Abstract

Introduction

Voluntary termination of pregnancy is considered a crime in Brazil, except in cases of rape, risk to the mother’s life, and fetal anencephaly. According to data from the Brazilian National Abortion Survey (PNA) in 2016 ¹1. Diniz D, Medeiros M, Madeiro A. Pesquisa Nacional de Aborto 2016. Ciênc Saúde Colet 2017; 22:653-60., one out of five Brazilian women had undergone an abortion by age 40. Among women 18 to 39 years of age, the reported abortion rate was 13%, without a significant difference in relation to the 15% rate found by the survey in 2010 ²2. Diniz D, Medeiros M. Aborto no Brasil: uma pesquisa domiciliar com técnica de urna. Ciênc Saúde Colet 2010; 10 Suppl 1:959-66..

Few studies in Brazil have analyzed the issue of induced abortion in women living with HIV/AIDS. In 2003-2004, a study in 13 Brazilian municipalities compared women 18 years and older living with HIV/AIDS (WLHA) and women not living with HIV/AIDS (WNLHA) and found a lifetime prevalence of induced abortion of 17.5% in WLHA and 10.4% in WNLHA (p < 0.001) ³3. Barbosa RM, Pinho AA, Santos NS, Filipe E, Villela W, Aidar T. Aborto induzido entre mulheres em idade reprodutiva vivendo e não vivendo com HIV/aids no Brasil. Ciênc Saúde Colet 2009; 14:1085-99.. Studies with a similar methodology, comparing WLHA and WNLHA, conducted subsequently in Porto Alegre ⁴4. Pilecco FB, Teixeira LB, Vigo A, Dewey ME, Knauth DR. Lifetime induced abortion: a comparison between women living and not living with HIV. PLoS One 2014; 9:e95570. in 2011 and in São Paulo ⁵5. Pinho AA, Cabral CS, Barbosa RM. Differences and similarities in women living and not living with HIV: contributions by the GENIH study to sexual and reproductive healthcare. Cad Saúde Pública 2017; 33:e00057916. in 2013-2014, found estimates for induced abortion of 13% and 14.1% in WLHA and 4.9% and 3.2% in WNLHA, respectively, and this difference between the groups was significant.

Time series with estimates of abortion by direct methods are not available in Brazil, since all the studies that estimated prevalence of abortion, whether in the general population ¹1. Diniz D, Medeiros M, Madeiro A. Pesquisa Nacional de Aborto 2016. Ciênc Saúde Colet 2017; 22:653-60.^,22. Diniz D, Medeiros M. Aborto no Brasil: uma pesquisa domiciliar com técnica de urna. Ciênc Saúde Colet 2010; 10 Suppl 1:959-66.^,66. Cecatti JG, Guerra GVQL, Sousa MH, Menezes GMS. Aborto no Brasil: um enfoque demográfico. Rev Bras Ginecol Obstet 2010; 32:105-11.^,77. Camargo RS, Santana DS, Cecatti JG, Pacagnella RC, Tedesco RP, Melo Jr. EF, et al. Severe maternal morbidity and factors associated with the occurrence of abortion in Brazil. Int J Gynaecol Obstet 2011; 112:88-92., in WLHA and WNLHA ³3. Barbosa RM, Pinho AA, Santos NS, Filipe E, Villela W, Aidar T. Aborto induzido entre mulheres em idade reprodutiva vivendo e não vivendo com HIV/aids no Brasil. Ciênc Saúde Colet 2009; 14:1085-99.^,44. Pilecco FB, Teixeira LB, Vigo A, Dewey ME, Knauth DR. Lifetime induced abortion: a comparison between women living and not living with HIV. PLoS One 2014; 9:e95570.^,55. Pinho AA, Cabral CS, Barbosa RM. Differences and similarities in women living and not living with HIV: contributions by the GENIH study to sexual and reproductive healthcare. Cad Saúde Pública 2017; 33:e00057916., used a cross-sectional design, generating point estimates for the study period. Studies that assessed hospitalization for complications from abortion using the Brazilian Hospital Admissions System of the Brazilian Unified National Health System (SIH/SUS), furnishing indirect estimates, have shown a reduction in abortion rates in Brazil from 1995 to 2013 ⁸8. Martins-Melo FR, Lima MS, Alencar CH, Ramos Jr AN, Carvalho FHC, Machado MMT, et al. Tendência temporal e distribuição espacial do aborto inseguro no Brasil, 1996-2012. Rev Saúde Pública 2014; 48:508-20.^,99. Monteiro MFG, Adesse L, Drezett J. Atualização das estimativas da magnitude do aborto induzido, taxas por mil mulheres e razões por 100 nascimentos vivos do aborto induzido por faixa etária e grandes regiões. Brasil, 1995 a 2013. Reprod Clim 2015; 30:11-8., but the results are subject to the method’s limitations ¹⁰10. Menezes GMS, Aquino EML, Fonseca SC, Domingues RMSM. Aborto e saúde no Brasil: desafios para a pesquisa sobre o tema em um contexto de ilegalidade. Cad Saúde Pública 2020; 36 Suppl 1:e00197918..

The National Institute of Infectious Diseases Evandro Chagas under the Oswaldo Cruz Foundation (INI/Fiocruz) is a referral unit for persons with HIV/AIDS. A cohort of women was launched in 1996 with the aim of providing gynecological care for WLHA, and the inclusion of women was concluded on December 31, 2016. A previous study ¹¹11. Friedman RK, Bastos FI, Leite IC, Veloso VG, Moreira RI, Cardoso SW, et al. Pregnancy rates and predictors in women with HIV/AIDS in Rio de Janeiro, Southeastern Brazil. Rev Saúde Pública 2011; 45:373-81. that assessed the incidence of pregnancies, deliveries, and abortions in the follow-up of women included from June 1996 to February 2003 found that 41.8% had a history of induced abortion recorded at the time of inclusion in the cohort.

