The study was conducted at Adigrat General Hospital in the Tigray region of Northern Ethiopia. Adigrat General Hospital is a publicly owned general hospital that also hosts Adigrat University as a teaching hospital. With over 142,765 patient flows annually, it also serves as a referral facility for catchment areas in the Eastern Zone of the Tigray region and some adjacent areas of the Afar region.

The study area was selected based on the Tigray Regional Health Bureau’s 2018 Annual Hospital Report. Adigrat General Hospital had the largest number of inpatient cases of spontaneous abortion, according to the data. Moreover, according to the 2019 Adigrat Hospital report, there were 1,009 abortion cases, 616 of which were spontaneous. The study was conducted between March 13, 2020, and June 12, 2020.

A facility-based, case–control study design was used, with the cases (with the condition) consisting of all women who were admitted to the hospital for post-abortion treatment or outpatient care following a spontaneous abortion. Whereas, pregnant women who came to the hospital for their routine third or fourth Antenatal Care (ANC) follow-up visits were considered controls (without the condition). The main reason for restricting those women in their 3rd or 4th ANC visit is to rule out an abortion, as abortion cannot be defined among those women in their 3rd or 4th visits and if we include those pregnant women who were in their 1st or 2nd ANC visits there will be a misclassification bias.

The sample size was determined using double population proportion formula using the Epi info version 7.2.0.1 software for Windows. Taking IPV as a primary exposure variable [26.1% (12)], and with the assumptions of a two-sided significance level (α = 5%), power (1-β) = 80, 95% confidence level, a ratio of cases to controls (R) 1:2. When comparing women exposed to IPV to controls, the likelihood of spontaneous abortion was 2.08 (13).

A 15% non-response rate was implemented due to the possibility of unpleasant pregnancy outcomes, the extremely sensitive and private nature of the exposure, and the outcome variable. As a result, the total sample size was expanded to 378, consisting of 126 cases and 252 controls. Until the necessary sample size was reached, all women admitted to post-abortion care and outpatient care services who had spontaneous abortions were contacted using a convenient sampling technique. Similarly, all pregnant women who arrived for their 3rd or 4th ANC follow-up visits were selected until the required sample size was attained.

A pre-tested (in 5% of the population outside of the study area), structured interviewer-administered questionnaire was utilized to collect the variables of interest, and a maternal medical record review was employed to acquire the maternal medical conditions like hypertension status during pregnancy, Human Immunodeficiency Virus (HIV) status, chronic heart disease, diabetes mellitus, and other conditions. The questionnaire was developed after reviewing previous similar studies (14–18). The IPV instrument was adapted from the WHO-2005 multi-country research tool (1) and the EDHS 2016 report (6). The questionnaire was developed in English, then translated into the local language, Tigrigna, and then retranslated back to English by independent professional translators to ensure consistency and understandability of the tool by the women.

The outcome variable of interest was spontaneous abortion and the independent variables were, socio-demographic characteristics, which contain maternal age, maternal education status, marital status, age of at first marriage, residence, religion, occupation, monthly household income, and partner education status. Medical factors include; hypertension during pregnancy, nutritional status of the mother, HIV status (positive/negative), any infectious disease, previously untreated Sexually Transmitted Diseases (STDs), chronic heart disease, and diabetes mellitus.

Maternal reproductive factors include birth interval, history of previous induced abortion, history of previous spontaneous abortion, history of previous cesarean delivery, gestational age, and parity. Risky behavioral factors include substance use during pregnancy, cigarette smoking, khat chewing, alcohol consumption, partner smoke cigarette status, and partner alcohol consumption.

Three female data collectors with bachelor’s degrees in midwifery who did not work for the hospital, as well as a senior public health professional with a master’s degree, served as a supervisor. The primary investigator trained data collectors two days before the actual data collection period about the study’s objective, the problem’s high level of sensitivity, the confidentiality of responses, and the questionnaires’ contents. Both cases and controls were interviewed using the same study tool. To protect their privacy and encourage their communicative motivations, data were collected in a separate, quiet, private room, and partners were not involved.

Spontaneous abortion or miscarriage: defined as the loss of a pregnancy without any intervention before 28 weeks gestational age from the Last Normal Menstrual Period (LNMP) (19).

Emotional IPV during pregnancy is defined as women experiencing any of the following; having been insulted by their husband by using abusive language that made them feel bad; being insulted in front of others; having been scared or intimidated on purpose, or having been threatened by their husband with an object such as a stick, belt, knife, gun, etc. by a current partner or boyfriend during the index pregnancy (1).

Physical IPV during pregnancy is defined as women experiencing any of the following; being slapped or having something thrown at her that could hurt her; being pushed or shoved; being hit with a fist or something else that could hurt; being kicked, dragged, choked, or burned on purpose; and/or being threatened with or having a gun, a knife, or another weapon used on her by a current intimate partner during the index pregnancy (1).

