Study design and setting
According to 2015 Tigray regional health bureau annual report there are a total of,one specialized hospital, 15 general hospitals, 22 primary hospitals, 204 health centersand 712 health posts and there are three private hospitals. There are 51 specialdoctors, 87 general practitioners, 3092 nurses, 792 midwives in the region. The studywas conducted from November 2017- June 2018. Institutional based cross-sectional study was used.
Data collection
There are 41 hospitals (1 specialized hospital, 15 general hospitals, 22 primaryhospitals and 3 private hospitals) which provide delivery service in the study area;health facilities were stratified in to private and public hospitals. Then 1 privatehospital and 8 public hospitals were selected by simple random sampling technique.Participants from each selected health facilities were sampled by systematic.Afterwards, every 3rd postpartum women were included until the required sample sizeis reached. Besides, consecutive participant was included if the selected participantwas not eligible. Average client load for each hospital was taken from 3 months’client flow prior to data collection period and proportional allocation to each hospitalwas made based on their respective quarterly client flow.The required sample size of the study participants is determined by using Open EpiInfo 3.03 software. The following assumptions are used while calculating the samplesize. Study from Ethiopia about intimate partner violence during pregnancy and lowbirth weight 25.8% [14], 95% confidence interval, margin of error 5%, design effectof 2 and expected non response rate 10%. Based on this the calculated sample size is648.Questionnaire was prepared first in English and then translated into Tigrigna andagain back translated to English by language expert to keep the consistency of thequestioners.Then data was collected through structured questioner from postpartum women andinterviewer administered technique was used. Data on still birth was collected fromchart and mothers. Outcomes of interest for this analysis pertained directly to neonataloutcomes, which was obtained through chart review within 72 hours of delivery. Birthweight (g) and gestational age (weeks) were taken directly from the chart.
Intimate partner violence
Maternal exposure to IPV was determined through thequestion: ‘‘when you were pregnant for this child did your current partner orboyfriend do any of the following things to you? The lists of potential offences wereas follows: Physical violence: slapped, pushed or shoved, hit with fist or somethingelse that could hurt her, beaten abdomen Choked or burnt on purpose, used or threatened to use knife, gun or weapon.Emotional violence: insult, humiliation intimidated on purpose, threatened to. Sexualviolence: Forced into sexual intercourse when you did not want, had sexualintercourse when you did not want to because she was afraid of what partner mightdo, forced to do something sexual that you found degrading or humiliating.
Stillbirth
Is typically defined as fetal death at or after 28 weeks of pregnancy. Itresults in a baby born without signs of life.
Data process and analysis
Double data entry was done by using EPI Info 2008 and exported to SPSS version 20software package for analysis. Experience of any physical, sexual or emotionalviolence was considered if a woman reported being exposed to at least one of the actsof violence exerted by her partner while she was a pregnant for current neonate.In preparation for bivariate and multivariable logistic regression, outcome variablewas coded as dichotomous: still birth (1 = yes and 0 = no). To estimate the associationbetween maternal exposure to intimate partner violence and risk of still birth, logisticregression analyses were performed and odds ratios (OR) with 95% confidenceintervals (CI) were calculated. Multivariable logistic regression analysis wasperformed where intimate partner violence plus other covariates that could influencestill birth such as age, educational level, occupation during pregnancy and alcoholintake were included. The degree of association between independent and dependentvariables were assessed using odds ratio with 95% confidence interval.Permission letter was obtained from regional health office and was presented toselected hospitals. Confidentiality was assured that their responses will not in any waybe linked to them. In addition, they were told they have the right not to participate andwithdraw from the study