Study Setting and Sites
This study included data from four emergency obstetric programs from three sub-Saharan African countries. Cesarean sections were performed at MSF-supported or MSF hospitals. MSF-supported was defined as a government hospital which received financial and human resources support from MSF. MSF hospitals were funded, built, and staffed solely by MSF. All hospitals received patients from various clinics that provided basic emergency obstetric care. In Masisi, DRC, Masisi General Hospital served a catchement of 337,000 persons and 29 clinics. In Lubutu, DRC, Lubutu General Hospital served a catchement of 103,000 persons and 18 clinics. In Kabezi, Burundi, the Centre d'Urgence Gynéco-Obstétrique, served a catchement of 581,000 persons and 55 clinics. In Bo, Sierra Leone, Gondama Referral Centre, served a catchement of 302,000 persons and 24 clinics.
All care was provided free of charge and administered with the permission of the local health authorities. All sites had electricity, clean water, sterilization equipment, an operating theater, post-operative surgical wards, post-anesthesia care units, blood banks, anesthetics, analgesics, and antibiotics. Cesarean sections were performed by expatriate and local general surgeons, obstetricians, and general doctors with surgical skills. Anesthesia was provided by nurse anesthetists and anesthesiologists. In all four programs, MSF was the only provider of comprehensive EmOc in the catchment area although other health facilities performed vaginal deliveries and some Cesarean sections.
Women undergoing Cesarean section from August 1 2010 to January 31 2011 were included in this study. All Cesarean sections were performed according to peri-operative protocols established by MSF.
Each site had 1–3 physicians performing Cesarean sections during the study period. Each physician was trained directly by either a physician or a midwife with experience in providing CS. Definitions of each indication were included in a study guideline which was reviewed with performing physicians.
Data was prospectively collected by trained data collectors using a standardized paper form and then entered into an electronic database. Regular consistency checks ensured that any missing data was tracked by the data collector and entered into the data. Baseline characteristics on age, program site, gravida, parity, and indication for cesarean section were collected. Prior to the study, various classifications of Cesarean section indications were discussed with a panel of experts with extensive experience working in sub-Saharan Africa including midwives, obstetricians and the MSF sexual health advisor. The initial indications, determined by consensus, were classified as follows: prolonged labor, pre-eclampsia/eclampsia, uterine rupture, antepartum hemorrhage (placenta previa or placental abruption), previous cesarean section, cephalopelvic disproportion, breech, malpresentation (transverse lie, face presentation, arm prolapse), multiple gestation (twins/triplets), cord prolapse, genital herpes, fetal distress, and ‘other’. For purposes of the analysis, prolonged labor and cephalopelvic disproportion were re-categorized as obstructed labor and malpresentation and breech as poor presentation. Indications for Cesarean section were classified by the physician on duty. Certain indications (such as uterine rupture) were confirmed after the procedure. Outcome data on maternal and early neonatal death were documented. Maternal death was defined as death of the mother during hospitalization. Early neonatal death was defined as death of the infant within 7 days of delivery. Cesarean section rates were calculated by dividing the total number of Cesarean section by the total number of expected deliveries. The expected number of deliveries was obtained by multiplying the crude birth rate of the country by the catchment population by the study period.
Baseline characteristics were described using medians and interquartile ranges (IQRs) for continuous variables and counts and percentages for categorical data. Logistic regression was used to model determinants of maternal and early neonatal mortality. Variables considered in the analysis included age, parity, program site, and indication for Cesarean section. Factors with a P<0.1 on univariate analysis were included in a multivariate model. All tests and confidence intervals were considered to be significant at a P≤0.05. All analyses were performed using STATA 10 (College Station, TX, USA).
Ethical approval was given by the independent Ethics Review Board of MSF. Individual patient consent was not separately obtained for the study or for patient information to be stored in the hospital database because this study was based on routinely collected data and all patient identifiers were removed prior to capture into the database and analysis.