The current study aimed to verify the prevalence of induced abortion and associated factors at inclusion in the INI/Fiocruz cohort of women from 1996 to 2016, with emphasis on the identification of changes in the prevalence of induced abortion over time.

Methods

This is an analysis of data from the appointment at inclusion in the INI/Fiocruz cohort of women. The INI/Fiocruz cohort is an open cohort whose inclusion criteria were: female sex at birth, age 18 years or older, and confirmed HIV infection. INI/Fiocruz is located in the Manguinhos neighborhood in the city of Rio de Janeiro, Brazil, and does not have a specific coverage area, rather receiving patients that come to the service spontaneously or are referred by other health services.

In the entry visit to the cohort, all the women were interviewed by the attending physician during individual appointments, using a standardized questionnaire on socioeconomic status, sexual and reproductive history, behavioral data, and data on HIV infection. After the cohort entry appointment, the women were followed up with biannual or annual visits, when clinical examination and laboratory tests were performed and a follow-up form was completed. All the participating physicians were trained for standardization of the data collection at both the initial and follow-up visits.

The current study only used data obtained in the cohort entry interviews. Inconsistencies or errors in completion of the questionnaire were corrected by means of review and consultation of the hospital charts. The analysis included the following variables: age, skin color, schooling in years, work, income in times the minimum wage, marital status (living versus not living with spouse/partner), gynecological and obstetric history (age at sexual initiation, number of pregnancies, births, spontaneous and induced abortions, age at first pregnancy), history of diagnosis of sexually transmissible diseases (STDs), current use of contraceptive methods, smoking (current or previous), lifetime use of illicit drugs (marijuana, inhaled cocaine, injecting cocaine, glue), lifetime sexual partners, and report of physical domestic violence and sexual violence. In 2006 the questionnaire with gynecological and obstetric information was modified to include information on the timing of the abortion (before or after diagnosis of HIV infection). This variable applied to women included in the cohort since 2006. However, for all the other analyses, the variable “induced abortion” refers to one or more lifetime induced abortions, independently of the timing in relation to HIV diagnosis.

The descriptive analysis presents the characteristics of the women included during the study period, with quantitative variables described as medians and interquartile ranges (IQR) and categorical variables as proportions. To verify whether the women’s characteristics changed over time (considering 1996-2000, 2001-2005, 2006-2010, and 2011-2016), the chi-square test for trend ¹²12. Dalgaard P. Introductory statistics with R. 2nd Ed. New York: Springer; 2008. was used for categorical variables, with statistical significance set at 0.05.

Factors associated with induced abortion were identified by simple and multiple logistic regression for the total cohort of women and for the group of women with previous pregnancy. Both analyses used as the outcome variable “previous induced abortion”, defined as the report of having undergone one or more induced abortions any time in life. The simple analysis included variables related to the woman’s life course: age, skin color (white, brown, and black, and the only indigenous woman excluded from the analysis), schooling, age at sexual initiation (< 15 years, 15-19 years, 20 or older), teenage pregnancy (under 20 years of age) ¹³13. Departamento de Ações Programáticas e Estratégicas, Secretaria de Atenção à Saúde, Ministério da Saúde. Proteger e cuidar da saúde de adolescentes na atenção básica. Brasília: Ministério da Saúde; 2017., total number of lifetime sexual partners (categorized as < 5 or ≥ 5), lifetime use of any illicit drug (regardless of duration and frequency), lifetime report of some episode of violence (physical or sexual), and time of inclusion in the cohort (1996-2000, 2001-2005, 2006-2010, 2011-2016). The variable “previous diagnosis of STD” was not included in this analysis due to the high proportion of unknown answers. Variables related to the timing of the pregnancy that resulted in the abortion, which may have changed in relation to the timing of the interview, such as income, marital status, contraceptive use, and number of pregnancies were also not included in this analysis. Variables with p-value < 0.20 in the simple logistic regression were included in the multiple logistic regression. All variables with level of significance < 0.05 were maintained in the final model, and the results were expressed as adjusted odds ratios (OR) with their 95% confidence intervals (95%CI). The final model’s fit was verified with the Hosmer-Lemeshow test. All the analyses used the R statistical package, version 3.5.1 (http://www.r-project.org).

The INI women’s cohort study was approved the Institutional Review Board of INI/Fiocruz. All the participating women signed a free and informed consent form upon inclusion in the cohort.

Results

From January 1996 to December 2016, 1,383 WLHA were included in the INI/Fiocruz gynecology cohort, and there were no refusals. More than one-third of the inclusions (32%) occurred in 2006-2010. Median age at inclusion in the cohort was 36 years (IQR 29, 43), and one third of the women belonged to the 30-39-year bracket. The majority of the women were non-white (61.8%), with low schooling (54.2% with 8 years or less of formal education), low family income (63.5% with income of up to twice the minimum wage), and were unemployed (52.3%) at entry in the cohort. More than half of the women were not living with a spouse or partner (56.1%), 30.9% had a current partner who was HIV-positive, and one fourth reported having a partner who had died of HIV/AIDS. Current or past smoking was reported by 42.4% of the women, and 16.6% reported lifetime use of any illicit drug, with marijuana and inhaled cocaine as the frequent (11% each). History of physical violence and sexual violence were reported by 31.3% and 16% of the women, respectively (Table 1). Approximately 90% of the women reported having been infected with HIV by the sexual route.

Median age at sexual initiation was 17 years (IQR 15, 18), with 18.7% of the women reporting sexual initiation before 15 years of age. Half of the women reported 5 or more lifetime sexual partners, and 41% reported a prior diagnosis of some STD (Table 2), the most frequent diagnoses being herpes or vesicles (20%), HPV/condyloma (15.8%), and syphilis (12.1%).

