| TYPE Original Research PUBLISHED 06 December 2022 DOI 10.3389/frph.2022.1040640 Social inequalities in utilization of a feminist telehealth abortion EDITED BY Singh Rajender, Central Drug Research Institute (CSIR), India service in Brazil: A multilevel REVIEWED BY Eileen Ai-liang Yam, analysis Carrot Fertility, United States Sydney Calkin, Sara Larrea1*, Laia Palència2,3,4, Mariana Prandini Assis5 Queen Mary University of London, 1,2,3,4 United Kingdom and Carme Borrell 1 *CORRESPONDENCE Departament de Ciències Experimentals i de la Salut, Universitat Pompeu Fabra, Barcelona, Spain, 2 Agència de Salut Pública de Barcelona, Barcelona, Spain, 3CIBER Epidemiología y Salud Pública, Sara Larrea Madrid, Spain, 4Institut d’Investigació Biomèdica (IIB Sant Pau), Barcelona, Spain, 5Faculty of Social slarrea80@gmail.com Sciences, Universidade Federal de Goiás, Goiânia, Brazil SPECIALTY SECTION This article was submitted to Reproductive Epidemiology, a section of the journal Frontiers The disruption caused by the COVID-19 pandemic on health services around the in Reproductive Health world boosted interest over telehealth models of care. In Brazil, where abortion is RECEIVED 09 September 2022 heavily restricted, abortion seekers have long relied on international telehealth ACCEPTED 17 November 2022 services to access abortion pills. We conducted a cross-sectional multilevel PUBLISHED 06 December 2022 study to assess the effect of individual and contextual social factors on utilization CITATION of one such service. For the individual-level, we analyzed data from the records Larrea S, Palència L, Assis MP and Borrell C of abortion seekers contacting this feminist international telehealth organization (2022) Social inequalities in utilization of a during 2019 (n = 25,920). Individual-level variables were age, race, education feminist telehealth abortion service in Brazil: A level and pregnancy length. Contextual-level units were states, for which we multilevel analysis. Front. Reprod. Health 4:1040640. used data from the national Demographic Census and Household Surveys. doi: 10.3389/frph.2022.1040640 Contextual-level variables were household income per capita, adjusted net COPYRIGHT school attendance rate, percentage of racialized women and income Gini Index. © 2022 Larrea, Palència, Assis and Borrell. This We fitted five multilevel Poisson Mixed-effects models with robust variance to is an open-access article distributed under the estimate prevalence ratios (PR) of service utilization, which was defined as terms of the Creative Commons Attribution License (CC BY). The use, distribution or receiving abortion pills through the service. We found that only 8.2% of reproduction in other forums is permitted, requesters got abortion pills through the service. Utilization was higher among provided the original author(s) and the women who were older, white, more educated and 5–8-weeks pregnant. copyright owner(s) are credited and that the original publication in this journal is cited, in Independently of this, service utilization was higher in states with higher income accordance with accepted academic practice. and education access, with lower proportions of racialized women, and located No use, distribution or reproduction is in the South, Southeast and Central-West regions. We concluded that while permitted which does not comply with these terms. feminist telehealth abortion initiatives provide a life-saving service for some abortion seekers, they are not fully equipped to overcome entrenched social inequalities in their utilization, both at individual and contextual levels. KEYWORDS self-managed abortion, misoprostol, mifepristone, social inequalities, abortion access, multilevel analysis, telehealth, Brazil Introduction Along with other countries in Latin America and the Caribbean, Brazil has a long legacy of colonialism and persistent socioeconomic inequalities that have diminished during the last couple of decades (1–3). This improvement has been attributed to the establishment of cross-sector welfare policies, labor formalization, increasing education Frontiers in Reproductive Health 01 frontiersin.org Larrea et al. 10.3389/frph.2022.1040640 access and other distributive policies (1, 3). The welfare policies marginalized and impoverished population groups (18). For established after the country’s redemocratization were example, while abortion incidence is higher among women strengthened during the first decade of the 2000s and include who are racialized and have lower educational levels, those the creation of the SUS (Sistema Único de Saúde), a unified same women are more likely to have unsafe abortions, to healthcare system based on the principles of universality, need hospitalization after a self-managed abortion, to face equity, integrality and social participation (4). As a result, barriers to access post-abortion care and to be criminalized most of the population now has access to institutional (8, 14, 18, 19). On the other hand, white, educated, and richer delivery and modern contraceptive methods, some of the key women resort to local private clinics that offer safe abortion reproductive health indicators (4). services clandestinely, or travel abroad to obtain legal However, improvements in socioeconomic and abortions (19, 20). reproductive health indicators are not distributed equally in Currently, nearly 50% of clandestine abortions in Brazil are the Brazilian territory. The North and Northeast regions done using pills, mainly misoprostol (8). While the transition to have a higher proportion of racialized women, particularly medication abortion as the most common abortion method black and brown, and their population has generally lower contributed to a decrease of abortion-related morbidity and educational levels and lower income (5). Higher mortality (18), misoprostol still features in the national list of development rates remain concentrated in the South, substances and drugs subjected to special control (21). People Southeast and Central-West regions, which is apparent in cannot access abortion pills in pharmacies, thus often lower formal education, higher rates of unemployment and procuring them through vendors in the informal market and lower per capita income in the North and Northeast regions international online organizations, while risking being (6). Contraceptive use is higher among women in the South, criminalized (19, 22, 23). International feminist organizations Southeast and Central-West regions (7) and abortion facilitating access to medication abortion via telehealth have incidence is higher among women living in the North and operated in Brazil since 2004. Based online, these initiatives Northeast regions (8). Regional inequality is a problem long operate from territories where abortion is legal and deliver acknowledged by political economists in Brazil (9, 10). Its mifepristone and misoprostol often by mail. They also use origin is attributed to the historical social formation of the email to provide practical and emotional support before, regions that up to this day determines the relative during and after the abortion process. Those who are in the population composition of each region, which has been economic position to donate are requested a voluntary reproduced throughout the centuries (10). contribution to access the service, the equivalent of 70–90 Moreover, abortion is highly restricted in Brazil, where it is Euros (22, 24). The model varies slightly from organization to only legal in cases of rape, when the pregnancy poses a risk to organization, but they have in common the use of the the pregnant person’s life and when the fetus is anencephalic Internet to provide access to safe abortion information and (11, 12). Abortion seekers often fall out of the very narrow online accompaniment while ensuring the delivery of the pills legal grounds for abortion or, when under their scope, find through different methods. various barriers to access legal services (13, 14). Many of Social and racial inequalities in abortion access and safety those who cannot access legal services resort to the use of have been studied in Brazil at the individual-level (8, 14, 18). misoprostol, which is an analog of prostaglandin first Multilevel analyses have been used to examine adolescent introduced in the Brazilian market in 1986 to treat gastric fertility and reproductive behaviors. These analyses indicate ulcers (15). Soon after, Brazilian women discovered its that state-level income inequalities, coverage of reproductive abortifacient effect and started using it to self-induce health services, maternal mortality rates and regions are abortions (15). Following this discovery, Brazilian researchers associated with adolescent fertility (25–27). To the best of our documented the use of misoprostol to induce abortions and knowledge, there are no studies assessing the combined effect health institutions and researchers around the world followed of individual and contextual variables on clandestine abortion suit (15). Currently, misoprostol is recognized as a safe access. Additionally, little is known about the relationship method to induce abortions with limited health professional’s between structural inequalities and utilization of initiatives involvement and its use –alone or in combination with that use the Internet to increase access to safe abortion. mifepristone—is the method of choice in a wide variety of During the last few years, and particularly with the advent of contexts globally, including settings where abortion services COVID-19, Internet-based abortion provision has been are offered legally (16, 17). continuously growing both in Brazil and around the world Despite severe legal restrictions, abortion is a common (28, 29). With this study, we aim to contribute to the practice in Brazil. It is estimated that 1 in 5 women will have understanding of inequalities in abortion access in Brazil by at least one abortion by the time they turn 40 (8). While assessing how individual and contextual social factors impact women with diverse socioeconomic positions have abortions, utilization of an online feminist service supporting self- barriers to safe abortion disproportionately affect historically managed abortion. Frontiers in Reproductive Health 02 frontiersin.org Larrea et al. 10.3389/frph.2022.1040640 Methods 25–29, 30–34, 35–39 and 40–49 years old). For the multilevel analysis we used the original continuous variable Study design, population and information without the categories. sources • Race: Requesters were asked “among the following alternatives, do you recognize or identify yourself as color We conducted a cross-sectional multilevel study with or race: white, black, brown, yellow, indigenous?” We women being individual-level units and Brazilian states being maintained the original responses as categories for contextual-level units. For the individual-level, we used data descriptive statistics and categorized responses as white and from the records of a feminist telehealth service. Study racialized for the multilevel analysis. Racialized category participants were living in Brazil, requested the service included brown, black, yellow and indigenous women. between January 1st and December 31st, 2019, were less than • Higher educational level accessed: Requesters are asked “what 9 weeks pregnant and between 14 and 49 years old. Upon is the highest level of education you have attained?” and given service request, requesters fill out an online survey containing the following options: Incomplete/complete primary questions to identify eligibility for medication abortion (elementary) education, Incomplete/complete secondary according to the latest WHO guidelines available (16). They education (high school), incomplete/complete tertiary are also asked for personal contact information, shipping (technical) education, incomplete/complete university address and socio-demographic data (age, educational level education, incomplete/complete post-graduate education, and race) and a donation of the equivalent of 75 Euro, which incomplete/complete doctorate. We categorized responses as can be done via credit card or international bank transfer. basic education (up to complete elementary education), People who cannot donate but has answered an automatic secondary education (up to complete secondary education) email about the donation are still offered the pills. During the and university or more (incomplete tertiary education or study period, around 6% of those who accessed the pills more). through the service did not donate. A package containing one • Pregnancy length: Requesters are asked their pregnancy pill of 200 mg of mifepristone and 8 pills of 200 mcg of length in weeks and can choose a number of weeks misoprostol is shipped to people who fulfill the service between 4 and ≥9 weeks. Being less than 9 weeks pregnant requirements. Instructions on how to use the pills as well as is a requirement to access the service. We categorized detailed information about the abortion process are provided pregnancy length as follows: 4 weeks, 5–6 weeks, 7–8 weeks. via email in the requester’s language. Users are instructed to take the mifepristone pill orally and to use a dose of 800 mcg of misoprostol buccally or sublingually 24 h after having taken the mifepristone. During the study period, 25,920 people in Contextual level Brazil requested a