The aim of this project is to study the safety, efficacy and effectiveness of medical treatment of incomplete abortion provided to women by physicians or midwives in Kisumu, Kenya.

In Kenya, Post Abortion Care (PAC), provided by physicians, nurse-midwives and clinical officers, has been integrated at private reproductive health facilities since 1998. Misoprostol as treatment of incomplete abortion was launched I Nyanza Province in April, 2012. The involvement of midwives in medical (Misoprostol) treatment of incomplete abortion has, however, not been systematically evaluated. There is a need to determine whether midwives and physician can perform medical treatment of incomplete abortion equally safe and effective in Kenya. The results will thus provide evidence-based information that can contribute to the development of strategies to increase women's access to Post Abortion Care in Kenya as well as in other low-income contexts.

The project will take place at Jaramogi Oginga Odinga Teaching and Referral Hospital (JOOTRH) and Kisumu East District Hospital (KEDH), in Kisumu County of Nyanza Province in Western Kenya. The project will be implemented within the Department of Obstetrics and Gynaecology at the two facilities. Altogether the two facilities admit around 60 women per month with incomplete abortion. Midwives and physicians included in the project will undergo a standardized training prior to the study. This will include basic knowledge on treatment of incomplete abortion and post abortion contraceptive counselling as well as interview technique. The training will be both theoretical and practical and on-going during the project in order to cover all / new staff.

The sample size has been calculated with the objective of showing two-sided equivalence, assuming that the overall complication rate could be as high as 4-5 percent and would apply to both types of providers. To demonstrate two-sided equivalence within a margin of 4-5 percent with 80% power and using a 95% CI (α=0.05), about 816 women would need to be recruited. In addition, about 10 percent could be expected lost to follow up and will be adjusted for. Thus, a total of 880 women will be recruited. The randomization will be conducted in blocks of 8 and will vary randomly. A computer random number generator will be used to generate a list of codes from 1 to 880 and each code is linked to one of the two groups - A= PAC care and misoprostol by midwife and B= PAC care and misoprostol by physician. The list will be used while sequentially numbered, opaque, sealed envelopes will be prepared by the research team. Each envelope contains a study protocol for the individual woman. At the time of allocation, the research assistant at the clinic will pick the envelope with the lowest number, write the participant's name and personal registration number on it, and then open it. Process evaluation will be conducted by intermittent check-ups in order to assure that the intervention procedures are performed correctly and that they follow the protocol. The check-ups will be made by a researcher and include both a review of the completed protocols and repetition and education of the physicians and midwives involved in the study.

All eligible women who consent to participation will undergo a clinical assessment by the provider they have been randomized to. The clinical assessment includes (i) medical history taking , Last Menstrual Period (LMP), Obstetric and Gynaecological history, contraceptive history, symptoms) (ii) General physical examination (pulse, blood pressure and temperature); (iii) Pelvic examination that include examination of size of the uterus (External genitalia, speculum examination, Bimanual examination). Before discharge all women will be given detailed information regarding bleeding and pain expected following treatment as well as abnormal symptoms (fever, and foul smelling vaginal discharge) and the importance of seeking care if such symptoms occur. All women will be followed-up after 7 - 10 days.

All analyses will be by Intention to Treat (ITT). Background characteristics for the two study groups and categorical outcomes will be presented using descriptive statistics. Differences between groups will be analysed using relative risks (95% CI). P-values equal to or lower than 0.05 will be considered statistically significant.

Inclusion Criteria:

• Women presenting with vaginal bleeding in the first trimester of pregnancy and diagnosed to have incomplete abortion

Exclusion Criteria:

• Women with unstable hemodynamic status and shock, signs of sepsis.