Intimate partner violence (IPV) is sexual, psychological and physical coercive acts used against persons by intimate partners. When IPV occurs during pregnancy (IPVp), it can result in adverse maternal and pregnancy outcomes. No policy nor practice direction exists to address the rates and risk factors of IPVp in Kenya. Determining the prevalence, types and determinants of IPVp in Western Kenya would aid in the identification of pregnant women affected by and/or at risk of IPVp, as well as informing the development of policy, practices and programmes to support preventive interventions. In this cross-sectional study of 369 women who had given birth at Moi Teaching and Referral Hospital, participants were recruited using systematic sampling and data collected via structured questionnaires adopted from the WHO Violence Against Women Instrument. Associations were made in relation to physical or sexual violence and psychological violence. Logistic regression was used to assess the association between determinants and occurrence of IPVp. The overall prevalence of IPVp was 34.1%. Prevalence of physical or sexual violence was 22.8%. Psychological violence emerged as the most common (27.4%) form of IPVp. A lower than tertiary level of education and previous experience of IPV were individually associated with physical/sexual IPVp, whereas psychological IPVp was associated with previous experience of IPV and was prevented by the intimate partner having formal employment. Preterm birth rates were found to be higher than the country’s rates. The prevalence rates of IPVp are high in Western Kenya. Strategies that address the promotion of respectful, nonviolent relationships and that interrupt the development of risk factors are required. Policies (clinical guidelines) targeting prevention of IPVp and screening and the identification of at-risk women and survivors of IPVp are needed urgently. Primary prevention through interrupting the occurrence of predisposing factors is key in addressing IPVp.
This was a cross-sectional study conducted at the MTRH, one of the two national referral hospitals in Kenya. Kenya is divided into regions called counties. MTRH is located in the Western Kenyan region of the country, representing at least 22 of the 47 counties of Kenya. It also serves parts of Eastern Uganda and Southern Sudan. It has a catchment population of about 24 million. The study site was opted for by virtue of it being a referral hospital which would capture a wider population that is more representative of the targeted Western region. Screening for IPVp at MTRH was not routinely taking place and therefore its prevalence and associated factors had not yet been established. Approximately 12 000 deliveries are conducted each year at the hospital. The study population consisted of women who had delivered (vaginal deliveries and caesarean sections) between April and June 2017 and who were in the MTRH postnatal ward and mothers’ hostel. All women delivering at MTRH, including emancipated minors (those <18 years), were included in the study sample. Those who were very ill and unable to respond to the questionnaire were excluded. Using the Cochran (1963) formula, based on the local prevalence of 37% of IPV among pregnant women (Makayoto et al., 2013), a margin of error of 5% and 95% confidence, a sample size of 359 was arrived at as shown below. where P is the prevalence of IPV, d = 0.05 is the margin of error and Z is the quantile of the standard normal distribution corresponding to 100 × (1−α) %. Systematic sampling was used to recruit the participants from the delivery register. An anticipated average population size of 1000 mothers delivering in the facility per month and an intention to carry out data collection within a period of three months were used in calculating the sampling interval. Therefore, to sample from an average population size of 3000, the sampling interval was 3000/359, which was ∼8, the denominator being the study sample size. IPV was categorized into physical, sexual and psychological aggression using standardized definitions (Saltzman et al., 2002). Data were collected through interviews and a review of clinical records by clinical research assistants (two nurses, two psychological counsellors) who were trained on sampling, data collection and confidentiality. The research assistants sampled potential respondents from the delivery register, checked for eligibility and initiated the informed consent or assent process in the wards. A separate assent form was administered to parents or guardians. Details of all vaginal deliveries and caesarean sections were recorded in the Maternity Services Health Facility Register kept in the delivery ward of MTRH, from where the respondents were sampled. The first respondent was selected randomly from the first eight entries on the register on the first day of data collection. Subsequently, every eighth client was sampled. When a sampled client met the exclusion criteria or did not consent to take part in the study, the next client on the register was sampled. A total of 381 respondents were sampled within a period of three months. Twelve respondents that had been sampled did not consent. A sample size of 369 was finally achieved. VAW is a phenomenon that is difficult to measure due to a variation in the types of acts considered violent by different populations and to the varied tools and methodologies used by different studies. In an attempt to minimize the methodological problems emanating from the different studies and to allow comparisons of the same studies across different cultures, the World Health