Research design and sample
A cross-sectional survey was conducted from July-October, 2019 among pregnant women in urban communities of Hengyang City, Hunan Province, China. A multi-staged cluster random sampling method was used in this study. There are 5 districts in urban Hengyang. In the first stage, a street from each district was randomly selected: Zhengxiang Street, Qingshan Street, Baishazhou Street, Guangdonglu Street and Zhurong Street. In the second stage, proportional sampling was carried out at a proportion of 1/3 to randomly select 4 communities in Zhengxiang Street, 3 communities in Qingshan Street, 3 communities in Baishazhou Street, 2 communities in Guangdonglu Street, and 2 communities in Zhurong Street. In total, 14 communities were selected for this study. All pregnant women who were registered in community health service centres and who met the inclusion criteria were potential subjects in this study. The inclusion criteria for the study were as follows: 1. women in the third trimester of pregnancy; 2. pregnant women over 16 years old; and 3. pregnant women who had local household registration, or migrant people who had lived in urban areas of Hengyang city for more than 6 months. The exclusion criteria were: 1. pregnant women with cognitive disorders, severe mental illnesses or other serious diseases who cannot fill out the questionnaire by themselves; and 2. pregnant women who refused to participate in the study. The sample size calculation formula for cross-sectional studies was used to calculate the minimum theoretical sample size for this study. According to the prevalence of DV which have been reported in a previous study, admissible error was 0.1, α=0.05. 763 people were required in order for the participants to represent the population. There were 819 registered samples in the 14 selected communities, of which 6 were excluded because of refusals to respond and failure to contact; therefore, 816 pregnant women who met the requirements participated in the study. The response rate was 99.3% (813/819).
All subjects gave their informed consent for inclusion before they participated in the study. The study was conducted in accordance with the Declaration of Helsinki, and the protocol was approved by the Ethics Committee of Xiangya School of Public Health, Central South University (XYGW-2019-056).
Data collection and measurements
General characteristics were collected, including age, ethnicity, marital status, monthly per capita household income, education level, occupation, partner’s education level, partner’s occupation, whether they had medical insurance，whether they had a smoking habit, whether they had a drinking habit, the partner’s smoking habit, the partner’s drinking habit, the mode of pregnancy, history of delivery, pregnancy intention, whether they had experienced multiple abortions/sterility, whether they had received antenatal examinations, whether they had pregestational diseases (depression or diabetes), and whether they had pregnancy complications (gestational diabetes, pregnancy-induced hypertension, intrahepatic cholestasis of pregnancy, and others). Being married was defined as being in a stable marriage. Unstable marriages included unmarried, divorced, and widowed.
Family factors consisted of four aspects: family relationships, family structure, family-related stressors and family function. Family relationships were evaluated on two dimensions: the relations with the mother-in-law and relations with other family members. Good or bad relationships with family members were judged subjectively by the participants. Family structure included the number of cohabitants and whether they lived with elders. In terms of family-related stressors, three family negative life events were listed: whether they had medium household debt, whether they were separated from their partner, and whether their partner had extramarital affairs. The Family Adaptation Partnership Growth Affection and Resolve Index (APGARI) is a tool for evaluating family functions. The Family APGARI has five items, and each item was answered on a 3-point Likert scale from “often” (2 points) to “rarely” (0 points). The total score was 0-10 points. High family APGARI scores ranging from 7 to 10 points indicated good family function, middle family APGARI scores ranging from 4 to 6 points indicated moderate family dysfunction, and low family APGARI ranges from 0 to 3 points indicated severe family dysfunction. Family APGARI has been widely used and has good reliability and validity[30, 31]. The Cronbach's α is 0.876.
Assessment tools for DV
The Abuse Assessment Screen Questionnaire (AAS) was compiled in 1995 by McFarlane and translated into Chinese by Leung of the University of Hong Kong and was used to assess DV during pregnancy. There are eight items in this scale assessing mental, physical and sexual violence as well as the psychological response to perpetrators in three periods: lifetime and 12 months prior to and during pregnancy. The scale provides participants several common perpetrators to choose from（husband, ex-husband, boyfriend, stranger and others）. The response to each item was either Yes or No. If the interviewee answered “Yes” to one or more of Questions 5 to 7, she was identified as a victim of DV during pregnancy[32, 33]. The scale’s Cronbach's α is 0.685. The retest reliability of the scale was 0.91, the specificity was 97%-99%, and the sensitivity to mild and severe violence was 32% and 61% -94%, respectively.[33, 34]These scales are widely known as self-management tools for screening DV against pregnant women with good validity and reliability.
EpiData 3.1 and SPSS 19.0 software were used for data entry and statistical analysis. Categorical variables are expressed as n (%), the χ2 test was applied for comparisons of general and family characteristics between women who had experienced DV and who had not (no DV group) during the pregnancy. The crude odds ratio (COR) was reported by multivariate binary logistic regression models. For instance, some general characteristics will be adjusted, and adjusted odds ratio (AOR) and 95% confidence interval (95% CI) were reported. The statistical significance level was accepted as p＜0.05. All statistical tests were 2-sided.