Study design and setup
The study was a baseline cross-sectional study of adolescents (10-19 years) and their parents/caregivers residing in Adaklu district of the Volta Region of Ghana. The Adaklu district comprises a total of 91 communities and the inhabitants are predominantly farmers. Farming forms about 78% of economic activities in the district and the inhabitants are mainly into crop production, bee keeping and livestock rearing. The district has a projected population of about 44,942 made up of 22,022 males and 22, 920 females, with females representing 51.0 percent of the population [20]. The estimated adolescent population in the district is 9, 887 and Women in Fertile Age account for 10, 394 out of the total population [20]. There are adolescent health clubs established in all the communities in the study area.
Study Population
The participants of this study were adolescents between the ages of 10-19 years, from 30 selected communities in the district who had lived in the area for a minimum of 6 months.
Sample size and Sampling Technique
A total of 30 communities (clusters) with a minimum population of 500 people were randomly selected from the district. The modified Expanded Program on Immunisation (EPI) cluster sampling technique was used to select seven households in each community. to which questionnaires were administered to adolescents (10-19 years) in each household. A conscious effort was made to include equal numbers of male and female respondents. Similarly, both adolescents who were members of adolescent health clubs and those who were not members had an equal chance of being recruited for the baseline study. The sample size was calculated using the single population proportion formulation under the following axioms: 17.6% proportion; confidence level was taken to be 95% with α=0.05 value, 3% margin of error and design effect of 2. Five percent was added for the expected non- response rate, constituting the final sample size of 221 adolescents aged 10-19.
Data collection tools and procedure
Data was collected electronically using RedCap software. The questionnaire was adapted from adolescent survey tools used in previous adolescents sexual and reproductive health studies of international standards and several literatures reviewed to achieve the study objectives [21]. The tools were pre-tested and all the necessary corrections and changes were made.
The tool designed in English but administered in the Ewe language was used to collect information on socio-demographic characteristics, sexual and reproductive health knowledge and services, sexuality, sexual communication between adolescents and their parents, risky sexual behaviour, sexual harassment and coercion (See supplementary file A). Data was collected by 12 well trained research assistants selected based on their proficiency in English and Ewe language commonly spoken in the study area.
Outcome and Predictor variables
The outcome variables of this study, were sexual communication with parents, comprised of three categories; sexual communication with (i) both parents, (ii) only one parent (iii) no parents, similar to the measurement used in previous studies by Manu and colleagues which explored the extent and patterns of parent-child sexual communication [16].
Predictor variables included sex[ male or female], age [10-14 and 15-19], level of education,[ no education, primary and secondary/higher] current school attendance[currently in school or out of school], adolescent club membership, household living arrangement[living with both parents, only father, only mother, only brother(s), only sister(s) or not], frequency and ease of communication with parents, sexual history[never had sex and ever had sex], history of contraceptive use[never used contraceptives and ever used contraceptives], sexual harassment experiences[kissed, breast fondled, touched) or not. The respondents were also asked about their participation in Adolescent health clubs which are common in the study district. The measurement of the predictor variables was guided by previous studies [36,37-43]
Data processing and analysis
Both descriptive (frequency distribution tables) and inferential statistics (bivariate and multivariate analysis) was used in the analysis. The bivariate analysis examined the association between the outcome and predictor variables. Multinomial logistics regression analysis was then used to examine which factors significantly influenced sexual communication between adolescents and their parents. All the variables in the multinomial logistic regression model were entered in one step. Our reference category in the regression analysis was sexual communication with no parents. SPSS was used to run all analysis, and the results of the multinomial logistic regression analysis were presented as odd ratios.