Sexual IPV during pregnancy is defined as women experiencing any of the following; being physically forced to have sexual intercourse when she did not want to, having sexual intercourse because she was afraid of what her partner might do, and/or being forced to do something sexual that she found humiliating or degrading to her by an intimate partner during the index pregnancy (1).

Any IPV during pregnancy is defined as women who experienced at least one of the above offenses being classified as having experienced any IPV during the index pregnancy (1). Risky behavior includes alcohol consumption and khat chewing. Khat (Catha edulis) is a plant with psychoactive properties, and the leaves and shoots are chewed (20).

The completeness and consistency of the obtained data were examined. For data input, Epidata Manager version 4.6.0.2 was utilized, and then the entered data were exported to STATA version 14 for Windows. Descriptive statistics for numeric variables were provided as means (standard deviations) when normally distributed, and as median with interquartile range (IQR) when skewed. Categorical variables were presented using frequency and percentages.

A binary logistic regression model was used to investigate the existence of crude association and to select candidate variables for the final multivariable logistic regression model. Finally, factors with a p-value <0.2 (21) were put into a multivariable binary logistic regression model to isolate the effect of IPV on spontaneous abortion, while controlling for other factors. The strength of the association was determined by both the crude odds ratio (COR) and the adjusted odds ratio (AOR) with a 95% confidence interval, and a p-value of <0.05 was used to indicate statistical significance.

A total of 378 study participants (126 cases and 252 controls) were approached; however, two cases and five controls were not willing to participate. From 371 women (124 cases and 247 controls) completing the interview, the response rate was 98.41% for cases and 98.02% for controls. In the case group, the median age was 26.5 years with an interquartile range of 22 to 32 years. Similarly, the median age of the control group was 26 years, with an interquartile range of 23–30 years. Ninety (72.58%) and 203 (82.19%) of the cases and controls, respectively, were people who lived in cities (Table 1).

Regarding the women’s gravidity, 88 (70.97%) cases and 153 (61.94%) controls were both multigravidas. In the case group, 53 (62.35%) women had an inter-pregnancy interval of at least 24 months, whereas in the control group only 122 (84.72%) women had an inter-pregnancy interval of at least 24 months. Regarding the presence of maternal hypertension, five (4.03%) and five (2.02%) of the cases and controls, respectively, had hypertension (Table 2). Regarding the social norm that supports IPV, 23 (18.55%) and 65 (26.32%) of the cases and controls, respectively, stated that there are such norms (Table 2).

From the total number of study participants, 1 (0.81%) cases and 2 (0.81%) controls indicated that they had ever chewed khat. Maternal alcohol intake during the index pregnancy revealed that 79 (83.16%) of cases and 172 (88.21%) of controls had consumed alcohol during their index pregnancies (Table 3).

The total percentage of pregnant women who were exposed to any form of IPV was 152 (40.97%), with 66 (53.23%) cases and 86 (34.82%) controls. In both the cases and the controls, the percentage of women who experienced physical IPV during the index pregnancy was 48 (38.71%) and 47 (19.03%) among cases and controls, respectively (Figure 1).

Bivariate analysis was carried out to identify those candidate variables for the final model, having a p-value <0.2 (Table 4). Two models were used for identifying the determinants of spontaneous abortion, where model one contains any form of IPV and all other variables, whereas model two contains specific types of IPV (we did not include any form of IPV here) and all other variables. Accordingly, after adjusting for all other factors, women who were exposed to any form of IPV during pregnancy were almost four times more likely to have a spontaneous abortion than those who were not exposed to any form of IPV (AOR = 3.66; 95% CI 1.69–7.95).

Similarly, pregnant women who were physically exposed to IPV had a threefold higher chance of spontaneous abortion than pregnant women who were not physically exposed to IPV (AOR = 3.06; 95% CI 1.49–6.63). Moreover, in models one and two, women who had an interpregnancy interval (IPI) of <24 months experienced a four-fold increase in the risk of spontaneous abortion compared to those who had an IPI of >24 months (AOR = 4.46; 95% CI 1.65, 12.07), and (AOR = 4.21; 95% CI 1.57, 11.29), respectively (Table 5).

The study’s strength was that it was the first local study to date to use the WHO’s standard questionnaires to evaluate the outcome variable and seek to demonstrate the association between intimate partner violence and spontaneous abortion. Due to the sensitivity of the subject, there are certain study limitations, and responses to intimate partner abuse during pregnancy may still be underreported.

Another limitation was the possibility that an induced abortion could have been mistaken for a spontaneous one, which could have led to an inaccurate assessment of the outcome. Since the study is facility-based using convenient sampling, generalizing the findings is difficult. We did not specifically address those types of social norms that support IPV; listing and determining those specific social norms that support IPV would have been important. Therefore, it is important to consider such potential limitations when interpreting the findings of this study.