Framework and period of study
This study was conducted in selected government hospitals in Arba Minch and Sodo towns, Southern Ethiopia, from April 15 to June 15, 2021. Arba Minch and Wolayita Sodo are towns located in the region of Southern Nations, Nationalities and Peoples (SNNPR) of Ethiopia. In the cities of Arba Minch and Wolayita Sodo, there are four hospitals, but the study was conducted in two selected public hospitals (Arba Minch General Hospital and Wolayita Sodo University Teaching & Referral Hospital). The total population of Arba Minch town 112,724 of which 50.2% are women while Wolayita Sodo 250,521 of which 79,871 (52%) are male and 73,650 (48%) are female. There are four public health institutions (one general hospital, one primary hospital and two health centers), thirty-two medium and high private clinics, one Marie Stopes clinic, twelve pharmacies and two community pharmacies in the town of Arba Minch. . According to the Medical Director’s report for fiscal year 2020, Arba Minch General Hospital provides 100 comprehensive abortion care services, 195 antenatal care services and 180 postnatal care services per month on average. There are two hospitals (one public and one private), three health centers, one Marie Stopes clinic and thirty medium and higher private clinics in the city of Wolayita Sodo. . According to the Medical Director’s Report for FY2020, Wolayita Sodo Teaching and Referral Hospital provides 150 comprehensive abortion care services, 250 antenatal care services and 210 postnatal care services per month on average. per month on average.
An institutional unmatched case-control study was conducted.
All women who received maternal health care services in public hospitals in Arba Minch and Wolayita Sodo town.
All women who received comprehensive abortion care and those who visited Maternal and Child Health (MCH) units for antenatal or postnatal care services in selected public hospitals in Arba Minch town and Wolayita Sodo during the data collection period.
Selection of cases and controls
Case : women who received induced abortion care services at selected government hospitals during the data collection period or who received postabortion care services after being presented to the selected government hospital with attempted abortion induced abortion.
Controls: women who have had at least one history of pregnancy in the past 12 months who have come for maternal health care services (antenatal or postnatal care) at selected government hospitals and who have never had a history of induced abortion.
Women presented to selected public hospitals for spontaneous abortion care after being diagnosed by a physician and unable to communicate or seriously ill until the end of the data collection period were excluded.
Determination of sample size
For a case-control study, the sample size for this study was determined using the Stat Cal application of the Epi-Info software version 7 and two population proportion formulas. The following assumptions were taken into account: 95% confidence level, 80% power, case/control ratio 1:3 and percentage of cases exposed 8.2% and percentage of controls exposed 21%. The percentages of cases and controls of the exposure variable were taken from a study conducted in Addis Ababa which found that the most determining variables for induced abortion were monthly income . Based on the above assumptions, the estimated minimum sample size for this study was 94 cases and 281 controls. After accounting for a 10% nonresponse rate, the final sample size for this study was 103 cases and 310 controls.
Sampling technique and procedures
In this study, Arba Minch General Hospital and Wolayita Sodo University Teaching and Referral Hospital were included. Reviewing the two-month report of the previous year during a data collection period, at Arba Minch General Hospital and Wolayita Sodo University Teaching and Referral Hospital, on average 40 and 80 women received induced abortion care and 295 and 340 women received maternal health care services respectively. The proportional allocation method was used to include 413 women in the sample. Cases were selected using a consecutive sampling technique until the required sample size was reached. Controls were selected using a systematic random sampling technique. The lottery method was used to select the first control, then every other control was queried until the required sample size was reached (Fig. 1).
Induced abortion was the dependent variable in this study. The independent variables were socio-demographic and economic factors (age, residence, marital status, professional status, level of education, monthly income), factors related to reproductive and maternal health (age at first marriage, age at first intercourse sex, multiple sexual partners, number of pregnancies, number of living children, pregnancy status) and factors related to contraception (use and knowledge).
Operational definition and measurements
Induced abortion: voluntary termination of pregnancy, by any means or by any person other than spontaneous . Contraceptive use: the percentage of women who used a modern contraceptive method before the current pregnancy . Knowledge of contraception: based on bloom cut of point, women who answered 75–100%, 50–74% and
Data collection tools and procedures
Data were collected using a standardized, pretested interviewer-administered questionnaire, which was adapted from previous related studies. [22, 29, 31, 33]. The questionnaires contain three sections: sociodemographic characteristics, factors related to reproductive and maternal health, and factors related to contraception (Supplemental File 1). Six midwives with a Bachelor of Science degree were recruited to collect the data. In addition to this, two supervisors with master’s degrees were recruited for supervisory activities. The principal investigator gave three days of theoretical and practical training to data collectors and supervisors on data collection tools, interview techniques, confidentiality of information, the objective and relevance of the study. The Kobo collect app version 3.1 was installed on the data collector’s Android device, and the blank form was downloaded from the Kobo toolbox server. Two weeks before the actual data collection, the tool was pre-tested on 20 women (5 cases and 15 controls) in the primary hospitals of Chencha and Bodti. After obtaining informed consent from the respondent, face-to-face interview and chart review were used to collect the data. Supervisors communicated regularly with data collectors to ensure that the data collection procedure was followed. To avoid misclassification, questionnaires coded cases and controls (1 = cases, 0 = controls) and women seeking maternal health care services were asked about their history of previous induced abortions. and their medical records were reviewed to ensure they were genuine witnesses. Finally, on a weekly basis, the data collectors sent the completed questionnaire forms to the Kobo Toolbox server.
Data quality management
The questionnaire initially prepared in English and translated into Amharic language, then translated back into English by the expert to check consistency. Data collectors and supervisors were trained over two days to familiarize themselves with all types of data, tools, and data collection methods and goals, plus a day of hands-on Kobo Collect sessions. A pre-test was carried out on 5% of the participants at Chencha and Bodti Primary Hospital, and any ambiguities, as well as missed points, were added in the final version of the questionnaire. Supervisors checked the completed questionnaires for key content before uploading them from the Android mobile phone to the Kobo Toolbox server to ensure data quality. All data was collected onsite using Android mobile devices and uploaded weekly to the Kobo server using Kobo collect version 3.1. The principal researcher also regularly checked the consistency and completeness of the files sent by each data collector.
Data processing and analysis
Data from the Kobo server was downloaded as an Excel file and exported to SPSS V.25 for cleaning, coding, completeness and accuracy, and then to Stata V.14 software for further analysis. A descriptive analysis was performed to describe the relevant characteristics of the study participants. After that, simple frequencies, percentages and summary measures were calculated. Bivariate and multivariate analyzes were used to assess the association between each independent variable and the outcome variable using binary logistic regression. The goodness of fit was checked by the Hosmer and Lemeshow test. Variables with a 95% confidence interval and a P value