Trends and gender differences in age at sex debut among adolescents and young adults in urban Cape Area, South Africa

Background Adolescents and young adults in South Africa are at increased risk of experiencing negative outcomes from early sex initiation due to the context they develop in. This study aimed at investigating trends and gender differences in early age at sexual debut. Methods Data from the Cape Area Panel Study (CAPS) of young adults' conducted between 2002 and 2009 in urban Cape Town were used. Correlates of early sex using ecological risk factors were analysed. Results Overall mean age at sexual debut at follow-up was 17.5 years (standard deviation (SD) =2.5), with males at 16.8 years (SD=2.5) and 18.1 years (SD=2.4) for females. Males consistently reported an earlier age at sex debut across the five waves of the survey. Significant risk factors for early sex appeared at all levels of the ecology to include individual, household, and community factors. Conclusions Integrated interventions for delaying early sex debut should consider factors within the ecology of the young adults' development context.


Introduction
Initiation into sexual activity among adolescents and young adults is an important marker of transition into adulthood and is usually embedded with social and personal meanings 1,2 . Early age at first sex (before 15 years as per WHO definition) exposes an adolescent to the risk of unwanted pregnancy and HIV infection alongside other sexually transmitted diseases [3][4][5][6] . Among teenagers in the US, age at first sex has steadily decreased since the 1980, and the proportion reporting sex before the age of 15 has been on the rise 7 . In South Africa, reports indicate that early sex debut is a major contributor to disease and disability among the youths, as calculated through disability-adjusted life years 6,8,9 . Understanding circumstances of early sexual debut among adolescents and young adults has important implications for theory, policy and practice. South Africa has the highest number of people living with HIV (PLHIV) of any country in the world as of 2016 with an estimated 7.1 million PLHIV and young people are disproportionally affected 10 . Even though South African youth make less than 1% of the world's youth, the country accounts for more than 15% of global youths living with HIV. At the same time, South Africa has one of the highest rates of teenage pregnancies around the globe. More than 35% of adolescents reportedly become pregnant before the age of 20 years with more than 30% giving birth at least once [11][12][13] . Even though a high rate of school resumption and completion has been reported for teenage mothers in South Africa compared to other countries of sub-Saharan Africa, disruption from schooling is associated with truncated educational attainment and lower future economic returns [14][15][16] . Furthermore, early childbearing put young women at risk of morbidity and mortality resulting from birth complications [17][18][19] .
Studies reporting on determinants of early sex debut have found variation in rural-urban residence, race, socio-economic status, academic performance, and peer influence on adolescent's sexual debut [20][21][22][23] . In South Africa, it is reported that males begin sexual activity earlier than females, and young adults of the Black/African race begin sexual activity at a significantly younger age compared to the Whites or Coloureds 24 . Other studies have reported on ecological factors that inform on adolescents' decision to engage in early sexual activities including family composition, staying with a single parent, parental supervision, perceved risk of STIs, available community health facilities, and media influence 22,25 . In addition, lower family incomes, living in disadvantaged neighbourhoods, lacking parental supervision, and community permissive attitudes have also been reported 24,26 . However, many of these findings emanate from studies conducted in developed countries, and even then, the contribution of risk factors has been reported in isolation.
In this study, we report findings of using an ecological perspective framework to investigate trends and gender differences in early sex activities. The ecological systems framework as proposed by Bronfenbrenner (1979), groups risk factors into the individual, household, and the community levels. The cumulative risk approach as proposed by Small and Luster (1994) is tested to study whether the effect of ecological risk factors is cumulative 27-29 . Using this approach, it is assumed that risk factors occurring at the multiple levels of the ecology act cumulatively to influence an adolescent' sexual behaviours, rather than risk factors acting in isolation30. Understanding circumstances of initiating into sexual activity is important, because it may be forced, and is a potential hazard for pregnancy and disease transmission 1,2 Data and methods Study design The Cape Area Panel Study (CAPS) was a longitudinal survey of young adults in urban Cape Town that was conducted as a collaborative project of the University of Michigan and the University of Cape Town. The CAPS was conducted in waves 1 to 5 between 2002 and 2009 by the time of these analyses. The CAPS sampling plan was designed to produce a young adult sample that was representative of the population aged between 14 and 22 years-old in metropolitan Cape Town. The surveys were conducted in waves with the first wave conducted between August and December, 2002. A third of the adolescents were again interviewed in 2003 (2a), and the remaining two-thirds re-visited in 2004 (wave 2b). The full youth sample was again re-interviewed in 2005 (wave 3) and 2006 (wave 4). The last wave was a face-to-face interview of any youth interviewed previously and was conducted in 2009 (wave 5).

