An institution-case-control study was conducted in Arba Minch Zuria district among high school female students attending regular education from March 20, 2023 to May 20, 2023. The woreda is situated around Arba Minch Town, the capital of the South Nationalities, Nations, and Peoples (Debub) Regional State. Hawassa, the Sidama region's commercial and administrative hub, is 275 km away, and Addis Ababa, the capital of Ethiopia, is 505 km South-West. In 2023, there were roughly 129,666 people living there. Located in the Gamo zone, it is one of 19 woredas with seven high schools. Located in the Gamo zone, it is one of 19 woredas with seven high schools offering both regular and irregular classes. Of the total number of students enrolled in all high schools, 1,639 were female.

The study participants were female high school students randomly chosen from the source population who met the inclusion and exclusion criteria. The source population for this study consisted of all female high school students enrolled in their education during the academic year 2022/2023.

An institutional (school) based unmatched case control study design was conducted. Reports of schoolgirls having been sexually assaulted were first obtained from the Arba Minch Zuria woreda which provide both day and night education program, Southern Ethiopia. Reports of schoolgirls being sexually assaulted were first obtained from the Arba Minch Zuria district police office. According to this report, there were seventy-five (75) cases who were attending or had attended the mentioned high school. A total of 150 female control students were also randomly selected from non-case students. Then, all cases were interviewed one after the other through home visits for those who had stopped attending school, or one-on-one interactions with school administrators at break time for those currently attending school.

Students who were critically sick or who attended classes on the weekends were excluded from the study due to the challenges in collecting the required data. Individuals who met the eligibility criteria but refused to give consent to participate in the research were also not included.

Before the two weeks of actual data collection, 5% of the female students at Chencha Woreda High School completed pretests of the questionnaires. In response, adjustments were made to the instrument, clarity, and ambiguity of the language tool. Five female teachers were chosen to act as data collectors and supervisors due to their previous experience in data collection and proficiency in local languages. They received two days of continuous training on the study's objectives, data collection tools, and techniques. Data was collected by self-administered questionnaires at break time in morning and afternoon without affecting teaching leaching learning processes. To maintain privacy, a separate space was prepared in advance for students, with data collectors and facilitators assigned to each room. The questionnaire should be filled out and placed on the designated table, but participants were instructed not to write their names or any other identifying information on it. As soon as the data was obtained, the supervisors and investigator checked it for accuracy, clarity, and consistency.

Data was collected by self-administered questionnaires and the coded, modified and input into the Epidata version 4.6 software, the data was ready for analysis. Descriptive statistics, which comprised relevant features including frequencies, proportions, and summary statistics, were used to characterize the study population. The association between the dependent and independent variables was evaluated using binary logistic regression. Multi-collinearity and the model's goodness of fit were checked using the VIF (<10) and Hosmer-Lemeshow tests (>0.05), respectively. Accordingly, any variable having p ≤ 0.25 was considered a candidate variable for multivariable analysis and entered into a multivariable logistic regression model using the backward elimination stepwise likelihood ratio method. All tests were two-sided, and p < 0.05 was considered statistically significant. The results were reported as the odds ratio (OR) and 95% confidence interval (CI). A p-value of less than 0.05 was chosen as the level for statistical significance.

The outcome variable of this study is the determinants of childhood sexual assault, which includes six items (components) that were extracted and adapted from previously published research (Table 1).

A total of 225 participants (75 cases and 150 controls) were included in the study, for a response rate of 100%. The mean age of the participants was 17.26 years (SD ± 1.54) with a minimum and maximum age of 14 and 22, respectively. Participants who were over 18 were asked only about their exposure to CSA that occurred before they turned 18. The majority of the students, 77.3% and 78.7% among cases and controls, respectively, came from rural areas. One-tenth of the participants, 22 (9.8%), were married; 9.3% of the cases and 10% of the controls. The majority of participants, 66.7% vs. 78.7%, among cases and controls, respectively, had both biological parents. The majority, 57.3% vs. 72%, of participants among cases and controls, respectively, were Gamo in ethnicity, while 48% of mothers among cases and 51.3% among controls were Protestant in terms of their religion.

