Menu
Menu
Menu
Menu
In Ethiopia, abortions are legal for minors and for rape, incest, foetal impairment or maternal disability. Knowledge of abortion legality and availability is low, and little effort has been made to disseminate this information for fear of invoking anti-abortion sentiment; instead, systems rely on health providers as information gatekeepers. This study explores how exposure to and interaction with family planning service delivery environment, specifically (1) availability of contraceptive and facility-based abortion services within 5 km of one’s residence and (2) contact with a health provider in the past 12 months, relate to women’s knowledge of the legality of accessing abortion services and of where to access facility-based abortion services. We used data from a nationally representative sample of 8719 women in Ethiopia and a linked health facility survey of 799 health facilities. Our outcome of interest was a categorical variable indicating if a woman had (1) knowledge of at least one legal ground for abortion, (2) knowledge of where to access abortion services, (3) knowledge of both or (4) knowledge of neither. We conducted multilevel, multinomial logistic regressions, stratified by residence. Approximately 60% of women had no knowledge of either a legal ground for abortion or a place to access services. Women who visited a health provider or who were visited by a health worker in the past 12 months were significantly more likely to know about abortion legality and availability. There were no differences based on whether women lived within 5 km of a facility that offered contraception and abortion services. We find that health workers are likely valuable sources of information; however, progress to disseminate information may be slowed if it relies on uptake of services and limited outreach. Efforts to train providers on legality and availability are critical, as is additional research on knowledge dissemination pathways.
This cross-sectional study uses data from Performance Monitoring for Action (PMA)-Ethiopia, a 5-year (2019–2023) research partnership between Addis Ababa University (AAU), the Ethiopian FMoH and the Johns Hopkins Bloomberg School of Public Health (JHSPH). PMA-Ethiopia generates cross-sectional and longitudinal data on a range of reproductive, maternal and newborn health indicators (Zimmerman et al., 2020). Data are collected from women, households and service delivery points (SDPs) that offer maternal and reproductive health services to inform policies and priorities at national and regional levels. This analysis uses two data sources from PMA-Ethiopia, the cross-sectional, nationally representative sample of women aged 15–49 years (Addis Ababa University School of Public Health and The Bill & Melinda Gates Institute for Population and Reproductive Health at The Johns Hopkins Bloomberg School of Public Health, 2020) and data from the cross-sectional SDP survey (Addis Ababa University School of Public Health; and the Bill & Melinda Gates Institute for Population and Reproductive Health at the Johns Hopkins Bloomberg School of Public Health, 2019) collected between October and November 2019. National cross-sectional data included both household and female surveys. Enumeration areas (EAs), groupings of approximately 200 households, were selected with probability proportional to size within regional and residential strata, using the national census as the sampling frame. Following a census and listing, 35 households were randomly selected within each EA; all households within a selected EA were eligible for the household survey. Women were eligible to participate if they were aged between 15 and 49 years, slept in the selected household the night prior or who were usual members of the household and were willing and able to provide informed consent. The SDP survey (herein referred to as the ‘facility survey’) was conducted among public and private facilities that served the selected EAs. Public facilities were included if they were a health post, health centre or hospital. A maximum of three private SDPs, including health clinics and health centres that offered maternal health services and pharmacies or drug vendors that provided reproductive health commodities, within the kebele (the smallest administrative unit in Ethiopia) in which the EA was located were randomly selected for interview. Data about facility readiness to offer essential health services, including provision of abortion services, were collected by trained resident interviewers using mobile phones equipped with Open Data Kit Software (Open Data Kit, Seattle, WA, USA). Study procedures, detailed elsewhere including additional information on consent procedures and ethical concerns, were approved by the Institutional Review Boards at Johns Hopkins School of Public Health and AAU (Zimmerman et al., 2020). Our primary dependent variable was a categorical measure assessing women’s knowledge of legal grounds for abortion and knowledge of service availability. We defined respondents as having knowledge of legal grounds for abortion using two conditions: if (1) a woman said yes to the question ‘Do you know if there is a law on abortion in Ethiopia?’ and (2) if she spontaneously reported one or more of the conditions under which abortion is legal in Ethiopia to the question ‘Under which circumstances, it is legal to have an abortion in Ethiopia?’