medical abortion from the service and • Rate of requests: We calculated the rate of requests for each 2,121 received the abortion pills. state to assess if the geographic distribution of requests was For the contextual level, we used state-level data from the homogeneous. To do so, we divided the number of 2010 Demographic Census and the Annual Continuous requests in each state by the number of women at National Survey of Households of 2015, 2019 and 2020, reproductive age in the same state and then multiplied it conducted by the Brazilian Institute of Geography and by 100,000. We used data of the 2010 Demographic Census. Statistics (IBGE). • Percentage of service utilization: Using data from the service records, we calculated the proportion of requesters from a given state that accessed the pills through the service among requesters from the same state that requested the Variables service. • School attendance percentage: It is the percentage of women Outcome variable 18–24 years old who attend school at the appropriate level • Service utilization: Defined as having received a package with for their age group, among the total number of women in abortion pills as recorded by the service staff among all those the same age group. This data was extracted from the who requested the service (yes/no). Annual Continuous National Survey of Households-2nd quarter 2019. • Percentage of racialized women: We calculated the Independent variables percentage of women who self-identify as indigenous, Individual-level black, brown or yellow, among the total population of • Age: We calculated age from the date of birth and women in each state using data from the 2010 categorized it for descriptive purposes (14–19, 20–24, Demographic Census. Frontiers in Reproductive Health 03 frontiersin.org Larrea et al. 10.3389/frph.2022.1040640 • Gini: The Gini index measures the extent to which the the inclusion of the contextual variable in the model by distribution of income deviates from an equal distribution, comparing the variance with that of Model 0. where 0 represents perfect equality and 1 indicates perfect We used STATA 14 to run our analyses and R to plot the inequality. We used the Gini index of monthly income maps. distribution of persons aged 15 years and over for the year 2015. Data was collected within the National Survey of Households 2015. Ethical issues • Region: We used the IBGE classification of regions (North, Northeast, Southeast, South, Central-West). Upon acceptance of the terms and conditions of the service, • Household income per capita: It is the ratio between the total requesters consent to the use of their anonymized data for earnings (in Brazilian Reais, R$) and the total number of research purposes. We followed the principles of the residents of a household. The source of the data was the Declaration of Helsinki on Human Research and the study Annual Continuous National Survey of Households – 2020. was approved by the Drug Research Ethical Committee CEIm For the multilevel analysis, we divided the variable by -Parc de Salut MAR, Barcelona (Code: 2018/8145/I). 1,000 so that coefficients were associated to a 1,000 R$ increase in household income per capita. In November 2021, 1,000 R$ are equivalent to 161€. Results Individual-level results Analysis Our study population included 25,920 requesters. Table 1 Individual-level analysis presents descriptive data on the individual-level variables as We described the distribution of our sample according to well as the results of our bivariate analysis on service the categories of our dependent and independent variables. utilization. Most requesters were between 20 and 24 years old We also compared people who requested the service but did (34.9%), self-identified as white (48.3%), had access to not use it (i.e., filled out the online survey but did not receive secondary or university education (46.5%) and were 6 weeks the pills) with those who used the service (i.e., received a pregnant or less (75.3%). package with pills), according to their socio-demographic 2,121 requesters (8.2%) accessed abortion pills through the characteristics using a Pearson’s X2 test. service during the study period. There was a clear utilization gradient based on age: 11.9% of people in the 40–49 category Context-level analysis received the pills through the service, but only 4.4% of people We first described all the contextual variables for states and 14–19 years old did. The proportion of white women who regions. We then mapped quintiles of all our contextual used the service almost triples that of indigenous women who continuous variables and calculated the correlations and did and is 3 percentage points higher that the proportion of significance levels among all the continuous variables at the brown, black and yellow women who used the service. contextual level. We also performed ANOVA tests (or Women with university education received the pills 4 times Kruskal–Wallis tests when the hypothesis of equal variances more than their peers with basic education. People who were was not fulfilled) to examine the correlation between our between 5- and 8-weeks pregnant received the pills at double categorical variable regions and the contextual continuous the rate than those who were earlier in pregnancy. The variables. association between service utilization and all individual-level variables was statistically significant. Multilevel analysis We fitted five multilevel Poisson Mixed-effects models with robust variance, with individual data in level 1 and state data in Context-level results level 2, to estimate prevalence ratios (PR) of service utilization with 95% confidence intervals (CI) and variance of random The distribution of context level variables by state and effects. Model 0 included all the individual-level variables. In region is presented in Table 2. Brazil has 27 states that are Models 1–5 we included all the individual-level variables and organized in 5 regions. The states’ population size varies one contextual variable in each model: region in Model 1, widely: the state with the smallest population is Roraima, household income per capita in Model 2, school attendance which has 631,181 inhabitants and 181,783 women at percentage in Model 3, percentage of racialized women in reproductive age, while São Paulo, the largest state in the Model 4 and income Gini index in Model 5. In each of these country, has more than 46 million inhabitants and more than models we calculated the percentage of variance explained by 12 million women at reproductive age. Frontiers in Reproductive Health 04 frontiersin.org Larrea et al. 10.3389/frph.2022.1040640 TABLE 1 Socio-demographic characteristics of study