More than 90% of the women reported one or more previous pregnancies (median of three pregnancies, IQR 1, 4), and 51.7% had their first pregnancy during adolescence (under 20 years of age). Of the women with previous gestations, 93% had given birth at least once and 52% reported at least one spontaneous or induced abortion. Of the women with previous spontaneous or induced abortion, 44.4% reported at least one spontaneous abortion and 64.5% reported at least one induced abortion. Of all the women (with and without previous pregnancy), 30.4% reported an induced abortion any time in life (Table 2), and 39.9% reported more than one induced abortion. In the period from 2006 to 2016, 96% of the induced abortions occurred before the woman was diagnosed with HIV.

At the cohort entry appointment, one fourth of the women under 50 years of age that did not report natural or surgical menopause were not using any contraceptive method, 23.9% had undergone tubal ligation, and more than 40% used condoms associated with natural or hormonal methods (Table 2).

Table 3 shows some of the women’s characteristics at inclusion in the cohort in the four periods. There was a significant increase in the proportion of women 50 years or older and a significant decrease in the 20-29-year bracket. There was also an increase in the proportion of women with 9 to 11 years of schooling, while the proportion of white women decreased. The report of sexual violence decreased significantly during the period, while the report of physical domestic violence and illicit drug use remained unchanged. There was an increase in the proportion of women that reported sexual initiation before 15 years and from 15 to 19 years of age, in the proportion of women with 5 or more lifetime sexual partners, without previous pregnancies, and with a history of teenage pregnancy. The proportion of spontaneous abortions increased, while the proportion of induced abortions decreased significantly over the 20 years, from 41.7% in 1996-2000 to 22.5% in 2011-2016. There was also a significant increase in the use of condoms associated with hormonal methods and a decrease in the proportion of women with tubal ligation.

In the simple logistic regression (Table 4), the variables “year since inclusion in the cohort”, age, skin color, schooling, age at sexual initiation, lifetime number of sexual partners, teenage pregnancy, report of violence (physical or sexual), and drug use were significantly associated with the outcome “lifetime induced abortion”. In the multiple logistic regression (Table 5), age and schooling presented an upward gradient with induced abortion, with an OR of 1.03 (95%CI: 1.02-1.05) for age and an OR of 3.33 (95%CI: 2.01-5.56) in women with 12 or more years of schooling. More lifetime partners (OR = 2.10; 95%CI: 1.60-2.77), history of teenage pregnancy (OR = 1.78; 95%CI: 1.35-2.36), and lifetime illicit drug use (OR = 1.51; IC95%: 1.06-2.15) showed a significant association with more abortions. The two entry periods after 2005 showed an association with a lower proportion of induced abortions when compared to 1996-2000, even after adjusting for all the women’s characteristics. The simple and multiple logistic regression analyses only with women with previous pregnancies showed a similar pattern (Table 5), with both models presenting good fit (Hosmer-Lemeshow test with p-values of 0.404 and 0.388).

Discussion

The study’s results revealed high lifetime prevalence of induced abortion in WLHA, namely 30.4%, in the period from 1996 to 2016 at the time of entry in the gynecology cohort in Rio de Janeiro. This rate is far higher than the rates estimated in the Brazilian National Demographic and Health Survey (PNDS) in 1996 ⁶6. Cecatti JG, Guerra GVQL, Sousa MH, Menezes GMS. Aborto no Brasil: um enfoque demográfico. Rev Bras Ginecol Obstet 2010; 32:105-11. and 2006 ⁷7. Camargo RS, Santana DS, Cecatti JG, Pacagnella RC, Tedesco RP, Melo Jr. EF, et al. Severe maternal morbidity and factors associated with the occurrence of abortion in Brazil. Int J Gynaecol Obstet 2011; 112:88-92. (2.3% and 2.4%), using direct interviews, and by the PNA in 2010 ²2. Diniz D, Medeiros M. Aborto no Brasil: uma pesquisa domiciliar com técnica de urna. Ciênc Saúde Colet 2010; 10 Suppl 1:959-66. and 2016 ¹1. Diniz D, Medeiros M, Madeiro A. Pesquisa Nacional de Aborto 2016. Ciênc Saúde Colet 2017; 22:653-60. (15% and 13%), using the ballot-box method. When comparing our results to specific studies with WLHA, our rates were also higher, both in comparison to studies using the ballot-box method (17.5%) ³3. Barbosa RM, Pinho AA, Santos NS, Filipe E, Villela W, Aidar T. Aborto induzido entre mulheres em idade reprodutiva vivendo e não vivendo com HIV/aids no Brasil. Ciênc Saúde Colet 2009; 14:1085-99. and those using direct interviews (13% ⁴4. Pilecco FB, Teixeira LB, Vigo A, Dewey ME, Knauth DR. Lifetime induced abortion: a comparison between women living and not living with HIV. PLoS One 2014; 9:e95570., 14.1% ⁵5. Pinho AA, Cabral CS, Barbosa RM. Differences and similarities in women living and not living with HIV: contributions by the GENIH study to sexual and reproductive healthcare. Cad Saúde Pública 2017; 33:e00057916.), with the caveat that the latter only analyzed women with prior pregnancies. Previous studies ⁶6. Cecatti JG, Guerra GVQL, Sousa MH, Menezes GMS. Aborto no Brasil: um enfoque demográfico. Rev Bras Ginecol Obstet 2010; 32:105-11.^,1414. 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. had already identified a high prevalence of induced abortion in Rio de Janeiro.

