage system. This slum is one of the largest and poorest settlements situated in the Baixada Santista (the metropolitan area that includes the city of Santos), an area with environmental degradation and extreme poverty.
Random household sampling8 was performed in the original project, including 1,067 women aged between 15 and 60 years, of whom 860 were in the age group of interest. However, the present study emphasized the analysis of the 735 women who have had one or more pregnancies.
Household selection was performed in partnership with the CDHU, which estimated that a total of 6,000 individuals lived in its area of coverage in the Mexico 70 slum, in 2007. Considering that each household, including businesses and churches, was comprised of three people on average, a total of 2,200 households were estimated in the area. The sample included all residential households with even numbers, while those with odd numbers comprised the back-up sample. Of all selected places, researchers decided to interview individuals from every other household, thus totaling nearly 1,100 households visited. In the case of more than one woman in the age group of interest being present, the one whose birth date was closer was selected to respond to the questionnaire. Up to three visits were made before considering a sample loss.
A pre-coded questionnaire was developed and used to collect data. All interviewers were female university students, as it was easier for them to discuss this issue. They received training on the content, management and application of questionnaires, which included the following: explanation about study objectives; description of the sampling scheme and instructions to identify eligible respondents; detailed description of the questionnaire; instructions on interview techniques; and interview execution.
The instrument was tested by conducting a pilot study, which enabled researchers to observe how easy it was to understand the questions formulated, to assess the interviewers' ability to perform the task attributed to them, and to identify the instructions required, which should be included in the questionnaire application manual.
The database of the sub-sample of the present study was structured using Excel software. Statistical analyses were performed with the SPSS software for Windows, version 17.0. The data were described as numbers and percentages for qualitative variables and as mean and standard-deviation for quantitative variables.
The assessment of prevalence ratios of women with induced abortion was performed with generalized linear models, using Poisson distribution with log link function and robust variance to approximate the binomial9 according to each of the following characteristics: age, paid work, income, level of education, marital status, contraceptive use, number of live births (LB), and acceptance of abortion, among others. The "Acceptance of abortion" categories were as follows: women who do not accept it under any circumstances; those who accept it in specific cases, such as lack of support from the child's father, presence of many children, unfavorable economic conditions, any conditions that prevent mothers from caring for their children adequately, risk of life for the mother and/or child, rape; and those who accept it in all cases, aiming to identify what characteristics were more associated with induced abortion.
Multiple regression analysis of the prevalence ratio was performed according to the factors investigated, using a backward stepwise variable selection strategy and the grouping of levels of factors that are statistically closer to one another. A level of significance (α) of 5% was adopted in all analyses, i.e. results with a p-value equal to or lower than 5% (<0.05) were considered to be significant. The strength of association between each of the explanatory variables and the response variable was assessed by calculating the prevalence ratio, followed by the respective 95% confidence interval (95%CI).
Among the 860 women aged between 15 and 49 years who participated in this study, the mean age was 32.4 years with a SD of 8.65. Of these, 735 (85%) had already become pregnant and their mean age was 32.9 years with a SD of 8.30. Table 1 shows the prevalence of women who became pregnant according to socioeconomic characteristics.
The lowest prevalences of women with a history of pregnancy were observed in the 15 to 20 year age group (53.5%), with an income higher than one minimum wage (83.3%) and high level of education (74.5%), and who were single (55.5%), did not use effective contraceptive methods (74.2%) and were against the practice of abortion (81.6%).
Table 2 shows the distribution of women who became pregnant according to the presence and absence of induced abortion (IA). The age at interview was shown in this table in a continuous way and, subsequently, as age groups.
Among women who became pregnant, 6.9% had an induced abortion. The highest frequencies of IA were observed in women aged between 40 and 45 years (12.9%), with an income equal to or lower than one MW (9%) and a low level of education (up to the 5th grade of primary school) (11.4%) and who always accepted the practice of abortion (34.8%). With regard to the "number of live births" variable, there was a high risk among women without children and among those with six or more live births. "Paid work", "marital status" and "use of contraception" did not show statistical significance.