participants, service utilization and p value of the association between socio-demographic characteristics and service utilization. Total Utilization Requested the Yes (received the No (did not receive p service pills) the pills) N (%) N (%) N (%) Variable 25,920 (100.00) 2,121 (8.18) 23,799 (91.82) Age (years) 0.001 14–19 6,436 (24.83) 285 (4.43) 6,151 (95.57) 20–24 9,051 (34.92) 724 (8.00) 8,327 (92.00) 25–29 5,075 (19.58) 524 (10.33) 4,551 (89.67) 30–34 2,999 (11.57) 330 (11.00) 2,669 (89.00) 35–39 1,735 (6.69) 184 (10.61) 1,551 (89.39) 40–49 624 (2.4) 74 (11.86) 550 (88.14) Race 0.001 White 12,510 (48.26) 1,232 (9.85) 11,278 (90.15) Brown 9,262 (35.73) 621 (6.70) 8,641 (93.30) Black 3,276 (12.64) 216 (6.59) 3,060 (93.41) Yellow 673 (2.6) 45 (6.69) 628 (93.31) Indigenous 199 (0.77) 7 (3.52) 192 (96.48) Higher educational level accessed 0.001 Basic education 2453 (9.46) 76 (3.10) 2,377 (96.90) Secondary education 11,427 (44.09) 567 (4.96) 10,860 (95.04) University or more 12,040 (46.45) 1,478 (12.28) 10,562 (87.72) Pregnancy length 0.001 4 weeks 9135 (35.24) 466 (5.10) 8,669 (94.90) 5–6 weeks 10,381 (40.05) 1,067 (10.28) 9,314 (89.72) 7–8 weeks 6404 (24.71) 588 (9.18) 5816 (90.82) Other regional characteristics also differ widely. For follow the same pattern than the other maps. Generally, example, the North and Northeast regions have the highest inequality is the highest in Central-Western and Northeastern proportions of racialized women (76% and 70% respectively), states and the lowest in the Southern states. the lowest school attendance rates (23.2% and 23.7% The relationship between contextual variables is shown in respectively) and the states with the lowest income per capita Table 3. There is a positive and statistically significant (676R$ for Maranhão, 796 R$ for Alagoas and 852R$ for correlation between utilization and household income per Amazonas). The Central-West region has the highest school capita (r = 0.727) and adjusted net school attendance rate attendance rate (37.7%), and the highest income inequality (r = 0.409). This means that the probability of using the (Gini index: 0.498). The South region has the lowest service is higher in states with higher income per capita and proportion of racialized women (20.8%) and the lowest with higher school attendance percentages. We also found an income inequality in the country (Gini index: 0.450). inverse correlation (r = −0.659) between utilization and the A selection of variables in Table 2 are mapped in Figure 1, percentage of racialized women, meaning that the probability which shows the geographic distribution of requests, percentage of accessing the pills through the service is lower in states of service utilization, household income per capita and income with higher proportions of racialized women. There is also a Gini index by state. The first three maps show the same pattern: statistically significant relationship between utilization and the number of requests, the percentage of service utilization and region, meaning that there is a higher probability of service the household income per capita are higher in the states of the utilization in the South, Southeast and Central-West regions. South, Southeast and Central-West regions and lower in the Utilization is not statistically significantly correlated with the North and Northeast regions. The maps of education level Gini index. and percentage of racialized women (not shown) follow the We also found positive correlations between the school same pattern. The map of the Gini index highlights the slight attendance percentage and income and region, meaning that differences in income inequality by regions and does not school attendance rate is higher in states with higher income Frontiers in Reproductive Health 05 frontiersin.org Larrea et al. 10.3389/frph.2022.1040640 TABLE 2 Context-level variables by region and state. Region and Population1 Women at School Percentage Household Gini Number of Utilization State reproductive attendance of income per index6 requests per (% of age (14–49 percentage3 racialized capita (R 100,000 requesters years)2 women4 $)5 women at that reproductive received the age pills through the service) Brazil 211,755,692 58,720,495 29.7 51.5 1,380 0.491 44.3 8.17 North Region 18,672,591 5,411,780 23.2 76.0 - 0.473 20.6 3.49 Rondônia 1,796,460 519,713 30.6 64.3 1,169 0.452 25.4 3.79 Acre 894,470 258,680 27.8 75.9 917 0.5 15.5 0 Amazonas 4,207,714 1,206,738 24.3 77.9 852 0.476 19.5 3.4 Roraima 631,181 181,783 26.8 78.0 983 0.5 41.8 7.89 Pará 8,690,745 2,532,687 18.4 77.6 883 0.465 17.2 2.98 Amapá 861,773 257,008 34.5 75.0 893 0.457 27.2 0 Tocantins 1,590,248 455,171 27.2 75.1 1,060 0.504 28.1 5.47 Northeast 57,374,243 16,531,575 22.7 70.0 - 0.484 21.6 5.57 Maranhão 7,114,598 2,067,949 20.2 77.4 676 0.506 13.0 2.6 Piauí 3,281,480 943,623 28.1 75.2 859 0.505 20.7 2.56 Ceará 9,187,103 2,647,793 25.8 67.3 1,028 0.453 22.8 6.12 Rio Grande do Norte 3,534,165 1,002,710 26.2 58.0 1,077 0.503 18.2 7.96 Paraíba 4,039,277 1,147,349 25.1 59.3 892 0.51 22.2 7.06 Pernambuco 9,616,621 2,752,837 23.1 62.5 897 0.492 26.0 6.98 Alagoas 3,351,543 988,690 17.1 68.2 796 0.438 21.6 7.48 Sergipe 2,318,822 685,381 22 71.6 1,028 0.46 20.7 5.63 Bahia 14,930,634 4,295,243 20.6 77.6 965 0.481 23.2 4.41 Southeast 89,012,240 24,099,986 32.5 44.0 - 0.477 59.6 9.13 Minas Gerais 21,292,666 5,777,126 29.7 54.1 1,314 0.478 50.7 8.81 Espírito Santo 4,064,052 1,110,013 29.2 57.1 1,347 0.471 43.5 7.45 Rio de Janeiro 17,366,189 4,661,660 33 51.6 1,723 0.454 59.0 10 São Paulo 46,289,333 12,551,187 34 35.2 1,814 0.47 65.3 9.05 South 30,192,315 7,992,461 36.8 20.8 - 0.45 53.1 8.93 Paraná 11,516,840 3,099,169 37.1 28.9 1,508 0.459 51.6 8.76 Santa Catarina 7,252,502 1,951,305 40.2 15.4 1,632 0.419 59.4 8.19 Rio Grande do Sul 11,422,973 2,941,987 34.2 16.2 1,759 0.487 50.4 9.7 Central-West 16,504,303 4,684,693 37.3 57.5 - 0.498 47.9 7.76 Mato Grosso do Sul 2,809,394 764,065 31.4 52.0 1,488 0.479 45.3 8.38 Mato Grosso 3,526,220 977,797 32.4 62.2 1,401 0.445 43.2 6.4 Goiás 7,113,540 2,010,623 38 57.6 1,258 0.436 39.8 5.25 Distrito Federal 3,055,149 932,208 45.4 57.1 2,475 0.555 72.3 11.28 Sources: (1) IBGE, Population estimations 2020. (2) IBGE, Population projections 2018. (3) IBGE, National Survey of Households- 2nd quarter 2019. (4) IBGE, Demographic Census, 2010. (5) National Survey of Households-2020. (6) National Survey of Households-2015. and in some regions. There was a negative association between Multilevel results the proportion of racialized women and school attendance rate and income per capita, meaning that in states where the Table 4 shows the results for our multilevel models. The proportion of racialized women is higher, school