Organization developed the WHO VAW instrument. The study questionnaire consisted of four parts: (1) a researcher-designed socio-demographic and health history section; (2) a modified WHO VAW screening tool (Garcia-Moreno et al., 2006); (3) intimate partner characteristics, such as age, use of alcohol and/or drugs, level of education and (4) perinatal characteristics. The WHO screening tool covered the occurrence of physical, psychological and sexual violence and was modified to include the following questions in the psychological section, as advised by a list of acts of psychological aggression compiled by Breiding et al. (2015): whether the partner isolated or confined the woman, and whether the partner prevented the woman from visiting her friends or relatives. Any positive response to the questions on the screening tool confirmed the occurrence of IPVp. Psychometric assessments of the WHO VAW instrument have shown that it demonstrates good internal consistency, indicating that it provides a reliable and valid measure of types of violence (Nybergh et al., 2013; Marizella et al., 2014). It has also demonstrated significant cross-cultural validity and reliability when comparing the IPV prevalence rates in Sweden (Nybergh et al., 2013). Data on the outcome of the pregnancy, including birth weights, 5-min Apgar score, foetal death, immediate neonatal death and gestation at birth, were obtained from patient records. In this study, foetal death is defined as death of a foetus between 28 weeks of pregnancy and delivery, whereas immediate neonatal death is the death of a baby within 24 h of delivery. LBW was defined as weight at birth of <2500 g (WHO, 2012). A premature baby is one born before 37 weeks of pregnancy, while a preterm birth is defined as a delivery that occurred before 37 weeks of pregnancy (Blencowe et al., 2012). The Apgar score is a scoring system that rapidly provides a standardized assessment of infants after delivery. It is divided into five components: heart rate, respiratory effort, muscle tone, reflex irritability and colour. Each component is given a score of 0, 1 or 2 at 1 and 5-min intervals. A 5-min Apgar score of 7–10 is reassuring, a score of 4–6 is moderately abnormal, whereas a score of 0–3 is low (American Academy of Pediatrics, 2006). The questionnaire underwent forward and back translation from English to Swahili according to WHO protocol (WHO, 2016a). The questionnaire was refined based on a pilot conducted in January 2017 at Uasin Gishu District Hospital with 40 respondents. Research assistants either oversaw participants filling in the questionnaires or read the questions aloud in English or Swahili for those who could not answer by themselves. Each administered questionnaire was numbered. The gathered data were cleaned and entered into an excel spreadsheet, and encrypted to ensure confidentiality. The password was available to the principal investigator alone. Back-up of the data was done to cushion against loss. Once the data had completely been converted into the electronic database, the questionnaires were kept in a locked cabinet, and access was allowed to the principal investigator alone. They will be shredded after five years. Categorical variables were summarized using frequencies and percentages. Continuous variables were summarized using median and the corresponding interquartile range (IQR) due to a violation of Gaussian assumptions, which were assessed using the Shapiro–Wilk test and histograms. The infant birth weight was summarized using mean and standard deviation (SD). For global comparison purposes, we conventionally reported the prevalence of physical or sexual IPVp as one indicator and psychological IPVp as the other indicator. Analyses of associations with and without adjustment for previous experience of IPV as a variable were conducted. This is because previous experience of IPV has been shown to be on the causal pathway for other determinants to IPVp, such as childhood violence or marital status of partner (Bell and Naugle, 2008). Logistic regression modelling was used to determine factors associated with sexual or physical IPVp while adjusting for previous history of IPV. Odds ratios (OR) and 95% confidence intervals (95% CI) were used in bivariate analysis of determinants of IPVp, with significant variables further analysed in the final regression models (Tables 1 and and2)2) using forced entry regression technique to achieve the documented results. Determinants of physical/sexual IPVp adjusted for previous experience of IPV aOR, adjusted odds ratio; Bold, significant variables; Ref, reference variables; uOR, unadjusted odds ratio. Determinants of psychological IPVp adjusted for previous experience of IPV aOR, adjusted odds ratio; Bold, significant variables; CI, confidence intervals; Ref, reference variables; uOR, unadjusted odds ratio. Characteristics of perinatal outcomes were presented in tables. A descriptive analysis of the perinatal outcomes was carried out. Analysis of associations of IPVp and perinatal outcomes could not be done, as the sample size was underpowered to provide strong analytical conclusions of these outcomes. The Institutional Research and Ethics Committee of Moi University and MTRH approved the study. All the participants gave written informed consent and their privacy and data confidentiality were maintained. Data collection was in accordance with the recommendations of WHO Ethical and Safety recommendations for intervention research on VAW (WHO, 2016b). Women who were found to have survived IPVp were offered a counselling session with a psychological counsellor.