As in other studies that assessed factors associated with lifetime abortion prevalence ³3. Barbosa RM, Pinho AA, Santos NS, Filipe E, Villela W, Aidar T. Aborto induzido entre mulheres em idade reprodutiva vivendo e não vivendo com HIV/aids no Brasil. Ciênc Saúde Colet 2009; 14:1085-99.^,66. Cecatti JG, Guerra GVQL, Sousa MH, Menezes GMS. Aborto no Brasil: um enfoque demográfico. Rev Bras Ginecol Obstet 2010; 32:105-11.^,1515. Souza MG, Fusco CLB, Andreoni SA, Silva RS. Prevalência e características sociodemográficas de mulheres com aborto provocado em uma amostra da população da Cidade de São Paulo, Brasil. Rev Bras Epidemiol 2014; 17:297-312.^,1616. Fusco CLB, Silva RS, Andreoni S. Unsafe abortion: social determinants and health inequities in a vulnerable population in São Paulo, Brazil. Cad Saúde Pública 2012; 28:709-19.^,1717. Machado CJ, Lobato ACL, Melo VH, Guimarães MDC. Perdas fetais espontâneas e voluntárias no Brasil em 1999-2000: um estudo de fatores associados. Rev Bras Epidemiol 2013; 16:18-29.^,1818. Pilecco FB, Knauth DR, Vigo A. Aborto e coerção sexual: o contexto de vulnerabilidade entre mulheres jovens. Cad Saúde Pública 2011; 27:427-39. or in pregnancies that ended in abortion ⁴4. Pilecco FB, Teixeira LB, Vigo A, Dewey ME, Knauth DR. Lifetime induced abortion: a comparison between women living and not living with HIV. PLoS One 2014; 9:e95570.^,1919. van Benthem BH, de Vincenzi I, Delmas MC, Larsen C, van den Hoek A, Prins M. Pregnancies before and after HIV diagnosis in a European cohort of HIV-infected women: European Study on the Natural History of HIV infection in women. AIDS 2000; 14:2171-8., we found that increasing age ³3. Barbosa RM, Pinho AA, Santos NS, Filipe E, Villela W, Aidar T. Aborto induzido entre mulheres em idade reprodutiva vivendo e não vivendo com HIV/aids no Brasil. Ciênc Saúde Colet 2009; 14:1085-99.^,44. Pilecco FB, Teixeira LB, Vigo A, Dewey ME, Knauth DR. Lifetime induced abortion: a comparison between women living and not living with HIV. PLoS One 2014; 9:e95570.^,66. Cecatti JG, Guerra GVQL, Sousa MH, Menezes GMS. Aborto no Brasil: um enfoque demográfico. Rev Bras Ginecol Obstet 2010; 32:105-11.^,1515. Souza MG, Fusco CLB, Andreoni SA, Silva RS. Prevalência e características sociodemográficas de mulheres com aborto provocado em uma amostra da população da Cidade de São Paulo, Brasil. Rev Bras Epidemiol 2014; 17:297-312., higher lifetime number of sexual partners ³3. Barbosa RM, Pinho AA, Santos NS, Filipe E, Villela W, Aidar T. Aborto induzido entre mulheres em idade reprodutiva vivendo e não vivendo com HIV/aids no Brasil. Ciênc Saúde Colet 2009; 14:1085-99.^,44. Pilecco FB, Teixeira LB, Vigo A, Dewey ME, Knauth DR. Lifetime induced abortion: a comparison between women living and not living with HIV. PLoS One 2014; 9:e95570.^,1616. Fusco CLB, Silva RS, Andreoni S. Unsafe abortion: social determinants and health inequities in a vulnerable population in São Paulo, Brazil. Cad Saúde Pública 2012; 28:709-19.^,1717. Machado CJ, Lobato ACL, Melo VH, Guimarães MDC. Perdas fetais espontâneas e voluntárias no Brasil em 1999-2000: um estudo de fatores associados. Rev Bras Epidemiol 2013; 16:18-29.^,1818. Pilecco FB, Knauth DR, Vigo A. Aborto e coerção sexual: o contexto de vulnerabilidade entre mulheres jovens. Cad Saúde Pública 2011; 27:427-39.^,1919. van Benthem BH, de Vincenzi I, Delmas MC, Larsen C, van den Hoek A, Prins M. Pregnancies before and after HIV diagnosis in a European cohort of HIV-infected women: European Study on the Natural History of HIV infection in women. AIDS 2000; 14:2171-8. and lifetime drug use ³3. Barbosa RM, Pinho AA, Santos NS, Filipe E, Villela W, Aidar T. Aborto induzido entre mulheres em idade reprodutiva vivendo e não vivendo com HIV/aids no Brasil. Ciênc Saúde Colet 2009; 14:1085-99. were associated with induced abortion. Age at sexual initiation was not associated with induced abortion as in other studies ³3. Barbosa RM, Pinho AA, Santos NS, Filipe E, Villela W, Aidar T. Aborto induzido entre mulheres em idade reprodutiva vivendo e não vivendo com HIV/aids no Brasil. Ciênc Saúde Colet 2009; 14:1085-99.^,1616. Fusco CLB, Silva RS, Andreoni S. Unsafe abortion: social determinants and health inequities in a vulnerable population in São Paulo, Brazil. Cad Saúde Pública 2012; 28:709-19.^,1717. Machado CJ, Lobato ACL, Melo VH, Guimarães MDC. Perdas fetais espontâneas e voluntárias no Brasil em 1999-2000: um estudo de fatores associados. Rev Bras Epidemiol 2013; 16:18-29.. However, teenage pregnancy showed a positive association, possibly reflecting early initiation of unprotected sexual activity.

