The data in this study come from a large-scale longitudinal study that aimed to investigate the predictors of denial of abortion services and examining the consequences of unwanted pregnancy for women and their children.
Study site
This study enrolled women from the period between April, 2019 to December 2020 from 22 health facilities across Nepal, including at least two from each province. We selected both public and private/NGO facilities based on volume of abortion patients, geographical distribution, and diversity of patients from a list of certified abortion facilities that provided 60 or more abortions per year in 2016–2017. A brief description of the study methods are below and more detailed information is described elsewhere (17).
Study population
The participants consisted of women between the ages of 15-49 years who were pregnant and seeking abortion care and living in Nepal.
Data collection
Between April 2019 and December 2020 (except for a few months when recruitment was paused due to Covid-19 Pandemic), we interviewed 1,831 women who sought abortions from one of 22 health facilities. In the first month of recruitment (April 2019), all women presenting for abortion care were eligible for the study. Starting in the second month, we limited recruitment to only those women who were presenting at or beyond 10 weeks’ gestation or who did not know their gestational age to collect a sufficient sample of women denied abortion services. We administered a short questionnaire before participants knew whether they would receive an abortion or not, in which we documented each participant’s gestational age at the time of the visit, regardless of their eligibility status for legal abortion. In case of non-eligibility, the reasons for denial were documented. Follow-up interviews were conducted at the participant’s home or elsewhere 6 weeks after the clinic visit, and then every 6 months for three years. Before starting each interview, written consent was obtained from the participants. After each interview, including the baseline survey, participants received financial compensation equivalent to $4. The interviewers asked all questions in Nepali or in the local language where applicable (Maithili/Bhojpuri). Survey answers were entered on tablets using Qualtrics and the data were synced. Interviews were conducted in person, either in a private setting at the home of the participants or another location, if preferred, and interviews were conducted without the presence of other adults in the household. We asked women at each follow up interview whether they were using a family planning method or not and, if not, the reasons for not using one.
In this paper, we analyzed data on postabortion contraceptive use among women who reported at the 6-week interview that they had an abortion and completed 7 rounds follow up interviews. Participants who reported being sterilized postabortion were dropped from the analysis.
Data Analysis
First, we described the socio-demographics of the sample, testing for differences between those who initiated a modern method of family planning by 6 weeks postabortion and those who did not using chi2 tests. Next, we explored predictors of adopting a modern method of family planning by 6 weeks, and, among those who took up a method at 6 weeks, predictors of continuation of use at each time point (6 months, 12 months,18 months, 24 months, 30 months and 36 months) using logistic regression models.
We ran parametric models of our survival-time data that is interval censored, since data on our main outcome, current use of family planning, was collected at each survey. Only women who reported that they were using family planning at 6 weeks postabortion were included in our model. A variable was created to indicate the time since adopting family planning at the last data collection time point that a woman reported using a method. A second variable was created indicating the time since adopting family planning at the first data collection time point that the woman reported that she was not using a method. After testing for model fit, we used a Weibull distribution in our models. Covariates included woman’s age, parity, ethnicity, years of schooling, living with husband, household income, autonomy level, and desire for more children. The level of women’s autonomy was determined by assessing variables related to decision making power in household and health care matters, women’s ability to travel outside of house independently, such as going to market, hospital or visiting outside of village. A composite index was created and the level of autonomy was categorized into three levels - low, medium and high. All women who adopted a method were married, so marital status was not included in any of our models.
Ethical approval
Ethical approval from the University of California, San Francisco (UCSF) institutional review board (IRB) in the United States and the Nepal Health Research Council (NHRC) in Nepal were obtained. Written informed consent was also obtained from all participants (age 18 and older). Before conducting interviews with minors (aged 15 to 17 years), written consent from the parents and the woman's assent was obtained.