Study design
A school-based comparative cross-sectional study was conducted to compare the extent of SRH services utilization among students from YFS implemented and non-implemented areas and to identify the associated factors among adolescents from 1-30, March 2019.
Study Setting
South Ari is one of the districts in South Omo Zone and it is located 735 Km away from Addis Ababa, the capital city of Ethiopia. There were a total of six high schools in the district; two of them were located in YFS implemented areas of the district.
Participants
The source population of this study was all adolescents between 15-19 years old who were attending their studies in the public high schools in South Ari district in 2019. The study population was all adolescents of age between 15-19 years in selected public schools during data collection period. Multi-stage cluster sampling technique was employed in order to select a representative sample of students. Four secondary schools (two from each of the YFS and non-YFS sites) were selected randomly out of the six secondary schools in the district. Samples were selected from the randomly selected schools based on proportion to the respective number of students. The total sample was allocated to each grade (9 -12) in proportion to their student size. From each grade, sections were selected randomly. Finally, the study participants were selected by using computer-generated random numbers based on a sampling frame prepared by using their attendance sheet obtained from their respective schools. On the date of data collection, the randomly selected students were invited to the classes arranged in advance for data collection.
Variables
The dependent variable of this study was SRH services utilization. Independent variables were predisposing, enabling and need factors of health care utilization. The predisposing factors considered in the study were sex, age, and place of residence and marital status (or relationship condition), knowledge and attitude. The enabling factors were access to services, cost of services, discussion of SRH issues with family members and perceived parental income status. The need factors include perceived need for SRH service, perceived personal health status, and worries (concern) about one’s health.
Measurement
Data collection tools and procedures
The data collection instrument was a self-administered questionnaire developed after reviewing previous publications from similar or related literatures. The questionnaire was prepared in English and translated to Amharic and back to English to check for consistency of meaning. Data were collected by four diploma holding health workers and two bachelor holding supervisors from the health discipline respectively.
Operational definitions
Adolescent: In this study adolescent stands for boys and girls between the ages of 15–19 years.
Utilization of sexual and reproductive health services: This was measured through the dichotomous response (yes or no) by asking whether a participant had utilized one or more of SRH service components within the last 12 months. The positive response was further validated with questions on the type of SRH services utilized. A positive (“yes”) response to any one of these services was regarded as service utilization[21].
Knowledge of SRH services: Adolescents who scored above the mean of questions assessing participants’ knowledge were labeled as having good knowledge and those scored a mean or below the mean were considered as having poor knowledge [1].
Attitude towards SRH services: Attitude was measured by 7 items each having three categories of responses: disagree, neutral and agree. It was further recoded in to agree and disagree by considering being neutral to attitude questions as a disagreement. Finally, adolescents who scored above the mean for attitude questions were regarded as having a “positive attitude” and those who scored a mean or below the mean on the questions assessing attitude were regarded as having a “negative Attitude” [21].
Bias
Since the study was institution-based, there is a risk of contamination across observation units. Therefore, to minimize this, data were collected at the same time from the selected high schools. Since students were asked to remind sexual and reproductive health services provided to them in the last one year, this might lead to a recall bias.
To assure the quality of data, a day-long training was provided for data collection facilitators and
supervisors to brief them on the aims of the study, its significance and the contents of the
instrument. The overall activity of data collection was closely supervised and coordinated by the principal investigator. The collected data were reviewed and checked for completeness and
consistency onsite in the schools by the supervisors and data collection facilitators and before data entry by the research team. Pre-test was conducted in a high school at a nearby district on 5% of the sample size two weeks before commencement of the actual data collection and where appropriate, modification to wording and content of questions was made based on the results from the pretest. To reduce Hawthorne effect owing to the physical presence of the data collection facilitators in the class rooms, they were trained to remain at the stages of each class room with minimal movement in between the students’ seats.
Sample size determination
The sample size for this study was determined based on assumptions of a 50% prevalence of SRH service utilization (as there was no study on the same topic) among school adolescents in the high schools where YFS was implemented (P2=percent of exposed having the outcome=50%) and a difference of 20% between YFS implemented and non-implemented areas (P1=percent of unexposed having the outcome=30%), a 5 % level of significance and 80% power and a ratio of unexposed to exposed equal to 1. With these considerations, a sample size of 208 was calculated. A design effect of two was considered as the sampling procedure had involved a multistage cluster sampling and an additional 10% was considered to compensate for possible non-response. Finally, a total sample of 458 was calculated as a sample size for the study.
Based on findings from the study, power was calculated using STATA software Version 13 with a significance level of 0.05, sample size (N)= 458, P1= proportion of youths from YFS not implemented areas and who utilized SRH services=9.9%=0.099, P2= proportion of youths from YFS implemented areas and who utilized SRH services=33.8%=0.338, and allocation ratio=1. Based on this assumptions and values, the power was estimated to be 1. Therefore, the sample calculated using this assumption was found to be adequate to show a significant difference in SRH utilization among the YFS implemented and non-implemented areas.
Statistical Methods
Data were entered in to Epi data version 4.4.1 software and then exported to SPSS version 20 for analysis. Descriptive statistics were computed and findings were summarized in tables and graphs with frequencies, mean, or standard deviations where appropriate. χ2 test was used to see significant difference in SRH service utilization among adolescents from YFS implemented and non-implanted areas. The association between SRH service utilization and the independent variables was examined by binary logistic regression. Crude odds ratios were computed to determine the strength of association of the selected explanatory variables with the dependent variable in the initial bivariable logistic regression analysis. To control for potential confounders and identify independent factors of SRH service utilization, variables which showed an association at a p-value ≤ 0.25 in the bivariable logistic regression model were selected as a potential candidate to fit to the final multivariable logistic regression analysis model. Model fitness was checked using Hosmer and Lemeshow goodness of fitness test (p-value ≥0.05).The association between SRH services utilization and the explanatory variables was reported by using odds ratio with its 95% CI and variables having p-value less than or equal to 0.05 in the multivariable logistic regression model were considered as statistically significant.