Frequency, risk factors, and complications of induced abortion in ten districts of Madagascar: results from a cross-sectional household survey.

Background: Madagascar has restrictive abortion laws with no explicit exception to preserve the woman's life. This study aimed to estimate the incidence of abortion in the country and examine the methods, consequences, and risk factors of these abortions.

Methods: We interviewed 3179 women between September 2015 and April 2016. Women were selected from rural and urban areas of ten districts via a multistage, stratified cluster sampling survey and asked about any induced abortions within the previous 10 years. Analyses used survey weighted estimation procedures. Quasi-Poisson regression was used to estimate the incidence rate of abortions. Logistic regression models with random effects to account for the clustered sampling design were used to estimate the risk of abortion complications by abortion method, provider, and month of pregnancy, and to describe risk factors of induced abortion.

Results: For 2005-2016, we estimated an incidence rate of 18.2 abortions (95% CI 14.4-23.0) per 1000 person-years among sexually active women (aged 18-49 at the time of interview). Applying a multiplier of two as used by the World Health Organization for abortion surveys suggests a true rate of 36.4 per 1000 person-year of exposure. The majority of abortions involved invasive methods such as manual or sharp curettage or insertion of objects into the genital tract. Signs of potential infection followed 29.1% (21.8-37.7%) of abortions. However, the odds of potential infection and of seeking care after abortion did not differ significantly between women who used misoprostol alone and those who used other methods. The odds of experiencing abortion were significantly higher among women who had ever used contraceptive methods compared to those who had not. However, the proportion of women with a history of abortion was significantly lower in rural districts where contraception was available from community health workers than where it was not.

Conclusions: Incidence estimates from Madagascar are lower than those from other African settings, but similar to continent-wide estimates when accounting for underreporting. The finding that the majority of abortions involved invasive procedures suggests a need for strengthening information, education and communications programs on preventing or managing unintended pregnancies.