Among the participants in the cases and the controls, 33.3% and 40.7%, respectively, their father had been working on daily labor. While about 61.3% and 56.7% of the cases and controls were living in families of more than five, 81.3% of the families of the participants among the cases and 88% of the participants among the controls were earning less than 1,317 ETB (Table 2).

Only 26.7% of both cases (16%) and controls (32%) slept with their mothers. The majority of the participants, 77.3% vs. 63.3%, had a boyfriend among the cases and control, respectively. Only 45.3% of the participants had an open discussion on sexual and reproductive health issues. About 27.6% of participants’ families (38.7% vs. 22.7%) among the cases and controls used substances like alcohol and chewing or smoking. The majority of participants started sexual practices between the ages of 14 and 16. The largest percentage, 40% vs. 26.7%, of participants among cases and controls were working with street food vendors, while 52% of participants among cases and 29.3% among controls claimed improper library arrangements (Table 3).

Most of the students (cases) were assaulted by non-relatives such as neighbors (26.7%), boyfriends (25.3%), and schoolmates (21.3%). About 17.3% of the participants did not know the perpetrator (strangers). Just 7 (9%) of the offenders were detained, placed under arrest, and brought to court. Regarding the place of assault, the majority of 24 (32%) of the students were assaulted at their own home, and 18 (24%) raped on the street or in the field (undefined area). The Among 75 rape survivors, only 4 (5.3%) were assaulted by family members.

Among the 75 cases adolescents who reported rape, more than half 40 (53.3%) had visited the clinic and received post exposure prophylaxis for pregnancy and infectious disease. Twenty (26.7%) of the respondents reported having suffered a physical harm during the incident. Twelve respondents (26%) reported that they ended their relationships with both their boyfriend and girlfriend following the rape, and nearly one-fifth (20%) felt they had lost confidence in themselves compared to before the incident (low self-esteem). Additionally, around ten students (13.3%) intended to stop their studies for one to two years due to feelings of humiliation, guilt, and fear. And nearly one-fourth of participants reported become substance users after the assault (Figure 1).

In the bi-variable logistic regression, ten variables: presence of both parents, parental conflict, having a boyfriend, having a history of sexual intercourse, open discussion with parents, history of alcohol drinking, living with non-biological parents, being street food vendors, and improper library arrangements were statistically significantly associated with childhood sexual assault at a p-value of 0.25. All these variables were included in the multivariable logistic regression model, but other variables that had a p-value greater than 0.25 in this model were not included in the multivariable regression model.

In the multivariate logistic regression model, four independent variables, namely, having ever drunk alcohol, with whom they had been living, being street food vendors, and improper library time arrangements, were found to be statistically significantly associated with life-time childhood assault at a 95% confidence level and a p-value of less than 0.05.

The study showed that students who had been using alcohol before incident were about four times more likely to experience childhood assault compared to students who did not use alcohol (AOR = 4.0, 95CI: 1.68–9.70). students who had been living with non-biological parents and alone were about seven and five times more likely to be exposed to sexual assault as compared to those who had been living with their biological parents (AOR = 7.49, 95CI: 2.72–13.65) and (AOR = 4.6, 95CI: 1.49–14.41), respectively. Similarly, participants who had been working as street food vendors (selling) were five times more likely to have experienced sexual assault compared to those who did not participate in this work (AOR = 4.5, 95CI: 1.48–13.70). Likewise, the odds of childhood sexual assault among students exposed to improper library time in school compared to their counterparts were fivefold (AOR = 5.0, 95CI: 1.87–13.47) (Table 4).

Generally, this study suggests that policymakers should modify the laws regarding illegal acts committed against children, even though there isn't a 100% safe place to prevent juvenile sexual assault. In addition, parents should carefully monitor their school-age children if they read this study. This work serves as a foundation (cornerstone) for further research on the subject of childhood rape by other scholars.

The nature of this study design allows for the assessment of temporal causality between exposure and the problem (disease). However, given the sensitive nature of the study, which addresses personal issues related to sexuality, there may be underreporting of sexual abuse experiences. Therefore, the findings should be interpreted in light of these limitations.