We found a gradient between schooling and induced abortion, an association that was also observed by Pilleco et al. ⁴4. Pilecco FB, Teixeira LB, Vigo A, Dewey ME, Knauth DR. Lifetime induced abortion: a comparison between women living and not living with HIV. PLoS One 2014; 9:e95570. in WLHA in Porto Alegre, Brazil. A multicenter study in three Brazilian cities with women 18-24 years of age ²⁰20. Menezes GM, Aquino EM, Silva DO. Induced abortion during youth: social inequalities in the outcome of the first pregnancy. Cad Saúde Pública 2006; 22:1431-46. showed that unintended pregnancy competes with the personal projects of young people seeking more education and participation in the work market. For women with less schooling, keeping a pregnancy (although unplanned) may be associated with less access to safe methods for termination of the pregnancy, but may also be a strategy for social recognition through motherhood or even a possible life project in a society with limited possibilities ²⁰20. Menezes GM, Aquino EM, Silva DO. Induced abortion during youth: social inequalities in the outcome of the first pregnancy. Cad Saúde Pública 2006; 22:1431-46.^,2121. Singh S, Darroch JE, Frost JJ. Socioeconomic disadvantage and adolescent women's sexual and reproductive behavior: the case of five developed countries. Fam Plann Perspect 2003; 33:251-8.. Thus, our results may reflect reverse causality, that is, women with more schooling at the time of the interview opted for termination of an unplanned pregnancy in the past.

Unlike other Brazilian studies that have reported higher rates of induced abortion in black women ¹1. Diniz D, Medeiros M, Madeiro A. Pesquisa Nacional de Aborto 2016. Ciênc Saúde Colet 2017; 22:653-60.^,33. Barbosa RM, Pinho AA, Santos NS, Filipe E, Villela W, Aidar T. Aborto induzido entre mulheres em idade reprodutiva vivendo e não vivendo com HIV/aids no Brasil. Ciênc Saúde Colet 2009; 14:1085-99.^,66. Cecatti JG, Guerra GVQL, Sousa MH, Menezes GMS. Aborto no Brasil: um enfoque demográfico. Rev Bras Ginecol Obstet 2010; 32:105-11.^,1616. Fusco CLB, Silva RS, Andreoni S. Unsafe abortion: social determinants and health inequities in a vulnerable population in São Paulo, Brazil. Cad Saúde Pública 2012; 28:709-19.^,1717. Machado CJ, Lobato ACL, Melo VH, Guimarães MDC. Perdas fetais espontâneas e voluntárias no Brasil em 1999-2000: um estudo de fatores associados. Rev Bras Epidemiol 2013; 16:18-29., our study did not find an association between induced abortion and self-reported skin color. This finding may result from the association between schooling and skin color, with white women reporting more schooling (14.6% of white women with 12 years of schooling or more, versus 7.7% of black and brown women, p < 0.001), which may have attenuated the association between induced abortion and non-white skin color, as reported in other studies.

An association between self-reported STD and induced abortion has also been reported in other Brazilian studies ³3. Barbosa RM, Pinho AA, Santos NS, Filipe E, Villela W, Aidar T. Aborto induzido entre mulheres em idade reprodutiva vivendo e não vivendo com HIV/aids no Brasil. Ciênc Saúde Colet 2009; 14:1085-99.^,1717. Machado CJ, Lobato ACL, Melo VH, Guimarães MDC. Perdas fetais espontâneas e voluntárias no Brasil em 1999-2000: um estudo de fatores associados. Rev Bras Epidemiol 2013; 16:18-29.. In our study, a high proportion of women were unable to report on previous diagnosis of STDs, which prevented the variable’s inclusion in the adjusted analysis. Among the women who were able to provide this information, the self-reported STD rate was high, especially for syphilis, herpes, and HPV, revealing this population’s vulnerability to STDs.

Prevalence of physical domestic violence and sexual violence was high, corroborating findings by Pinho et al. in São Paulo ⁵5. Pinho AA, Cabral CS, Barbosa RM. Differences and similarities in women living and not living with HIV: contributions by the GENIH study to sexual and reproductive healthcare. Cad Saúde Pública 2017; 33:e00057916.. However, although history of violence was associated with induced abortion in the simple regression, this association lost significance after adjusting for the other variables, contrary to findings from international studies demonstrating an association between domestic violence and induced abortion ²²22. Kaye DK, Mirembe FM, Bantebya G, Johansson A, Ekstrom AM. Domestic violence as risk factor for unwanted pregnancy and induced abortion in Mulago Hospital, Kampala, Uganda. Trop Med Int Health 2006; 11:90-101.^,2323. Stöckl H, Filippi V, Watts C, Mbwamboo JK. Induced abortion, pregnancy loss and intimate partner violence in Tanzania: a population based study. BMC Pregnancy Childbirth 2012; 12:12.^,2424. Palitto CC, García-Moreno C, Jansen HA, Heise L, Ellsberg M, Watts C. Intimate partner violence, abortion, and unintended pregnancy: results from the WHO Multi-country Study on Women's Health and Domestic Violence. Int J Gynaecol Obstet 2013; 120:3-9.. In these studies, domestic violence is defined as physical and/or sexual violence by an intimate partner and can include psychological violence in the definition ²²22. Kaye DK, Mirembe FM, Bantebya G, Johansson A, Ekstrom AM. Domestic violence as risk factor for unwanted pregnancy and induced abortion in Mulago Hospital, Kampala, Uganda. Trop Med Int Health 2006; 11:90-101.. In our study, the exclusion of psychological violence, the lack of specification of the aggressor, and the use of different instruments for measuring domestic violence limit the results’ comparison. In the study of WLHA by Barbosa et al. ³3. Barbosa RM, Pinho AA, Santos NS, Filipe E, Villela W, Aidar T. Aborto induzido entre mulheres em idade reprodutiva vivendo e não vivendo com HIV/aids no Brasil. Ciênc Saúde Colet 2009; 14:1085-99., the association between sexual violence and induced abortion lost significance after adjusting for lifetime casual partners and self-reported STDs, both heavily and jointly associated with history of sexual violence and lifetime induced abortion. In our study, three factors associated with induced abortion were also associated with violence, which may explain the loss of association in the adjusted analysis. Women with more sexual partners and report of drug use showed higher prevalence of sexual violence (5 or more partners 20.1% versus < 5 partners 11.4%, p < 0.001; drug use 26% versus no drug use 14.3%, p < 0.001) and physical domestic violence (5 or more partners 36.7% versus < 5 partners 23.7%, p < 0.001; drug use 52.5% versus no drug use 28%, p < 0.001), while teenage pregnancy was associated with physical domestic violence (39.2% versus 27.4%, p < 0.001).