. Interviewers were trained to select one or more of the options that included rape, foetal impairment, risk to the life of the mother or foetus or if the mother is unable to raise the child due to physical or mental infirmity. We chose to define ‘knowledge’ based on identifying a single exemption, rather than all exemptions, as the percentage of women with knowledge of all legal exemptions based on previous research were less than 5% of women (Sheehy et al., 2021). Abortion is also legal in Ethiopia for women under age of 18 years; however, due to a survey programming error, this legal ground was not assessed in the questionnaire. Women could also respond that they knew that there was a law about abortion and that abortion was not legal under any circumstances; these women were not categorized as having knowledge of the legal grounds for abortion. We measured women’s knowledge of where to access services through a single item, ‘Do you know where a woman can access facility-based abortion services in the community where you live?’ (Yes/no). Based on responses to these three questions, women were classified into one of the four categories: (1) no knowledge of either legality or service availability, (2) knowledge of legal grounds only, (3) knowledge of service availability only or (4) knowledge of legality and service availability. Our key independent variables measured a woman’s exposure to and use of the contraceptive service environment, which we examined via three measures captured across the female and facility surveys. First, using the female survey, we assessed two measures: (1) whether the respondent had been visited by a health provider who discussed FP1 in the last 12 months (yes/no) and (2) whether the respondent had visited a health facility in the past 12 months and spoken to someone about FP (categorical: no visit, visit but no FP discussion and visit and FP discussion). Second, using data from the facility survey, we assessed availability of facility-based contraceptive and abortion services for each woman in our sample. Specifically, we identified all facilities within a 5 km radius (geodetic distance) of the woman’s residence and defined a three-level categorical variable indicating availability of services: (1) no contraceptive or abortion services within 5 km, (2) contraceptive services only within 5 km and (3) contraceptive and abortion services available within 5 km. No facilities provided abortion without contraceptive services. Finally, we explored the role of women’s contraceptive use status, which we defined as either using or not using a modern contraceptive method at the time of the survey. Due to significant differences in availability of services and wealth distributions between urban and rural areas (Table 1), all analyses were stratified by residence. Analyses adjusted for a number of additional socio-demographic characteristics and potential confounders that we hypothesized to be associated with exposure to the FP service environment and knowledge of abortion legality and services, including women’s age (5-year age groups), marital status (married/in-union, not in union), wealth quintiles, education (none, primary, secondary or above), parity (0, 1–2, 3–4 and 5+ children) and region. Given high correlation between age and parity (ρ > 0.65 across residence), age was excluded in final models. Wealth quintiles were heavily skewed in urban and rural areas when stratified (Table 1). For subsequent analyses, we thus created wealth quintiles specific to urban/rural women to assess the effect of wealth separately within each residence. Sample characteristics of women participating in the PMA-Ethiopia 2019 cross-sectional survey Of 8976 eligible women who slept in the house the night before (de facto residents), 8839 completed the interview, with a response rate of 98.4%. For this analysis, we dropped 120 women who were missing outcome data (described further later) for a total population of 8719 women (unweighted: n = 3738 urban and n = 4981 rural; weighted: n = 2854 urban and n = 5870 rural). Of 815 SDPs identified for the survey, 799 completed the interview, with a response rate of 98.0%. All observations were included. We used design-based analysis to account for survey weighting due to differential probability of selection and clustering of responses within EAs (Heeringa et al., 2017). Exploratory analyses assessed the distributions of the key outcome and predictor variables and women’s socio-demographic characteristics. We then assessed outcomes by urban and rural strata separately and tested for differences using Pearson chi-square statistics with the Rao and Scott second-order correction. Finally, we used stratified multinomial multilevel regression models, with EA as the second level, to estimate the relative risk ratios (RRRs) of having knowledge of legal grounds only, service availability only or knowledge of both, relative to having no knowledge of either. Models 1 and 3 (rural and urban, respectively) included only covariates that related to exposure to the FP environment, specifically distance to services, visit to a health facility, visits from a health worker, modern contraceptive method use and region. Models 2 and 4 (rural and urban, respectively) additionally adjusted for relevant individual-level socio-demographic characteristics. We tested for shared versus separate random effects using Akaike Information Criterion and Bayesian Information Criterion and treated random effects as separate but correlated. All analyses were conducted using Stata 16.1 (College Station, TX, USA).
N/A