attendance null model indicates a positive association between utilization rate and income per capita are lower. and the individual-level variables of race, higher educational Frontiers in Reproductive Health 06 frontiersin.org Larrea et al. 10.3389/frph.2022.1040640 FIGURE 1 Number of requests per 100,000 women at reproductive age, percentage of service utilization, household income per capita and Gini index according to Brazilian states. TABLE 3 Correlations and their p-values between variables of the contextual level. Utilization Adjusted net school Percentage of Household income Gini Index attendance rate racialized women per capita Utilization 1 Adjusted net school attendance rate 0.4095 1 0.0339 Percentage of racialized women −0.659 −0.6283 1 0.0002 0.0004 Household income per capita 0.7273 0.8177 −0.6877 1 <0,0001 <0,0001 0.0001 Gini Index 0.0586 −0.0298 0.2702 0.0784 1 0.7716 0.8828 0.1728 0.6974 Regiona 0.0018 0.0001 <0.0001 0.0007 0.5949 a Region is a categorical value so the p-value is the one for the ANOVA test with the other variables. level accessed and age. This means that the probability of service education and 69% higher for those who attended secondary utilization is 21% higher for white women than for racialized education than for women with basic education; and increases women; 2.7 times higher for women who accessed university 1% for the increase of each year of age. For pregnancy length Frontiers in Reproductive Health 07 frontiersin.org Larrea et al. 10.3389/frph.2022.1040640 TABLE 4 Prevalence ratios (PR) of service utilization with 95% confidence intervals and residual variance for multilevel logistic regressions. Model 0 Model 1 Model 2 Model 3 Model 4 Model 5 PR (IC 95%) PR (IC 95%) PR (IC 95%) PR (IC 95%) PR (IC 95%) PR (IC 95%) Race Racialized 1 1 1 1 1 1 White 1.21 (1.16–1.27) 1.20 (1.14–1.25) 1.21 (1.16–1.27) 1.20 (1.15–1.26) 1.20 (1.14–1.26) 1.21 (1.15–1.27) Higher educational level accessed Basic education 1 1 1 1 1 1 Secondary education 1.69 (1.34–2.12) 1.69 (1.34–2.12) 1.69 (1.34–2.12) 1.69 (1.34–2.12) 1.69 (1.34–2.13) 1.69 (1.34–2.12) University or more 3.73 (3.05–4.57) 3.75 (3.06–4.70) 3.72 (3.04–4.54) 3.73 (3.05–4.56) 3.74 (3.05–4.58) 3.73 (3.05–4.57) Age 1.01 (1.01–1.02) 1.01 (1.01–1.02) 1.01 (1.01–1.02) 1.01 (1.01–1.02) 1.01 (1.01–1.02) 1.01 (1.01–1.02) Pregnancy length 4 weeks 1 1 1 1 1 1 5–6 weeks 1.87 (1.70–2.05) 1.87 (1.70–2.06) 1.87 (1.70–2.05) 1.87 (1.70–2.05) 1.87 (1.70–2.05) 1.87 (1.70–2.05) 7–8 weeks 1.86 (1.65–2.09) 1.86 (1.65–2.09) 1.86 (1.65–2.09) 1.86 (1.65–2.09) 1.86 (1.65–2.09) 1.86 (1.65–2.09) Region North 1 Northeast 1.53 (1.06–2.22) Southeast 2.43 (1.76–3.35) South 2.21 (1.58–3.09) Central–West 2.00 (1.30–3.07) Household income per capita (1,000 Reais) 1.59 (1.27–2.00) Adjusted net school attendance rate 1.02 (1.00–1.04) % of racialized women 0.992 (0.98–0.99) Gini Index (2015) 0.052 (0.00–4.29) Variance 0.232 0.015 0.011 0.098 0.072 0.237 % variance reduction with respect to model 0 93.4 95.1 57.7 68.8 −2.2 the association is positive, meaning that when compared with indicate an association between utilization and income those who are 4 weeks pregnant, people between the 5th and inequality. the 8th week of pregnancy use the service 86%–87% more. Results for Model 1 show a positive association between service utilization and region, meaning that when compared Discussion with the North region, the probability of using the service is 53% higher in the Northeast, 100% higher in the Central- We found evidence of both individual and contextual West, 121% higher in the South and 143% higher in the inequalities in utilization of the telehealth abortion service we Southeast region. Region accounts for 93.4% of the state-level analyzed. The prevalence of service utilization was higher variance in utilization. Model 2 indicates that utilization among women who were older, white, with higher levels of increases 59% (95% CI 27%–100%) for each increase of 1,000 education and between 5 and 8 weeks pregnant. R$ in household income per capita, and 95.1% of the variance Independently of individual characteristics, service utilization in state-level utilization is attributed to income on its own. was higher in states with higher incomes, higher percentage of Model 3 indicates that utilization is also positively educated women, lower percentage of racialized women and associated with the adjusted net school attendance rate, located in the South, Southeast and Central-West regions. meaning that the probability of using the service increases 2% Income inequality was not associated with utilization. (95% CI 0.1%–4.0%) for each percentage point increase in Our analysis of requests showed that women of all age school attendance. Independently, school attendance rate groups, races and education levels resorted to the telehealth accounts for 57.7% of the state-level variance in utilization. service when seeking an abortion, which is in line with Model 4 indicates a negative association between utilization previous studies showing that Internet is an important source and percentage of racialized women, meaning that the of information and abortion access in Brazil (24, 30). probability of using the service decreased 1% for each point However, women from regions that are richer, have higher increase in percentage of racialized women. Model 5 did not education levels and lower proportions of racialized people Frontiers in Reproductive Health 08 frontiersin.org Larrea et al. 10.3389/frph.2022.1040640 were more likely to request the service, which contrasts with the younger women may not have information about its existence, higher abortion incidence in the North and Northeast regions lack access to private devices to connect to the Internet, face (8). One possible explanation for this territorial inequality are more difficulties to gather the money for the requested the differences in Internet access: while more than 70% of donation and to find a way to transfer it internationally. In Brazilians have some kind of Internet connection, the any case, young women who do not get abortion support will coverage and quality of the Internet vary widely according to get the abortion by other means and may be at higher risk of region and income. Most of the North and Northeast regions being criminalized, as research in the region has shown (34, 35). depend on satellite or mobile connections and, while 99% of Race and educational level are also important determinants of households with higher socioeconomic status are connected to utilization in this abortion service. In Brazil, abortion incidence is the Internet, only 43% of households of lower socioeconomic higher among racialized women and women