The WHO previously conducted an international, multicentre study examining obstetric outcomes in women by FGM status. The cohort contained women without FGM, as well as women with FGM, categorized by the WHO classification system. Previous papers have reported on the risks of different obstetric outcomes for both the woman and the neonate, as well as estimated costs to the health system [3, 18]. In this sub analysis, we focus on the association between type of episiotomy and maternal outcomes in women with FGM Type 3.
Women who presented for singleton delivery at 28 obstetric centres in Burkina Faso, Ghana, Kenya, Nigeria, Senegal and Sudan between November 2001 and March 2003 were screened for study eligibility. Women with multiple gestations, or presenting for elective caesarean delivery or in advanced labour (unable to complete a pelvic exam prior to delivery) were excluded from the study, along with women who were unwilling or unable to give informed consent. Women and their infants were then followed until time of maternal discharge from the hospital. All participants provided informed consent prior to enrolment. Institutional review boards at all participating hospitals and the World Health Organization (WHO) Secretariat Committee on Research Involving Human Subjects gave ethics approval.
We used descriptive statistics and bivariate measures of association to describe the study population and the population of women by type of FGM. Bivariate and multivariable logistic regression models were used to examine the association of type of episiotomy and maternal outcomes (anal sphincter tears, intrapartum blood loss requiring intervention, and postpartum haemorrhage) among women with type 3 FGM.
We included only women having a vaginal delivery; this included normal vaginal delivery, assisted operative delivery (forceps or vacuum) and assisted breech delivery. Women giving birth by caesarean were excluded. Participants had an antepartum examination of the external genitalia, by a trained study midwife, to determine whether or not they had undergone FGM. If they had FGM, the type was categorized according to the WHO classification system (Table 1). The pelvic exam also included an assessment of outlet obstruction: the dimension of the introitus was evaluated by fingerbreadths. For the analysis of the association between episiotomy and maternal health outcomes, we limited our sample to women who were living with FGM Type 3 with data on episiotomy status.
Our key independent variable for analysis was episiotomy type. If an episiotomy was performed, the study investigator recorded the type. Episiotomy was classified as follows: no episiotomy, anterior (deinfibulation), posterior lateral, and anterior with simultaneous posterior lateral episiotomy. The dimension of the introitus was assessed by finger breadths and coded as one, two, three, or more than three fingerbreadths. For the multivariable models, we included the following demographic characteristics of the woman: her age, place of residence (urban/rural), socioeconomic status (low, medium, high) and level of education.
Three maternal health outcomes served as our dependent variables—anal sphincter tears, intrapartum blood loss requiring an intervention, and post partum haemorrhage. Degree of tear was included as a dichotomous variable, with comparing more extensive lacerations (anal sphincter tears, also called 3rd and 4th degree obstetric tears) to no tear or 1st or 2nd degree tears. Intrapartum blood loss was dichotomized comparing women who required an intervention (e.g uterotonics, dilation and curettage, transfusion) to those who did not. Postpartum haemorrhage, blood loss occurring within 24 h of delivery, was coded as a binary variable using the standard threshold of exceeding 500 ml .
We examined the association between episiotomy type among women living with FGM Type 3 and each of the following outcomes—anal sphincter tears, intrapartum bleeding requiring intervention, and postpartum haemorrhage. Each type of episiotomy was compared with no episiotomy. Theoretically relevant model covariates included parity, pelvic introitus width, age, socioeconomic status, and education level. Initially we planned to enter the covariates in blocks—obstetric factors, sociodemographic factors and then the combination for fully adjusted models. However, the adjustment variables had minimal impact so we present only the unadjusted and then fully adjusted models. Odds ratios (OR) with 95 % confidence intervals were assessed for each of the three maternal outcomes. As the data were clustered in the 28 centres, robust standard errors were used to adjust for this clustering .