Methods women use for induced abortion and sources of services: insights from poor urban settlements of Accra, Ghana - BMC Women's Health - BMC Blogs Network

Study setting and sampling

Data used in this paper are from the Willows Impact Evaluation project implemented in Ghana (hereafter, referred to as "WIE-Ghana) between 2017 and 2018. WIE-Ghana collected reproductive health information from women living in two urban-poor neighborhoods of Accra, the capital city of Ghana, West Africa. The two study settings were purposively identified on the basis of having similar demographic, ethnic and socio-economic characteristics. One study setting was in the coastal area while the other was located approximately 18 km inland. The communities surveyed in the coastal area included Osu Klotey, La, Teshie and Nungua while those surveyed in the inland area were La Nkwantanang (Madina), Abogba and Old Ashongman. These communities are disproportionately vulnerable to shocks including floods, congestion, water scarcity, sanitation problems, cholera and other health hazards [24,25,26].

WIE-Ghana employed a three-stage cluster sampling technique to create representative samples from each of the two study settings (Fig. 1). In the first stage, a simple random sampling technique was used to sample 200 census-based geographic clusters (100 in each for the coastal and inland areas). The clusters were generally equally-sized subdivisions of census Enumeration Areas (EAs) obtained from the Ghana Statistical Service consisting of approximately 60–100 households. A complete listing of all eligible women ages 16–44 years in households within the sampled clusters was conducted. In the second stage, approximately 25 households with at least one eligible woman were randomly sampled from each cluster. In the final stage, if the sampled households had more than one eligible woman, one woman was randomly selected for the interview. Our final analytic sample for this paper included 1252 women (representing 30% of WIE-Ghana sample) who reported ever having had an induced abortion. Of these, eleven refused to answer the question on type of method, three did not know, and two had missing information, leaving 1236 women. Three other women who indicated the use of IUD as an abortion method were dropped during data cleaning, leaving 1233 women for analysis.

Fig. 1

Source: Constructed by authors

Sampling and procedure for obtaining final analytic sample.

Full size image

Data collection

Data were collected between January and July, 2018. Data collection was conducted using questionnaires programmed in CommCare version 2.40.1 for use in tablets [27]. Face-to-face interviews were conducted by trained research assistants in either English or the local language after obtaining written consent. The interviews took place in or within the vicinity of the respondents' homes or places of work. The interviews captured household, demographic, socio-economic, and sexual and reproductive health information, including abortion (Additional file 1). Given the sensitive nature of some of the topics, research assistants were extensively trained by the supervising team from the Regional Institute for Population Studies at the University of Ghana, the Harvard T.H. Chan School of Public Health and Ghana Health Service on interviewing techniques, and topics such as abortion law and methods for abortion.

Measures

The questionnaire had a series of questions on abortion for women who reported that they have ever had a pregnancy that ended in miscarriage, induced abortion or stillbirth (Additional file 1). Women who reported that they had ever intentionally terminated a pregnancy or induced an abortion were further asked about the number of pregnancies that ended in induced abortion, when the last induced abortion occurred, and what was done to have their recent pregnancy terminated. To capture information on methods used to induce abortion, women were asked, "What was done to have the pregnancy terminated?". Response categories included: surgical (operation), injection, took pills/medicines/medication abortion, inserted herbs or object in womb, took homemade medicine and other (specify). Women who responded that they used a surgical method were further asked, "What was the name of the surgical procedure used to end pregnancy?". Response categories include: MVA or EVA, Dilation and Curettage (D&C)/Sharp Curettage and specify other. Women who responded that they took pills/medicines or medication abortion were, on the other hand, asked, "What kind of medication did you use to end the pregnancy?". Response categories included: Mifepristone and Misoprostol (Medabon/Mariprist), Misoprostol alone (Cytotec/Misoclear), Oral contraceptive pills (OCP) (Lydia, Microgynon), Other pills and Don't know, and Other (Specify). If women mentioned either the brand-name or the actual medication, these options were recorded under appropriate response options. However, the women were not prompted and if they did not recall the name or brand "Other pills" was selected. These questions collectively formed the basis for categorizing types of abortion methods in this study.

For source of service for the last induced abortion, women were asked, "From whom/where did you receive induced abortion services the last time?". There were 21 possible response categories that were ultimately compiled into the following: government hospital/polyclinic, government health center/clinic, government health post/CHPS, private hospital/clinic, private doctor, pharmacy/chemical/drug stores, friends/relative/partner, drug peddler and other (specify).

Other explanatory variables included: age of woman (16–19 years, 20–24 years, 25–29 years, 30–34 years, 35–39 years and 40–44 years); marital status (never in union, currently in union and formerly in union); level of education (no formal education, completed primary, completed middle school/junior high school (JHS), completed secondary and higher); religion (Moslem, Catholic, Anglican/Methodist/Presbyterian, Pentecostal, other Christian and other/no religion); ethnicity (Ga-Dangme, Akan, Ewe, other Ghanaian and non-Ghanaian); household wealth index (poorest, poorer, middle, richer and richest); knowledge of abortion law (knowledge of at least one legal condition to induce abortion and no knowledge of any legal condition in Ghana); and when last induced abortion was conducted (induced abortion more than 3 years ago and induced abortion recently within 3 years).

Analysis

Data from CommCare was exported to STATA version 14.2 for analysis [28]. Statistical tests with p-values (p < 0.05) were considered significant. We generated descriptive statistics for socio-economic, demographic and abortion-related attributes, including sources of services. We then estimated a multinomial logistic regression model to examine the factors associated with the type of method used for the last induced abortion. For the regression modelling, we classified the types of abortion methods into three broad categories: Surgical (D&C and EVA/MVA), Medication (misoprostol alone and misoprostol and mifepristone combination) and Non-medical methods (insertion of objects, and homemade herbs/concoctions/objects). Medication abortion was used as the base outcome for the analysis. Women who reported using other pills (approximately 12%) were dropped from the analysis since the use of specific medication abortion pills may be significantly different from the use of OCP or other non-recommended pharmaceutical tablets. However, it is also possible that women used approved MA pills but did not remember the brand name. A sensitivity analysis was therefore conducted by including OCP and other non-recommended pharmaceutical tablets to the medication category. We also dropped women who reported using injection (about 3%) from the regression analysis. Injection abortion (methotrexate alone or combination of methotrexate and misoprostol) uses similar regimens as medication abortion. However, it could not be classified under MA because the medium of use is significantly different and is often used by doctors for rare abortion cases such as ectopic pregnancy and for women who are allergic to mifepristone [29, 30]. The explanatory variables included in the regression analysis were selected based on the existing literature and their likely role in influencing choice of abortion methods [20, 31, 32]. The results are presented as relative risk ratios (RRR) with 95% confidence intervals (CI).

Ethical considerations

Ethical approvals were obtained from Ghana Health Service Ethical Review Committee (GHS-ERC #: 005/08/2017), University of Ghana Ethics Committee for the Humanities (ECH #: 020/17–18) and the Institutional Review Board (IRB) of Harvard T.H. Chan School of Public Health before implementing the study. Written informed consent was obtained for all respondents before participation in the study. Before obtaining the written informed consent, an information sheet which contained a summary of the study and all ethical issues related to the study was given to the participant to read. For participants with no formal education, the research assistants read and explained the information sheet to them. Since data were collected from women aged 16 to 44 years, no parental/guardian consent was required. In Ghana, minors are persons under the age of 16 years.

Adblock test (Why?)