Participants and setting
An estimated 4,800 young adults aged between 14 and 22 years were randomly selected and interviewed in the first wave of the CAPS survey conducted in 2002. The survey also collected information on all members of these young people's households. Additionally, a random sample of households that did not have members of the age grop 14 to 22 years was also interviewed. The young adults were followed-up in successive surveys with the latest conducted in 2009. The study was conducted in metropolitan area of the City of Cape Town in the Western Cape province of South Africa.

Procedures
Questionnaires were administered to the adolescents and young adults to collect current and retrospective data about themselves, their households, and their communities. Data are captured in the form of modules detailing life-history, schooling, employment, sexual behaviours, and family relationships.

Outcome measures
Early age at sex debut was the dependent variable in this study and was defined as sex intercourse before the age of 15 years. In the CAPS surveys, young adults reporting sexual activity also reported age at firs sex. We restricted our outcome to only young adults transitioning into sexual activity during follow-up to establish a causal relationship between baseline risk factors and onset of early sexual activity.

Risk factors
Baseline risk factors are categorised into the individual, household, and community levels. At the individual level, factors reported included race, school achievements and aspirations, history of sexual abuse, marital status, employment, family size, ability to make personal decisions on health, and knowledge of HIV/AIDS. At the household level, risk factors included socio-economic status of the family, living with single or both parents, parental monitoring and supervision, closeness to the parents, and parents' education attainment. At the community level, mean age at marriage for men and women, socio-economic profile of the community including proportion of employed and education status of men and women. Furthermore, knowledge and attitudes towards HIV, religiosity, graduation rates, peer influence, and availability of health facilities in the community will be considered.

Statistical analysis
Statistical analysis and data manipulation were carried out in STATA version 13 (STATA Corporation, College Station, TX). Proportions, means and standard deviations were used to summarize the data, while cross-tabulations and Chi-square tests were applied for differences between bivariate categorical groups with associated p-values reported. For multivariate analysis, discriminant function analysis (DA) was used to describe and classify participants into early and late sexual debutants groups. Assumptions of the DA were met as the sample was randomly selected thus independence of the observations, multivariate normality of the observations, and equal covariance matrices for the two groups under investigations. A test of differences between the categorical outcome and multivariate predictor variables used the F-value tests of significance where resulting p-values were reported. A cumulative risk index was constructed by summing the number of risk factors that were found to be significant in the discriminant analyses. The proportion of sexually active young people was plotted against the number of risk factors reported to determine the effect of increasing risk factors on engaging in early sex debut. These analyses were performed separately for males and females, with risk factors occurring at the individual, household, and community levels.

Ethical issues
This study used CAPS datasets and has acknowledged the CAPS survey team at the Centre for Social Science Research, University of Cape Town.

Results
Of the 4,704 adolescents and young adults responding to the young adult questionnaire by wave 1 survey of 2002, a total of 2,162 (46%) reported to have experienced sexual debut and were excluded from further analysis. Out of the 2,542 adolescents who were sexually inexperienced at baseline, 1,712 (67.4%) transitioned into sexual activity by wave 5 survey of 2009. Overall mean age at sexual debut at follow-up was 17.5 years (SD=2.5), with males at 16.8 years (SD=2.5) and 18.1 years (SD=2.4) for females. Table 1 and Table 2 report on the descriptive statistics of the analysis sample separately conducted for males and females. For males, sexual experience was reported at 46% at Wave 1 survey increasing to 95% by the Wave 5 survey as reported in Table 1. Majority of males reported to have had consensual sex relationships throughout the surveys, with 96% of males reporting sexual experience at baseline, having been willing to have first sex. to females, reported consenting to first sex. The mean age at first sex for males was 15.7 (SD 1.8) years for females and 16.6 (SD 1.7) years for females at Wave 1, implying that males became sexually experienced at a younger age compared to females. Figure 1 shows that over the follow-up, males consistently reported lower average age at first sex.  ported in Wave 1. In summary, many of the independent factors were significantly different between males and females, and further analysis were conducted stratified by gender.