with lower status have internet access (31). Thus, differences in Internet socioeconomic status (8), but they face more barriers to access access could act as the intermediate mechanism by which safe abortion methods and post-abortion care (14). They are also geographic and economic inequalities determine access and more often targeted by the criminal justice system (14, 18). Our utilization to this telehealth service. results show that this pattern still stands for those who resort to We also found that only a small proportion of requesters this feminist service, thus revealing that structural inequality (8.2%) actually obtained abortion pills through the service, operates both in the formal healthcare system and in civil society which generally indicates poor access to the pills and strong initiatives and highlighting the need for policies that target barriers to service utilization. Moreover, among the small entrenched racial inequalities in abortion access. percentage of requesters who were able to access the pills Service utilization was also associated with pregnancy through the service, most were white, 5–8 weeks pregnant and length, with women who are between 5 and 8 weeks pregnant educated, indicating the existence of entrenched social utilizing the service more often. The fact that women who are inequalities in this service utilization. While the organization very early in their pregnancies use this service at lower rates has a policy of providing the service also to those who cannot could be explained by some people requesting the service donate, only 6% of users received the pills without donating. before confirming the pregnancy, combined with a high One possible explanation for the small proportion of people proportion (between 11% and 30%) of early pregnancies with lower economic status utilizing the service is that those ending in natural miscarriages (36) or because being early in with difficulties to donate do not believe that they will be able the pregnancy translates into having more time to find local to access the service without making a payment, which would solutions. However, those who are 9 weeks pregnant or more indicate that there are symbolic barriers deterring people from and thus do not fulfill the service requirements, are likely to continuing communicating with the service after filling out face more difficulties accessing safe abortion elsewhere, which the service request. From our perspective, both the amount of could cause further delays or even mean they will have to the requested donation and the requirement of an carry the pregnancy to term, as has been demonstrated by international payment method – credit card or bank transfer- other studies in the region (37, 38). could also act as barriers for this service utilization. Our context-level analysis showed that the probability of Supporting our hypothesis, a study in Kenya found that some using the service increases when income and school payment methods can constitute barriers to access and can attendance are higher at the state-level and is lower in states trigger delays in abortion services utilization (32). Regarding with higher proportions of racialized women. Furthermore, the amount of the requested donation, in 2021 it was some of these variables are also correlated among them, equivalent to 50% of a monthly minimum salary in Brazil, confirming the compounded nature of inequality and which makes the donation impossible to afford for most exclusion shown by previous research showing that under- women and would explain why many of them do not follow- served communities are also often racialized, with lower up on the communication after having requested the service. income and access to formal education (9, 39). Similarly, in line with previous studies, we found that older The results of our multi-level analysis showed that while women accessed the service at higher rates than their younger service utilization is associated with individual characteristics, peers. Being young can be a barrier for abortion access in most of its state-level variability is explained by contextual several ways. For example, in a review of the regional variables such as region and state income per capita, which evidence on pregnancy and abortion among girls 15 years old account for more than 90% of the variance in state-level or younger, authors found that girls and young women face utilization. Thus, our results indicate that social inequalities in particular barriers to access legal abortion services, such as the abortion access are not a unique feature of this service, nor a lack of clarity on clinical protocols for underage women, as result of its users’ individual characteristics, but they also well as norms that require parental consent or judicial reflect the broader social context in which Brazilians authorization when parents do not consent or are not seek abortion services. While an important promise of involved in the process (33). Regarding this specific service, medication abortion provision is to dismantle geographic Frontiers in Reproductive Health 09 frontiersin.org Larrea et al. 10.3389/frph.2022.1040640 barriers in abortion access (40), we showed that inequalities However, the study’s population was not representative of the based on age, race, income, and formal education, can be Brazilian population but only of women who requested one reproduced even in a remote, Internet-based service. On top specific service, which caused a selection bias as shown in the of unequal access to internet and income, other reasons such results. Also, interpretations regarding the analysis of the as lower levels of health literacy, familiarity with online pregnancy length in our study should be cautious: Because we transactions and empowerment to act on their reproductive found a high percentage of missing cases in the data on last desires may explain why women from the most menstrual period, we used limited data coming from a self- disenfranchised population groups and territories are not able assessment of pregnancy length in number of weeks, which due to to access abortion pills through this service. the characteristics of the service, might reflect an information bias. The evidence of inequalities in the utilization of this abortion service highlights, first, the relevance and need of working to improve equity in access to abortion support. Second, our research Conclusion points to the pressing need to consider the local context in structuring services so that telehealth models of care can fulfill Our study shows that while feminist telehealth abortion their aim of confronting contextual and individual inequalities in services, such as the one we