The results of the final regression analysis model for induced abortion are shown in Table 3.
The variable that had the greatest influence on the practice of abortion was "always accepts abortion" (PR=, p<0.0001, 95%CI 2.98-11.02), followed by "no live births" (PR=, p=0.0002, 95%CI 3.23-46.19), "from two to five live births" (PR-, p=0.0108, 95%CI 1.42-14.40), "six or more live births" (PR=, p=0.0162, 95%CI 1.35-19.78), "current age" (one year age increment in PR= p=0.0216, 95%CI 1.01-1.07) and "income" < R$ 484.97 (p=0.0345, 95%CI 1.04-2.96).
The present study reveals the socio-demographic characteristics of women with a history of pregnancy, emphasizing those who had an induced abortion, and finally the regression model that exposes the main variables that jointly explain the induced abortion process.
The age group with the highest prevalence of women who became pregnant was that from 30 to 35 years (91.4%), while more than half of the women in the 15 to 20 year age group already had a history of pregnancy (53.5%).
The results found in the Pesquisa Nacional de Demografia e Saúde da Criança e da Mulher (PNDS - National Child and Women Health and Demographic Survey)6 referring to the socio-demographic characteristics of women all over Brazil show that the age structure of Brazilian women of childbearing age can be considered as that of a relatively young population, where the less than 30 years age group represents approximately 50% of the entire population. When the mean age of the interviewed women is estimated according to the macro-region of residence, women living in the South, Southeast and Center-West regions had a mean age higher than 31 years, similarly to what was found in the present study, with the 25 to 30 year age group having the highest percentage of women who became pregnant, 22.4%.
With regard to mean income, the present study showed that it was equal to or lower than one MW among 88.7% of women who became pregnant and 9% among those who had an abortion. These results were well below those obtained by the PNDS6, where the Southeast region had a mean income of R$ 1,674.98 or per household.
In terms of level of education, women who became pregnant and those who had an abortion most frequently reported having completed up to the 5th grade of primary school. These results are in disagreement with those found by the PNDS6, which showed that 20% of Brazilian women of childbearing age stated that they had completed up to the 5th grade of primary school, whereas 50% had more than eight years of education, i.e. they had completed primary school or higher.
With regard to the number of pregnancies and IA, the practice of abortion was high among women who did not have live births and it increased among those who had more than two live births, emphasizing the assumption that the number of abortions is directly associated with the number of pregnancies. In Colombia, Zamudio et al.10 found that housewives had less experience with abortion and that the percentage of women who had abortions also increased with the number of pregnancies. Silva and Fusco11, in a study conducted in the Inajar de Souza slum, in the city of São Paulo, found that only 33.3% of women had their first IA during the first pregnancy, whereas 66.6% had induced abortion in subsequent pregnancies, which is in agreement with the results observed in the present study.
Women living in the Mexico 70 slum had a high fertility rate of 2.5 children per woman. This rate is similar to that found by Sorrentino12 in the state of São Paulo (2.5) and higher that that found by Yazaki13 (1.74). However, these results differ from those from the Fundação Seade (São Paulo State Data Analysis System Foundation)14, which shows that the fertility rate of Brazilian women has decreased throughout time, totaling 2.26 children per woman in 1995 and 2.16 children in 2000; and those from Cuba15, where there has been a great reduction since the 1970s, reaching a rate of 1.5 children per woman in 1992 and 1.55 in 2000, one of the lowest ones in Latin America. The most important factor for this decrease in fertility was the growing practice of abortion.
The prevalence of IA in the present study was 6.9%, with a higher frequency in the 40 to 45 year age group. This result was higher than that found in the city of São Paulo16 (4.5%), and lower than those found in the majority of household surveys conducted in this city (8.3% in the district of Vila Madalena17 and 13.6% in the Inajar de Souza slum11). In a household survey conducted in the city of Rio de Janeiro, Martins et. al.18 interviewed 1,784 women aged between 15 and 49 years living in low-income areas, where one third of them reported having had an abortion, of which 16.9% were induced. It could be assumed that the several prevalences found in different studies are due to the methodological peculiarities of each study and to the social and religious factors associated with the local dynamics and response biases caused by fear or prejudice of reporting the practice of abortion.