A factor that was not investigated in previous studies and is thus unprecedented is the change in abortion rates over time. Our study found a significant reduction in induced abortion in the most recent periods (2006-2010 and 2011-2016) when compared to the late 1990s and early 2000s, even after adjusting for all the women’s characteristics that also changed in these same 20 years. The last period showed an increase in the proportion of spontaneous abortion when compared to induced abortions, which may have resulted from a measurement error if women in the more recent periods omitted induced abortions, leading to a false reduction in them. However, there was a drop in the total number (spontaneous plus induced abortions) throughout the period, and even if all the fetal losses reported as spontaneous abortions were actually induced, which is unlikely, there still would have been a decrease in the number of abortions.

There are no other studies with direct estimates that have assessed the trend in induced abortion over time and that would allow comparison with our findings. The two PNDS, conducted with a 10-year interval (1996-2006) ⁶6. Cecatti JG, Guerra GVQL, Sousa MH, Menezes GMS. Aborto no Brasil: um enfoque demográfico. Rev Bras Ginecol Obstet 2010; 32:105-11.^,77. Camargo RS, Santana DS, Cecatti JG, Pacagnella RC, Tedesco RP, Melo Jr. EF, et al. Severe maternal morbidity and factors associated with the occurrence of abortion in Brazil. Int J Gynaecol Obstet 2011; 112:88-92., and the 2 PNA, with a 6-year interval (2010 ²2. Diniz D, Medeiros M. Aborto no Brasil: uma pesquisa domiciliar com técnica de urna. Ciênc Saúde Colet 2010; 10 Suppl 1:959-66., 2016 ¹1. Diniz D, Medeiros M, Madeiro A. Pesquisa Nacional de Aborto 2016. Ciênc Saúde Colet 2017; 22:653-60.), did not identify a significant decrease in the estimated number of induced abortions. Indirect estimates based on hospital admissions due to complications from abortion in hospitals affiliated with the Brazilian Unified National Health System (SUS) show a decline in the number of abortions ⁸8. Martins-Melo FR, Lima MS, Alencar CH, Ramos Jr AN, Carvalho FHC, Machado MMT, et al. Tendência temporal e distribuição espacial do aborto inseguro no Brasil, 1996-2012. Rev Saúde Pública 2014; 48:508-20.^,99. Monteiro MFG, Adesse L, Drezett J. Atualização das estimativas da magnitude do aborto induzido, taxas por mil mulheres e razões por 100 nascimentos vivos do aborto induzido por faixa etária e grandes regiões. Brasil, 1995 a 2013. Reprod Clim 2015; 30:11-8.. However, these estimates are subject to error because they use parameters to correct for hospitalizations due to spontaneous abortions, admissions to private hospitals, and abortions that did not result in hospitalization. Thus, the decrease in hospitalizations may reflect a change in other aspects (proportion of admissions to private hospitals due to an increase in coverage by health plans; less occurrence of complications and hospitalizations due to the use of safer methods for termination of pregnancy; alteration in the proportion of spontaneous abortions due to changes in women’s age at pregnancy), and not in the practice of induce abortions itself.

Among the factors analyzed here that presented a significant association with induced abortion, four showed increasing prevalence during the period: women’s age, more schooling, teenage pregnancy, and more sexual partners. Thus, other factors not investigated here must explain the decrease in prevalence of induced abortion since 2005. We can raise several hypotheses related to changes in Brazilian women’s socioeconomic and reproductive characteristics and the scenario of HIV infection in the country. There was a decrease in the social vulnerability of the cohort’s women over time, with an increase in years of schooling and less report of sexual violence. We also found a significant increase in the proportion of women without previous pregnancies. In 1996-2006, data from the PNDS 2006 showed a reduction in Brazilian women’s fertility and an increase in the use of contraceptive methods, although with varying intensity according to economic class ²⁵25. Centro Brasileiro de Análise e Planejamento, Ministério da Saúde. Pesquisa Nacional de Demografia e Saúde da Criança e da Mulher PNDS 2006: dimensões do processo reprodutivo e da saúde da criança. 2009. http://bvsms.saude.gov.br/bvs/publicacoes/pnds_crianca_mulher.pdf (acessado em 05/Jun/2019).
http://bvsms.saude.gov.br/bvs/publicacoe... . PNDS data subsequent to 2006 are not available, which prevents comparison with the periods 2006-2010 and 2011-2016 in our study. Our study also showed an increase in contraceptive use, especially dual contraception (condoms associated with another highly effective method), although this information refers to the time of the interview and not to the moment when the abortion occurred. Brazilian data also show an increase in condom use, especially in casual sexual relations ²⁶26. Berquó E, Barbosa RM, Lima LP; Grupo de Estudos em População, Sexualidade e Aids. Trends in condom use: Brazil 1998 and 2005. Rev Saúde Pública 2008; 42 Suppl 1:34-44.^,2727. Departamento de DST, AIDS e Hepatites Virais, Secretaria de Vigilância em Saúde, Ministério da Saúde. Pesquisa de conhecimento, atitudes e práticas na população brasileira. http://www.aids.gov.br/system/tdf/pub/2016/59392/pcap_2013.pdf?file=1&type=node&id=59392&force=1 (acessado em 19/Out/2018).
http://www.aids.gov.br/system/tdf/pub/20... .

