Determinants of change in unintended pregnancy in Ethiopia using the 2005 and 2016 EDHS: non-linear multivariable decomposition analysis

BACKGROUND: Unintended pregnancy is a pregnancy either mistimed or unwanted. The main consequence of unintended pregnancy is inducing abortion. In Ethiopia, more than half of unintended pregnancies end up in abortion. OBJECTIVE: This study aims to measure the change in unintended pregnancy among women of reproductive age between survey years 2005 and 2016 and to identify the socio-demographic factors that most significantly contributed to the change. METHODS: Data from the two most recent Ethiopian Demographic and Health Surveys (EDHS) were analyzed. We quantified the contribution of socio-demographic factors in the change of unintended pregnancy, using Oaxaca-Blinder decomposition for non-linear regression models by applying the STATA command ‘mvdcmp’. RESULT: Unintended pregnancy decreased from 37% in 2005 to 27% in 2016 in Ethiopia. Both changes in population characteristics and coefficient were the contributing elements to the observed change in unintended pregnancy. Among population characteristics factors, being a partial decision-maker and being a slum in the Somali region contributed 10 and 14% to the change of unintended pregnancy between the 2005 and 2016. Of the coefficient factors, knowledge of modern family planning, being a partial decision-maker, media exposure, distance to health facilities, and health facility visits contributed to the change by 93, 43, 17, and 10% respectively. CONCLUSION: The majority of the change in unintended pregnancy from 2005 to 2016 survey was due to differences in coefficients (85%). The principal contributing factors to the change of unintended pregnancy were FP knowledge, decision making, media exposure and health facility visits. Therefore, an interventional plan will be efficient, better, and more effective if focused on the larger contributing factors.