The final model reveals that the set of variables that explain the decision in favor of abortion are low income, number of LB, age at interview and acceptance of the practice. The following three categories of LB were included in the final model composition, showing an association between induced abortion and the number of live births: no live births (PR=12.22), between two and five live births (PR=4.52), and six or more live births (PR=5.18). Consequently, this reveals that the number of live births can be a good indicator to predict future fertility. If the number of children considered to be ideal had been asked, the need to improve or not the access to the contraceptive methods available could be assessed.
To have an income < 1MW (R$484.97 was included in the final model of this study as a factor that was significant (PR=1.76) when correlated to the practice of abortion. Silva and Fusco11 found a significant association between IA and current per capita income (PCI), thus corroborating with the finding from the present study. The same authors also worked with -"spontaneous abortion" and "no abortion" groups and observed that IA was still associated with a lower PCI, when compared to these two groups.
When study participants were asked whether they accepted the practice of abortion, the most frequent response among both women with a history of pregnancy and those who had had an abortion was that they accepted this practice. This is in accordance with what was found by the PNDS6, where 66.6% of women who had been pregnant reported that they accepted the practice of abortion.
One of the limitations of the present study, apart from those inherent in cross-sectional studies, was that researchers opted for the presence of an interviewer due to participants' low level of education and possible difficulties in understanding and writing their responses, although this resulted in fewer women reporting the practice of abortion.21,22
The fact that the highest percentage of women with induced abortion was found among older participants could be justified, as these had been more exposed to the occurrence of an unwanted pregnancy in general. On the other hand, the positive association between the presence of induced abortion and the acceptance of this practice is more difficult to be interpreted, as women could have had an abortion because they were in favor of this practice or because they began to accept it after having an induced abortion. The cross-sectional design of the study did not enable researchers to find out what occurred first.
However, having no children was considered as a factor that promotes induced abortion and this was very significant, especially because it indicated that poor women aim at low fertility, like wealthier women.13 Thus, the risk of having an abortion increases by 12.22 times when women have "no children", when compared to those with "one child". Additionally, when considering the fact that, among poor women, the poorest ones are those most frequently having induced abortions, a clear public health deficit stands out in terms of meeting the contraceptive needs of this population, precisely the ones most affected by the morbi-mortality caused by this practice. It is essential for this field of knowledge to change such situation.
Meanwhile, the legalization of abortion could certainly promote equity, in the sense that not only those who can afford this type of service, but rather all women could be cared for by doctors when facing the extreme situation of deciding for an induced abortion. The low prevalence of women who have had an abortion clearly indicates that this is not routinely performed. Based on the data from countries where induced abortion has been legalized, it is believed that the recurrence of this practice will tend to decrease as effective unwanted pregnancy prevention programs are implemented.