A factor not investigated in our study was the woman’s religion at the time of entry in the cohort. Although the PNAs in 2010 ²2. Diniz D, Medeiros M. Aborto no Brasil: uma pesquisa domiciliar com técnica de urna. Ciênc Saúde Colet 2010; 10 Suppl 1:959-66. and 2016 ¹1. Diniz D, Medeiros M, Madeiro A. Pesquisa Nacional de Aborto 2016. Ciênc Saúde Colet 2017; 22:653-60. found that Brazilian women of all religions reported previous induced abortions, the prevalence rates differ according to the woman’s religion. In the PNDS 1996, having a religion protected against abortion ⁶6. Cecatti JG, Guerra GVQL, Sousa MH, Menezes GMS. Aborto no Brasil: um enfoque demográfico. Rev Bras Ginecol Obstet 2010; 32:105-11.. Since we lack this information, we do not know to what extent changes in Brazilian women’s religious profile may have contributed to the reduction in the prevalence of induced abortion at entry in the cohort.

Most of the abortions reported in this study probably occurred before the diagnosis of HIV infection. However, Brazil’s advances in care for persons living with HIV/AIDS during the period of women’s inclusion in the cohort may have affected in some way the decision to not terminate pregnancy after diagnosis of the infection. These advances feature the reduction in mother-to-child transmission (MTCT) of HIV. France reported a significant drop in the proportion of induced abortions following the introduction of zidovudine (AZT) to reduce MTCT in 1994 ²⁸28. Bongain A, Berrebi A, Mariné-Barjoan E, Dunais B, Thene M, Pradier C, et al. Changing trends in pregnancy outcome among HIV-infected women between 1985 and 1997 in two southern French university hospitals. Eur J Obstet Gynecol Reprod Biol 2002; 104:124-8.. Although the INI cohort was only launched in 1996 (after the introduction of AZT), it is known that measures to prevent MTCT of HIV in Brazil were implemented gradually, not reaching 100% of WLHA ²⁹29. Domingues RMSM, Saraceni V, Leal MC. Mother to child transmission of HIV in Brazil: data from the "Birth in Brazil study", a national hospital-based study. PLoS One 2018; 13:e0192985.. According to a study in the South of Brazil in 2011, the main reason for termination of pregnancy after diagnosis of HIV infection was related to the infection itself (being infected or fear of bearing an infected child), reported by 47.6% of the women that underwent abortion ⁴4. Pilecco FB, Teixeira LB, Vigo A, Dewey ME, Knauth DR. Lifetime induced abortion: a comparison between women living and not living with HIV. PLoS One 2014; 9:e95570.. It is thus possible that with the expansion of measures to prevent MTCT of HIV, Brazilian women have opted less for termination of pregnancy if the concern with infection was the main reason for the induced abortion.

Other improvements in care for persons living with HIV/AIDS in Brazil include encouragement of early diagnosis and expansion of access to highly active antiretroviral therapy (HAART). A study at our service found an increase from 1996 to 2016 in the proportion of women with undetectable viral load, high CD4 count, and use of HAART at entry in the cohort ³⁰30. Zachek CM, Coelho LE, Domingues RMSM, Clark JL, De Boni RB, Luz PM, et al. The intersection of HIV, social vulnerability, and reproductive health: analysis of women living with HIV in Rio de Janeiro, Brazil from 1996-2016. AIDS Behav 2018; 23:1541-51., demonstrating that these women have also benefited from these advances. The declining mortality and increasing survival of persons living with HIV/AIDS ³¹31. Secretaria de Vigilância em Saúde, Ministério da Saúde. Boletim Epidemiológico: HIV/AIDS 2017. http://www.aids.gov.br/pt-br/pub/2017/boletim-epidemiologico-hivaids-2017 (acessado em 05/Out/2018).
http://www.aids.gov.br/pt-br/pub/2017/bo... , the reduction in MTCT of HIV infection, and the expansion of more effective contraceptive methods in our cohort, based not only on exclusive condom use, may have resulted in less use of abortion as a way of regulating fertility, affecting the reproductive choices of these WLHA. However, Brazilian WLHA still face numerous barriers to sexual and reproductive health services. A recent systematic review of studies with WLHA in Latin America revealed higher prevalence of unmet contraceptive needs, unplanned pregnancies, induced abortions, risk of sterilization in the immediate postpartum, and exposure to sexual and institutional violence in WLHA compared to WNLHA. The reduction in institutional barriers, facilitation of access to services, and reduction of stigma and discrimination by health professionals are necessary in order for WLHA to make their sexual and reproductive choices free of violence and coercion ³²32. Gómez-Suárez M, Mello MB, Gonzalez MA, Ghidinelli M, Pérez F. Access to sexual and reproductive health services for women living with HIV in Latin America and the Caribbean: systematic review of the literature. J Int AIDS Soc 2019; 22:e25273..

This study has some limitations. Since it was an outpatient-based study, the women included in the cohort could represent a group of WLHA with specific characteristics related to the service offered by INI/Fiocruz. However, the characteristics of the women observed in the four periods of inclusion in the cohort over the course of the 20 years - with a predominance of heterosexual transmission, increasing age and schooling, and reduction in the proportion of white women - are similar to the pattern observed in the HIV/AIDS epidemic in women in the Brazilian scenario ³¹31. Secretaria de Vigilância em Saúde, Ministério da Saúde. Boletim Epidemiológico: HIV/AIDS 2017. http://www.aids.gov.br/pt-br/pub/2017/boletim-epidemiologico-hivaids-2017 (acessado em 05/Out/2018).
http://www.aids.gov.br/pt-br/pub/2017/bo... , suggesting absence of selection bias in the women included in the cohort.

