Prevalence of Induced Abortion in Iran: A Comparison of Two Indirect Estimation Techniques.

Abstract

Surveys that use direct questions to ascertain women's history of induced abortion tend to underestimate abortion prevalence, especially in such contexts as Iran where the procedure is legally restricted and highly stigmatized. No previous study has compared two indirect techniques for estimating abortion prevalence. A sample of 708 married women were recruited from one public hospital in Tehran between August and December 2013. Participants completed a survey, which included induced abortion estimation using the randomized response technique (RRT) and the unmatched count technique (UCT), as well as questions about demographic characteristics, trust in direct questions about abortion, and comprehensibility of and trust in RRT and UCT. Prevalence of induced abortion was calculated for each technique. Spearman correlation was used to evaluate whether comprehensibility of and trust in estimation methods were associated with women's age and education. The prevalence of induced abortion was estimated to be 14% using RRT and 12% using UCT; the estimates were not significantly different. Ninety-one percent of women reported that UCT was very easy to comprehend; the proportion for RRT was 78%. Sixty-three percent of women reported completely trusting in the confidentiality of UCT; the proportion for RRT was 50%. Age was inversely associated with comprehensibility for UCT (correlation coefficient, -0.13), and with trust for both RRT and UCT (-0.12 and -0.08, respectively); education was directly associated with trust for both methods (0.24 and 0.22). Of the two indirect methods, UCT may be simpler and more dependable for the estimation of induced abortion prevalence in low-literacy, abortion-restricted settings.