1. World Health Organization. Unsafe abortion: global and regional estimates of the incidence of unsafe abortion and associated mortality. Geneve; 2003. [ Links ]
2. World Health Organization. Unsafe abortion: global and regional estimates of the incidence of unsafe abortion and associated mortality. Geneve; 2011. [ Links ]
3. BRASIL. Código penal. 4. ed. São Paulo: Editora Revista dos Tribunais; 2002. [ Links ]
4. Laurenti R, Mello Jorge MHP, Gotlieb SLD. A mortalidade materna nas capitais brasileiras: algumas características e estimativas de um fator de ajuste. Rev Bras Epidemiol 2004; 7(4): 449-60. [ Links ]
5. Ministério da Saúde. Norma Técnica para Atenção Humanizada ao Abortamento; 2005. [ Links ]
6. Instituto de Medicina Social/UERJ e Ipas Brasil. ENSP/Fiocruz 2005. Rev Radis 2008; 66. [ Links ]
7. Comissão Parlamentar de Inquérito sobre a Situação da Mortalidade Materna no Brasil, Relatório 2001. [ Links ]
8. Ministério da Saúde, CEBRAP. Pesquisa Nacional de Demografia e Saúde da Criança e da Mulher 2006. Brasília; 2009. Disponível em http://bvsms.saude.gov.br/bvs/publicacoes/pnds_crianca_mulher.pdf. [Acessado em 25 de dezembro de 2010] [ Links ]
9. RELEASE Bahia: Dossiê sobre a Realidade do Aborto Inseguro na Bahia: a Ilegalidade da Prática e os seus Efeitos na Saúde das Mulheres em Salvador e Feira de Santana, 2008. Disponível em http://www.ipas.org.br/arquivos/RELEASE_Dossie_BA.pdf. [Acessado em 11 de fevereiro de 2011] [ Links ]
10. Cardoso MRA. Comparação entre três métodos de amostragem que visam à estimação da cobertura vacinal [dissertação de mestrado]. São Paulo: Universidade de São Paulo; 1990. [ Links ]
11. Barros AJ, Hirakata VN. Alternatives for logistic regression in cross-sectional studies: an empirical comparison of models that directly estimate the prevalence ratio. BMC Medical Research Methodology 2003; 3: 21. [ Links ]
12. Zamudio Cárdenas L, Rubiano Blanco N, Wartenberg L, Viveros M, Salcedo Fidalgo H. El aborto inducido en Colombia. Cuadernos del CIDS. Bogotá: Centro de Investigaciones sobre Dinámica Social (CIDS); 1999. [ Links ]
13. Fusco CLB, Andreoni S, Silva RS. Epidemiologia do aborto inseguro em uma população em situação de pobreza Favela Inajar de Souza, São Paulo. Rev Bras Epidemiol 2008 mar; 11(1): 78-88. [ Links ]
14. Sorrentino SR. "Aborto inseguro". Rede nacional feminista de saúde e diretos reprodutivos. Saúde da mulher e direitos reprodutivos: dossiês. São Paulo: Rede Nacional Feminista de Saúde e Direitos Reprodutivos 2001; 41-72. [ Links ]
15. Yazaki LM. Análise da Fecundidade no Estado de São Paulo. São Paulo em Perspectiva 2008; 22(1): 48-65. [ Links ]
16. Fundação Sistema Estadual de Análise de dados, SEADE. Secretaria de Economia e Planejamento. Perfil Municipal de São Paulo, 2009. Disponível em: http://www.seade.gov.br/produtos/perfil/perfil.php [Acessado em 15 de Novembro de 2010] [ Links ].
17. Cabezas-Garcia E et al. Perfil sociodemográfico del aborto inducido. Salud Pública de México 1998; 40: (30). [ Links ]
18. Silva RS. Padrões de aborto provocado na Grande São Paulo, Brasil. Rev Saúde Pública 1998; 32(1): 7-17. [ Links ]
19. Silva RS. Aborto provocado: sua incidência e características: um estudo com mulheres em idade fértil (15 a 49 anos), residentes no subdistrito de Vila Madalena [tese de doutorado]. São Paulo: Universidade de São Paulo; 1992. [ Links ]
20. Martins IR, Costa SH, Freitas SRS, Pinto CS. Aborto induzido em mulheres de baixa renda: dimensão de um problema. Cad Saúde Pública 1991; 7: 251-66. [ Links ]
21. Alan Guttmacher Institute. Sharing responsibilities: women, society and abortion worldwide. New York: The Alan Guttmacher Institute; 1999. [ Links ]
22. Grimes DA, Benson J, Singh S et al. Unsafe abortion: the preventable pandemic. Lancet 2006; 368: 1908-19. [ Links ]
research project was approved by the research ethics committee of the Universidade
Federal de São Paulo UNIFESP, under number 0196/08.
Tássia Ferreira Santos
Rua Afonso Celso, 718 Apto, 23, Vila Mariana
São Paulo, SP CEP 04.119-060
Final version: 13/10/11
This research project was approved by the research ethics committee of the Universidade Federal de São Paulo UNIFESP, under number 0196/08.