The second limitation relates to the form of measurement of induced abortion. Since induced abortion is illegal in Brazil, there are methodological difficulties in measuring this event, and the information obtained by direct interview may have underestimated its occurrence. However, the observed frequency was much higher than in Brazilian studies that measured abortion by direct estimates in WLHA and WNLHA, whether through interviews or the ballot-box technique, the latter considered less subject to under-recording since it guarantees the information’s secrecy ³³33. Olinto M, Moreira Filho DC. Estimating the frequency of induced abortion: a comparison of two methods. Rev Panam Salud Pública 2004; 15:331-6.^,3434. Medeiros M, Diniz D. Recommendations for abortion surveys using the ballot-box technique. Ciênc Saúde Colet 2012; 17:1721-4.. In the current study, even when we only analyzed the period 2011-2016 and women in the 18-39-year age bracket, the inclusion criterion for the period in which the two PNA were performed, we obtained higher prevalence rates for induced abortion (18.5%) than estimated in the two national surveys (15% and 13%). Although comparison with the two surveys are limited by differences in the study populations, the data suggest that there was no underestimation of induced abortion.

A third limitation was the lack of identification of the timing of the induced abortion in relation to HIV infection, information that was only available for 2006-2016, when nearly all of the induced abortions were performed before diagnosis of the infection. Brazilian studies that verified factors associated with induced abortion in WLHA and WNLHA ³3. Barbosa RM, Pinho AA, Santos NS, Filipe E, Villela W, Aidar T. Aborto induzido entre mulheres em idade reprodutiva vivendo e não vivendo com HIV/aids no Brasil. Ciênc Saúde Colet 2009; 14:1085-99.^,44. Pilecco FB, Teixeira LB, Vigo A, Dewey ME, Knauth DR. Lifetime induced abortion: a comparison between women living and not living with HIV. PLoS One 2014; 9:e95570., whether analyzing lifetime induced abortion ³3. Barbosa RM, Pinho AA, Santos NS, Filipe E, Villela W, Aidar T. Aborto induzido entre mulheres em idade reprodutiva vivendo e não vivendo com HIV/aids no Brasil. Ciênc Saúde Colet 2009; 14:1085-99. or pregnancies that ended in abortion ⁴4. Pilecco FB, Teixeira LB, Vigo A, Dewey ME, Knauth DR. Lifetime induced abortion: a comparison between women living and not living with HIV. PLoS One 2014; 9:e95570., found no important differences in the factors identified in each of these populations. The hypothesis raised by these studies is that the factors associated with induced abortion are similar in the two populations, but since WLHA display increased vulnerability, with higher prevalence of these factors, the prevalence of induced abortion is also higher ³3. Barbosa RM, Pinho AA, Santos NS, Filipe E, Villela W, Aidar T. Aborto induzido entre mulheres em idade reprodutiva vivendo e não vivendo com HIV/aids no Brasil. Ciênc Saúde Colet 2009; 14:1085-99.^,44. Pilecco FB, Teixeira LB, Vigo A, Dewey ME, Knauth DR. Lifetime induced abortion: a comparison between women living and not living with HIV. PLoS One 2014; 9:e95570.^,55. Pinho AA, Cabral CS, Barbosa RM. Differences and similarities in women living and not living with HIV: contributions by the GENIH study to sexual and reproductive healthcare. Cad Saúde Pública 2017; 33:e00057916.^,3535. Santos NJ, Barbosa RM, Pinho AA, Villela WV, Aidar T, Filipe EMV. Contexts of HIV vulnerability among Brazilian women. Cad Saúde Pública 2009; 25 Suppl 2:S321-33.. In Barbosa et al. ³3. Barbosa RM, Pinho AA, Santos NS, Filipe E, Villela W, Aidar T. Aborto induzido entre mulheres em idade reprodutiva vivendo e não vivendo com HIV/aids no Brasil. Ciênc Saúde Colet 2009; 14:1085-99., the difference in prevalence of induced abortion in WLHA and WNLHA lost significance when age, marital status, and lifetime number of sexual partners were included in the adjusted analyses. Thus, lack of identification of the timing of induced abortion in relation to HIV diagnosis should not be an important limitation to this study, since the factors associated with abortion do not appear to differ between these two populations of women.

One last limitation was the lack of inclusion of variable related to the pregnancy that resulted in abortion, which limited identification and discussion of factors associated with induced abortion in this population of WLHA.

Finally, the regression model with the total population of women has limitations, since women without prior abortion included women who had never become pregnant (and thus could not have aborted). However, we opted to present this model in order to allow comparison with other Brazilian studies that used the total population of women in their analyses. Since 90% of the women at inclusion in our cohort had already been pregnant, the two models (with the total population of women and only with those with previous pregnancy) presented the same results.

Conclusion

The study’s data show a high rate of lifetime induced abortion at inclusion in the cohort of WLHA in the city of Rio de Janeiro, with a significant reduction after 2006. This is an unprecedented result, since there are no previous studies in Brazil that have measured change over time in the frequency of induced abortion using direct estimates, whether in WLHA or in WNLHA.

The factors associated with abortion - older maternal age, more sexual partners, teenage pregnancy, drug use - and high prevalence of these factors are consistent with the results of other studies in WLHA and indicate their increased vulnerability, both to HIV infection and to unplanned pregnancies and induced abortion.

It was not possible to verify the reason for the observed drop in the prevalence of induced abortion. Some hypotheses include socioeconomic and reproductive changes and changes in the context of HIV infection/AIDS in Brazil, such as a reduction in the prevalence of risk factors, decreasing fertility, and greater access to contraceptive methods and HIV/AIDS services. Further studies, preferably population-based and using direct measurement methods, are necessary to confirm the downward trend in induced abortion and its determinants in Brazil.

References

Publication Dates

History