analyzed, secure safe abortion abortion access, instead of reproducing them. Much can be access for some people in highly restrictive contexts, they are learned from feminist initiatives that have implemented specific not equipped to overcome entrenched social inequalities in strategies to tackle social inequalities in access to their support. For safe abortion access. Individual-level inequalities in service example, some feminist organizations use phones, rather than utilization are associated to age, education level and race. Internet, to provide abortion information, which amplifies access Contextual-level inequalities are associated to region, income to disadvantaged population groups. Others organize in-person per capita, percentage of racialized women and schooling. meetings with large groups of women to ensure that all the Thus, our findings suggest that telehealth abortion services necessary information as well as the medication can be delivered need to structure their models of care with specific attention in one single opportunity (37, 41, 42). While not all feminist local to how local and structural inequalities interact with abortion initiatives in the region request a payment or donation in provision. In addition, large scope measures, such as exchange for their support, those who do are able to establish an decriminalization and local provision of public abortion care, amount that is better suited for the specific context they work in, are necessary to secure equity in safe abortion access. making the service affordable for at least most people in their own territories. Other activists have established alliances with health professionals to secure the people they support will have access to Data availability statement the public health system if needed and desired (43). More research is necessary to understand how social The datasets presented in this article are not readily inequalities in access to remote abortion services operate in available because of security concerns of the organization. different contexts, including less restrictive and less unequal Requests to access the datasets should be directed to settings. In terms of services and policies, our results suggest that slarrea80@gmail.com. remote Internet-based initiatives supporting abortion access are not sufficient to reach the most marginalized social groups in a very unequal context such as Brazil, which underscores the need Ethics statement for local and in-person provision of abortion services or for remote alternatives that are specifically tailored to local contexts. The studies involving human participants were reviewed and Thus, to secure equity in safe abortion access, abortion approved by Drug Research Ethical Committee CEIm -Parc de decriminalization and provision of public abortion care are needed. Salut MAR, Barcelona (Code: 2018/8145/I). Written informed consent for participation was not provided by the participants’ legal guardians/next of kin because participants had already Strengths and limitations consented to the use of their use of their anonymized data for research purposes upon acceptance of the terms and conditions To our knowledge, this is the first multilevel analysis seeking of the service. to explain social inequalities in feminist telehealth abortion services. Moreover, the large study population made it possible to arrive to sound conclusions. Thus, our study Author contributions contributes with fresh data on the relationship between social contexts, individual characteristics, and inequality in safe SL, LP and CB contributed to the conception and design of abortion services utilization, which is the study’s major strength. the study. SL and LP performed the statistical analysis. SL Frontiers in Reproductive Health 10 frontiersin.org Larrea et al. 10.3389/frph.2022.1040640 drafted the first version of the manuscript. SL and MPA revised Publisher’s note the draft. All authors contributed to the article and approved the submitted version. All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the Conflict of interest editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its The authors declare that the research was conducted in the manufacturer, is not guaranteed or endorsed by the absence of any commercial or financial relationships that could publisher. be construed as a potential conflict of interest. References 1. de Andrade LOM, Filho AP, Solar O, Rígoli F, de Salazar LM, Serrate PC-F, 2018. Cad Saude Publica. (2020) 36(Suppl 1):e00190418. doi: 10.1590/0102- et al. Social determinants of health, universal health coverage, and sustainable 311X00190418 development: case studies from Latin American countries. Lancet. (2015) 385 19. Diniz D, Medeiros M. Itinerários e métodos do aborto ilegal em cinco (9975):1343–51. doi: 10.1016/S0140-6736(14)61494-X capitais brasileiras. Cien Saude Colet. (2012) 17(7):1671–81. doi: 10.1590/S1413- 2. Landmann-Szwarcwald C, Macinko J. A panorama of health inequalities in 81232012000700002 Brazil. Int J Equity Health. (2016) 15(1):1–3. doi: 10.1186/s12939-016-0462-1 20. Biller D, Calatrava A, Pollastri T. Brazilian women head to Argentina to 3. Góes C, Karpowicz I. Inequality in Brazil : A Regional Perspective.; 2017. avoid abortion ban. AP News. Available from: https://apnews.com/article/brazil- Available from: https://www.imf.org/en/Publications/WP/Issues/2017/10/31/ abortion-ban-884b4f8e96d773d15d503cefc4930c32 Published 2021. Inequality-in-Brazil-A-Regional-Perspective-45331 21. Brasil. Regulamento Técnico Sobre Substâncias e Medicamentos Sujeitos a 4. França GVA, Restrepo-Méndez MC, Maia MFS, Victora CG, Barros AJD. Controle Especial. Brasil; 1998. Available from: https://bvsms.saude.gov.br/bvs/ Coverage and equity in reproductive and maternal health interventions in saudelegis/svs/1998/prt0344_12_05_1998_rep.html Brazil: impressive progress following the implementation of the unified health system. Int J Equity Health. (2016) 15(1):1–12. doi: 10.1186/s12939-016-0445-2 22. Gomperts R, Van Der Vleuten K, Jelinska K, Da Costa CV, Gemzell- Danielsson K, Kleiverda G. Provision of medical abortion using telemedicine in 5. IBGE. Desigualdades sociais por sor ou raça no brasil. Estud e Pesqui Brazil. Contraception. (2014) 89(2):129–33. doi: 10.1016/j.contraception.2013.11.005 Informações Demográficas e Socioeconômicas. (2019) 41:1–12. Available from: https://www.ibge.gov.br/estatisticas/sociais/populacao/25844-desigualdades-sociais- 23. Assis MP, Erdman JN. In the name of public health: misoprostol and the por- cor-ou-raca.html%0AEstudos.018v21n3p452 new criminalization of abortion in Brazil. J Law Biosci. (2021) 8(1):1–20. doi: 10.1093/jlb/lsab009 6. dos SG, Pales RC, Rodrigues