Data source and sampling techniques

We used the dataset of women aged 15–49 years included in the EDHS 2016. Ethiopia is sub divided in to eleven geographical regions which are sub divided into kebeles (the smallest administrative unit). Each region was stratified in to urban and rural areas. The 2016 EDHS sample is stratified and was selected in two stages.

In the first stage, 645 enumeration areas were randomly selected: 202 in urban areas and 443 in rural areas. In the second stage, a fixed number of 28 households per cluster were selected randomly for each enumeration areas. The 18,060 households were randomly selected and 16,650 households were eligible and interviewed. Additional information about the methodology of EDHS 2016 can be accessed in the published report of the main findings of the survey [12].

The main focus of this study was women aged 15–49 years, from the EDHS 2016 dataset with various socioeconomic, obstetric and nutrition variables.

Sample design

In-depth secondary analysis was conducted; the data was obtained from 2016 EDHS, which was taken from Central Statistical Agency (CSA). It is the fourth survey conducted in Ethiopia, following the 2011. A community-based unmatched case–control study was conducted among reproductive age women to identify the factors associated with obstetric fistula.

Variables and measurements:Dependent variable

The main outcome variable was obstetric fistula, which is defined as reproductive aged women experiencing lifelong obstetric fistula.

Independent variables

The selection of the independent variables was guided by the literature and availability of the variables in the dataset. The variables were categorized in to five groups: Maternal, Household, Obstetric, Anthropometric and wealth indices factors.

Maternal factors

Maternal age, maternal educational status, maternal antenatal care follow up, marital status, mother’s current employment status.

Household factors

number of household members, residence, wealth indices ranked in to five categories (poorest, poorer, middle, richer and richest), sex of household head.

Obstetric factors

Place of delivery, ANC follow up, size of child at birth, postnatal checkup, Preceding birth interval, Height (Cm) and ever had a terminated pregnancy.

Anthropometric measurements

Anthropometric data were collected through measurement of height and weight among all reproductive age women. Among four possible anthropometric indicators to evaluate women’s chronic under nutrition height less than 145 cm; body mass index (BMI) < 18.5 (thinness); weight less than 45 kg; and mid-arm circumference (MUAC) < 22.5 cm. BMI is defined as weight in kilograms divided by height squared in meters (kg/m2) [13].

Wealth index

A wealth index in the EDHS survey was measured based on household asset and data to classify individuals into 5 wealth indices (poorest, poorer, medium, richer and richest). Variables incorporated in the wealth index were ownership of chosen household assets (television, bicycle or car), size of agricultural land, number of livestock and materials used for house construction [13, 14].

Data analysis

Data analysis was carried out using STATA version 14 (Stata Corp, College station, Texas United states). The data explore for inconsistency and missing value. In all analysis, sample weights have done due to two stage cluster sampling design in the EDHS dataset to adjust for the imbalance probability selection among the strata [12]. Categorical type of data was analyzed by descriptive statistics (frequency and percentage).

Logistic regression analysis was used to identify factors associated with obstetric fistula. Bivariable analysis was carried out to see the crude association of each independent variable with the outcome variable (Obstetric fistula). Those independent variables with P-value ≤ 0.05 in the bivariable analysis were included in the final multivariable logistic regression analysis to adjust for confounders and to identify the final factors associated with obstetric fistula.

Logistic regression enter method was used during the multivariate logistic regression analysis. Before inclusion of predictors to the final logistic regression model, the multi-collinearity was checked using VIF < 10/Tolerance > 0.1 for continuous independent variables. The goodness of fit of the final logistic model was tested using Hosmer and lemeshow test at p value of > 0.05. The strength of association of the predictors and outcome variables has been indicated by adjusted odds ratio at 95% confidence interval. The significant association was declared at p ≤ 0.05 for the final logistic regression model.

Ethical considerations

The study permission was obtained from measure DHS project website to access the dataset (http://www.measuredhs.com).