Correlates of early sexual activity
Further analyses were conducted for adolescents and young adults who were not sexually experienced at baseline, and were followed-up to monitor transitioning into sexual activity by the end of the wave 5 survey. Of the 1,712 adolescents experiencing sexual debut during follow-up, 1,615 had non-missing age at first sex and were used for these analyses, with 140 (9%) reporting early sexual activity. Early sexual debut significantly differed by gender, with 14% of males reporting early sex compared to 4% of females (p<0.01). A significantly higher proportion of African blacks reported early sexual debut at 15% compared to Coloured at 4.6% and Whites at 3.6% (p<0.01). Table 3 reports on transitions into sexual activity of the adolescent and young adults' sample. Significant individual level risk factors for sexual transitions included gender, population group, years of schooling, having worked in the last year, and expecting to work in the future. At the community level, households' employment, female headed households, dwelling in informal settings, households with African population, and households reporting to be below poverty lines. Table 4 reports on means and standardised scores from Discriminant analyses for considered risk factors.  Table 4 reports on results from the multivariate analyses on factors discriminating against early age at sexual debut.
In males, significant factors at the individual level included; population group, years in education, age of first sex partner, and having worked in the last year. At the household level, parents ever married, and father spending time in the household were mildly important but had no statistically significant effect on early sexual debut. Significant community factors included employment levels, female headed households, and informal dwellings in the community, poverty levels, and proportion of families classified as Africans. Nine risk factors were selected, and only seven were found to contribute significantly to the group classification in the multivariate analysis. Age of first sex partner and years in education were important factors at the individual level, while proportion of unemployed households' heads, informal dwellings, households headed by a female, households with individual Africans, and families below poverty line were major determinants at the community level. For females, individual level factors discriminating early sexual debut included years of education, population group, currently being in school, age of first sex partner, and having worked in the last year. At the household level, number of siblings, parents married, spending time with the mother, siblings who helped with homework, number of meals in day, and most influence on future jobs. At the community level, proportion of household heads unemployed, living in informal dwellings, proportion of the community below poverty level, and proportion of households classified as black Africans had significant effect on early sexual activity. Six risk factors were found to discriminate significantly the group classification in the multivariate analyses. At the individual level, age of first sex partner and years in education were important, while only parents ever married was significant at the household level. The proportion of unemployed household heads, proportion of household with individual Africans, and proportion of households below poverty line were important at the community level.

Discussion
Previous studies reporting on age at first sex have identified slight gender differences at mean age of first sex, with males becoming sexually experienced much earlier compared to females 8,31-33 . Pettifor et al observed that among [15][16][17][18][19] year old respondents from a nationally representa-tive survey in South Africa, 50% of males, and 47% of females reported ever having sex 34 . Further, reporting that mean age at first sex was at 16.4 years for males, and 16.8 years for females, findings that were consistent with this study. However, this study does not identify substantial gender differences in terms of sexual experience at both Wave 1 and Wave 5. Findings from the current study on trends and genders differences may indicate an increase in the age of sex debut and are consistent with other reported studies 35 . Contrary to previous findings where gender inequality for young women was manifested through sexual coercion and partnering with older men, these findings suggest that young women in South Africa mainly had first sex with partners of similar age 36 . Reported sexual partner at first sex was mainly the boyfriend or girlfriend of the respondent, reporting 88% of all the time. Majority of females had first sex in committed relationships either in marriage, or when engaged to partners or living together. Therefore, safe sexual messages could be targeted to schooling institutions, mainly targeting similar age-group relationships as these are more frequently exercised and have the potential to be most effective in delaying sex debut. In summary, for school attending respondents, their first sexual partners were likely to be their girlfriends or boyfriends in the same age group and could be targeted with appropriate interventions.

Limitations
These data lacked important indicators of peer influence, which might further explain on findings of sexual relationships among respondents of the same age group. In addition, studies of sexual behaviours may be prone to social desirability biases that may distort responses, differences that may be exacerbated by gender of respondents.

Conclusion
This study finds that a high proportion of adolescents were initiating sexual activity at an early age and that gender differences in early sex persisted throughout the study period. Other important indicators of early sexual debut identified were age of first sex partner, population group, years in education, and markers of community socio-economic status. These findings have also been previously reported in studies conducted in South Africa and other developed countries 13, 37 . The presence of risk factors at African Health Sciences Vol 19 Issue 4, December, 2019 multiple levels of the ecology suggests that interventions aimed at delaying early sexual debut should not only target individual attributes but also both the household and community context.
This study was significant in that analysing ecological factors for early sexual activities adds significantly informs on interventions to improve on sexual and reproductive health outcomes for adolescents and young adults in South Africa.

Acknowledgement
The Authors' contributions E.M and C.O conceived the idea. E.M performed data analyses, derived the models. All authors wrote the manuscript and provided critical feedback that shaped the manuscript.