SG. Desigualdades regionais no brasil – 1991- 2010. Intersci Place. (2014) 1(31):145–73. doi: 10.6020/1679-9844/3106 24. Wollum A, Larrea S, Gerdts C, Jelinska K. Requests for medication abortion support in Brazil during and after the Zika epidemic. Glob Public Health. (2021) 7. Farias MR, Leite SN, Tavares NU, Oliveira MA, Arrais PS, Bertoldi AD, et al. 16(3):366–77. doi: 10.1080/17441692.2020.1809694 Use of and access to oral and injectable contraceptives in Brazil. Rev Saude Publica. (2016) 50(supl 2):1s–10s. doi: 10.1590/S1518-8787.2016050006176 25. Filho ADPC, Kawachi I. Income inequality is associated with adolescent fertility in Brazil: a longitudinal multilevel analysis of 5,565 municipalities. BMC 8. Diniz D, Medeiros M, Madeiro A. National abortion survey 2016. Cienc e Public Health. (2015) 15(1):1–7. doi: 10.1186/s12889-015-1369-2 Saude Coletiva. (2017) 22(2):653–60. doi: 10.1590/1413-81232017222.23812016 26. Borges ALV, Chofakian CdN, Sato APS, Fujimori E, Duarte LS, Gomes MN. 9. Bacelar TA. “Questão regional”e a “questão nordestina.”. In: F Maués, editors. Fertility rates among very young adolescent women: temporal and spatial trends in Celso furtado e o brasil. 1st ed (2000). p. 71–92. São Paulo: EDITORA Brazil.. BMC Pregnancy Childbirth. (2016) 16(1):1–9. doi: 10.1186/s12884-016-0843-x FUNDAÇÃO PERSEU ABRAMO. Available from: https://www.ifibe.edu.br/arq/ 201507272115271743262037.pdf#page=75 27. Borges ALV, Duarte LS, Lay AAR, Fujimori E. Individual and context correlates of the oral pill and condom use among Brazilian female adolescents. 10. Rands A. Desigualdades regionais No brasil: Natureza, causas, origens e BMC Womens Health. (2021) 21(1):1–11. doi: 10.1186/s12905-021-01447-6 soluções. Rio de Janeiro: Elsevier (2011). 28. Aiken ARA, Starling JE, Gomperts R, Scott JG, Aiken CE. Demand for self- 11. Brasil. DECRETO-LEI No 2.848. Brasil; 1940. Available from: https://www2. managed online telemedicine abortion in eight European countries during the camara.leg.br/legin/fed/declei/1940-1949/decreto-lei-2848-7-dezembro-1940-412868- COVID-19 pandemic: a regression discontinuity analysis. BMJ Sex Reprod Heal. publicacaooriginal-1-pe.html (2021) 47(4):238–45. doi: 10.1136/bmjsrh-2020-200880 12. Brasil. Supremo Tribunal Federal. ADPF n° 54, del 11/12 de abril de 2012. 29. Diniz D, Ambrogi I, Carino G. Telemedicine abortion: a channel to Brazilian 13. Madeiro AP, Diniz D. Serviços de aborto legal no brasil – um estudo women. Int J Gynecol Obstet. (2021) 155(1):161–2. doi: 10.1002/ijgo.13821 nacional. Cienc e Saude Coletiva. (2016) 21(2):563–72. doi: 10.1590/1413- 30. Leone T, Coast E, Correa S, Wenham C. Web-based searching for abortion 81232015212.10352015 information during health emergencies: a case study of Brazil during the 2015/ 14. Freitas Goes E, de Souza Menezes GM, Chagas de Almeida MDC, Velho 2016 Zika outbreak. Sex Reprod Heal Matters. (2021) 29(1):1–13. doi: 10.1080/ Barreto-de-Araújo T, Valongueiro Alves S, Seabra Soares de Britto Alves MT, 26410397.2021.1883804 et al. Barriers in accessing care for consequence of unsafe abortion by black 31. Farrugia B. Pandemic exposes “digital inequality” in Brazilian internet women: evidence of institutional racism in Brazil. J Racial Ethn Heal access. LABS- Lat Am Bussiness Stories. 2020. Available from: https://labsnews. Disparities. (2021) 8(6):1385–94. doi: 10.1007/s40615-020-00900-w com/en/articles/society/pandemic-digital-inequality-brazil/ 15. Ministerio da Saúde. 20 Anos de Pesquisas Sobre Aborto No Brasil. Brasilia- 32. Mutua MM, Manderson L, Musenge E, Ochieng Achia TN. Policy, law and DF; 2009. Available from: www.saude.gov.br post-abortion care services in Kenya. PLoS One. (2018) 13(9):1–18. doi: 10.1371/ 16. WHO. Medical Management of Abortion. Geneva; 2018. Available from: journal.pone.0204240 http://apps.who.int/ 33. Escobar JC, Ariza Navarrete S, Keller V. Embarazo y maternidad en niñas y 17. Popinchalk A, Sedgh G. Trends in the method and gestational age of adolescentes menores de 15 años: aportes de evidencia socio-sanitaria y jurídica en abortion in high-income countries. BMJ Sex Reprod Heal. (2019) 45(2):95–103. la región. 1st ed Lima: CLACAI (2019). doi: 10.1136/bmjsrh-2018-200149 34. Cruz Sánchez V. Fin a una década de criminalización por aborto contra 18. Domingues RMSM, Fonseca SC, Leal MDC, Aquino EML, Menezes GMS. mujeres pobres en guanajuato. Debate Fem. (2011) 43:176–91. doi: 10.22201/ Aborto inseguro no brasil: revisão sistemática da produção científica, 2008- cieg.2594066xe.2011.43.852 Frontiers in Reproductive Health 11 frontiersin.org Larrea et al. 10.3389/frph.2022.1040640 35. Citizen’s Coalition for the Decriminalization of Abortion on Grounds of 39. Comisión Económica para América Latina y el Caribe (CEPAL). Panorama social Health Ethics and Fetal Anomaly. From hospital to jail: the impact on women de américa Latina 2020. Santiago: CEPAL (2021). doi: 10.22201/fesa.rdp.2021.3.06 of El Salvador’s Total criminalization of abortion. Reprod Health Matters. 40. Calkin S, Freeman C. Trails and technology: social and cultural geographies (2014) 22(44):52–60. Available from: https://www.jstor.org/stable/43288361% of abortion access. Soc Cult Geogr. (2019) 20(9):1325–32. doi: 10.1080/14649365. 0AJSTOR doi: 10.1016/S0968-8080(14)44797-9 2018.1509114 36. Dellicour S, Aol G, Ouma P, Yan N, Bigogo G, Hamel MJ, et al. Weekly miscarriage rates in a community-based prospective cohort study in 41. Larrea S, Palència L, Borrell C. Medical abortion provision and rural western Kenya. BMJ Open. (2016) 6(4):1–11. doi: 10.1136/bmjopen-2016- quality of care: what can be learned from feminist activists? Health 011088 Care Women Int. (2021):1–20. doi: https://doi.org/10.1080/07399332.2021. 1969573. [Online ahead of print] 37. Zurbriggen R, Vacarezza N, Alonso G, Grosso B, Trpin M. Second trimester medical abortions A feminist socorrista study. 1st (English) Ciudad Autónoma de 42. Drovetta RI. Safe abortion information hotlines: an effective strategy for Buenos Aires: La Parte Maldita (2019). increasing women’s Access to safe abortions in Latin America. Reprod Health Matters. (2015) 23(45):47–57. doi: 10.1016/j.rhm.2015.06.004 38. DePiñeres T, Raifman S, Mora M, Villarreal C, Foster DG, Gerdts C. “I felt the world crash down on me”: women’s Experiences being denied legal 43. Burton J, Peralta G. Redes en torno al aborto clandestino: vínculos de socorristas abortion in Colombia. Reprod Health. (2017) 14(133):1–9. doi: 10.1186/s12978- y sistema de salud en Neuquén, Argentina. Clivajes Rev Ciencias Soc. (2016) 0(6):159. 017-0391-5 Available from: http://revistas.uv.mx/index.php/Clivajes/article/view/2156. Frontiers in Reproductive Health 12 frontiersin.org