Examining the association between motivations for induced abortion and method safety among women in Ghana

ABSTRACT

This article draws on data from 552 women interviewed in the 2007 Ghana Maternal Health Survey to examine the association between motivations for women’s pregnancy terminations and the safety of methods used. Women’s reasons for induced abortions represented their vulnerability types at the critical time of decision making. Different motivations can result in taking various forms of action with the most vulnerable potentially resorting to the most harmful behaviors. Analysis of survey data pointed to spacing/delaying births as the main reason for abortion. Furthermore, women were more likely to terminate pregnancies unsafely if their main motivation for abortion was financial constraints. Especially among rural women, abortions for any reason were more likely associated with safe methods than if for financial reasons. These findings suggest a theme of vulnerability, resulting from poverty, as the motivations for women to resort to harmful abortion methods. Therefore, interventions formulated to reduce instances of unsafe pregnancy terminations should target reducing poverty and capacity building with the aim of economic advancement, in addition to curbing the root of the problem: unintended pregnancy.

KEYWORDS:

• Ghana
• induced abortion
• motivations
• unsafe abortion
• vulnerability

Acknowledgments

Versions of this article were presented at the 2015 Population Association of America (PAA) Meeting in San Diego, CA, as well as the 2015 University of Ghana’s School of Social Sciences Colloquium in Legon, Accra. The authors are grateful to all those who reviewed or provided feedback at the conferences to refine the article.

Notes

^1. Koster-Oyekan (1998) did discuss reasons women in Zambia resort to illegal (unsafe) methods although Zambia has one of the more relaxed abortion laws in sub-Saharan Africa. Also, Osur (2012) investigated the drivers, decision-making processes, and outcomes of unsafe abortion among women in Kenya. However, for these studies, multivariate analyses were not used to test the statistically significant links between the reason for abortion and use of an unsafe method.

^2. In traditional settings, young girls are forbidden to engage in sexual relationships prior to undergoing the puberty rites. Some examples include Dipo rites among the Ga-Dangmes and Bragro among the Akans (Anarfi and Owusu 2011).

^3. The sample of 552 is the weighted total. A total of 564 women (unweighted) had abortions in the 5 years preceding the survey. Six women were removed since they provided no reasons for their abortion, resulting in 558 (unweighted) cases, the final sample used in the article.

^4. This variable was created using principal components analysis and took into account the various items each household owned (e.g., a radio, television, freezer, car, truck, mobile phone, etc.), as well as the type of water and toilet facilities they had access to and the type of homes they resided in. The first factor that loaded was then split into quintiles and labeled poorest, poorer, middle, richer, and richest. For this article the poorer and poorest categories were merged to obtain a poor category.

^5. In using the “fitstat” STATA command to compare two models, the full logistic regression model with and without an interaction term (between place of residence and reason for abortion), the BIC results showed stronger support for the model without the interaction term whose results were later presented. Despite this, indicators showed the model with the interaction also had a good fit statistics. Therefore, since an interaction was identified between “place of residence” and “reason for abortion” we went further to stratify the samples and run regression models by place of residence. The fact that the model without the interaction was deemed as better fitting does not take away from the fact that the model with the interaction was also a good fitting one and did indicate different abortion reasons among urban and rural dwellers.