Factors associated with early sexual initiation among preparatory and high school youths in Woldia town, northeast Ethiopia: a cross-sectional study

An institution based cross-sectional study was conducted on March 7, 2016 in Woldia, northeast Ethiopia. Woldia is found 520 km from Addis Ababa, the capital of Ethiopia. According to the 2016 Woldia town administrative report, the estimated population was 75,446, of which 50.6% were males [21]. In the town, there was one preparatory and three high schools with a total of with a total of 4671 students. There were 87 sections, of these 59 sections were in high schools.

All regular preparatory and high school students in the town during the study were included. The required sample size was calculated using the single population proportion formula by considering a 95% confidence interval (CI), 3% margin of error(d), and the proportion(p) of early sexual debut 19% from the previous study in Shire-Endasellassie town, Tigray region [22]. Considering a 10% non-response rate, the final sample was 723 students. The simple random sampling technique with proportional allocation to each grade level was applied to reach participants. Thus, 269, 177, 143 and 129 students were selected from grade 9, 10, 11, and 12, respectively.

The data were collected using a structured self-administered questionnaire (Additional file 1). The questionnaire was first prepared in English and translated to Amharic (the local language) and back-translated into English by language experts to ensure an accurate translation. Ten data collectors were selected from Woldia University 4th year Midwifery students. Interviewers received 2 days intensive training prior to data collection. The questionnaire was pretested on 37 students in Kobo preparatory and high school out of the study area. The purpose and objectives of the study were clearly explained to participants before data collection. No personal identifiers were used on data collection forms, and teachers were requested to leave the classroom. To maintain privacy, seats were arranged far apart and copies were collected on the ballot.

Household wealth index adopted and developed from EDHS 2011 [23] was assessed using Principal Component Analysis (PCA) by considering the numbers and kinds of goods ranging from television to car, agricultural land ownership, quantity of cereal products, livestock and housing condition, such as flooring materials, household cooking material and place. Firstly, variables were coded between 0 and 1; then, the variables were entered and analyzed using PCA. Variables which had communality values greater than 0.5 were considered to produce factor scores. Next, the produced factor scores were computed to produce common factor scores. Finally, common factor scores were summed and categorized into quintile as lowest, second, middle, fourth, and highest. Similarly, paternal monitoring was assessed using a four-item Likert scale which was adapted and developed from previous studies [24, 25]. The items used to assess parental monitoring were ‘How often do your parents try to know who are your friends’, ‘Do your parents know where you are outside of home or school’ and ‘Do your parents know with whom you are outside home or school’. The responses ranged from ‘never’ (coded as 1) to ‘always’ (coded as 4). Finally, via PCA, it was classified as ‘poor’ and ‘good’ parental monitoring.

Student school connectedness was assessed using a five-item Likert scale adopted and developed from a previous study [26], such as ‘Do you feel close to people who are at school’, ‘Do you feel happy to be at this school’, ‘Do you feel as if you are a part of this school’, ‘Do school teachers treat students fairly at this school’, and ‘Do you feel safe being at this school’. The responses ranged from ‘strongly disagree’ (coded 1) to ‘strongly agree’ (coded as 5). Using PCA, the scale was classified into quantiles as poor and good school connectedness. In addition, knowledge of students on HIV/AIDS was assessed based on the transmission and prevention methods of HIV/AIDS which was adopted from EDHS 2011 [23], and the questions included condom use, number of sexual partners, awareness on healthy-looking person may have HIV/AIDS, and rejection of the two most common local misconceptions about HIV/AIDS that (HIV/AIDS can be transmitted by mosquito bite and supernatural means). Using PCA, the scale was ranked into ‘poor’ and ‘good’ knowledge.

After the data collection, the data were checked for completeness, coded manually, and entered into EpiData version 3.1 and then exported to SPSS version 20 for analysis. Descriptive analysis was carried out and presented using texts, tables, and graphs. Bivariate and multivariable logistic regression analysis were done to identify factors associated with early sexual initiation. Variables in the bivariate logistic regression with p-value less than 0.2 were fitted into the multivariable logistic regression to control the possible effects of confounders. The enter method was employed. Odds Ratios with their corresponding 95% confidence intervals (CI) was calculated to see the presence and the strength of associations. Model fitness was checked by the Hosmer-Lemeshow goodness of-fit test. In the multivariable analysis, variables with p-value < 0.05 were considered as statistically significant.

A total of 718 students participated in the study with a response rate of 99.3%. The median age of the participants was 17 (IQR ±1.6) years. Approximately, half (50.7%) of the students were female. The majority (91.8%) were in the age range of 15–19 years. In addition, 59.2% of the students did not take life skills training, and 15.6 and 7.7% had deceased fathers and mother, respectively, (Table 1).

About 46% of students reported that their parents strongly monitored them, and 49.6% said they had good school connectedness. Moreover, 55.3% of the participants discussed at least one sexual and reproductive health issue with their parents. The most common sexual and reproductive health issue which students discussed with their parents was on HIV/AIDS (Table 2).

Among the study participants, 39.4, 7.8, 17.3 and 45% reported that they drank alcohol, smoked cigarettes, chewed khat, and were exposed to pornographic materials, respectively, (Fig. 1).

The overall prevalence of early sexual initiation in the study area was 18.4% (95% CI:15.50%, 21. 30%). Out of the participants, 38.3% were sexually active, of whom 55.5% initiated the first sex with their regular boy/girlfriends and 56.7% had two or more sexual partners (Table 3).

In the bivariate logistic regression analysis, sex, marital status, educational status, attending religious programs, parent-youth discussion, drinking alcohol, smoking cigarettes, chewing khat, smoking shisha, life skills training, parental monitoring, school connectedness, peer pressure, and exposure to pornographic materials were associated with early sexual initiation at p-value < 0.2. In the multivariable logistic regression analysis, sex, attending religious programs, peer pressure, smoking cigarettes, parental monitoring, and pornographic materials remained statistically significant and were independently associated with early sexual initiation. The odds of having early sexual initiation were three times (AOR = 3.2; 95% CI: 1.84, 5.44) as high among students who didn’t attend religious programs compared to their counterparts. Similarly, the odds of having early sexual initiation were nearly three times (AOR = 2.8; 95% CI:1.77, 4.53) as high among students who had poor parental monitoring than those who had good parental monitoring. Compared to students who had not been exposed to peer pressure, students who had peer pressure were at increased odds of having early sexual initiation by 90% (AOR = 1.9; 95% CI:1.14, 3.25). Moreover, early sexual initiation was about three times (AOR = 2.7; 95% CI: 1.68,4.40) as high among students exposed to pornographic materials than their counterparts (Table 4).