“I enjoy sex more without condoms": revisiting barriers and facilitators of consistent condom use among adolescents and young adults in South Africa

Background Previous studies have examined consistent condom use correlates in South Africa, focusing on sociodemographic factors, HIV risk perceptions, relationship conict, multiple sexual partners, and masculinity. However, the effect of family nancial support, HIV testing, partner communication and self-ecacy for HIV prevention is less studied. We drew from a cross-sectional survey to address this gap and highlight the key barriers and facilitators of consistent condom use among young people. Methods We analysed data obtained from 631 unmarried sexually active male and females students selected using stratied sampling from a university in Eastern Cape Province of South Africa. Consistent condom use was dened as regular use of condoms in all sexual encounters in the past year. We used an open-ended question to probe the reasons for inconsistent condom use. Adjusted and unadjusted regression analysis were tted to examine factors associated with consistent condom use. gender and differences. After family partner young people on the need for partner communication about HIV, HIV testing uptake, and condom use should happen more rigorously in school and through the media to improve their consistent use of condoms.


Introduction
South Africa remains the epicentre of the HIV pandemic, with 7.7 million people living with HIV and 240,000 new infections every year [1]. Horizontal transmission remains the predominant route for HIV transmission in the country, particularly among young people [2]. Compelling scienti c evidence has shown that the correct and consistent use of condoms during sexual intercourse represents one of the most viable means of preventing HIV and sexually transmitted infections (STIs) [3,4]. In South Africa, condoms are widely available at no cost to young adults through several outlets, such as youth centres and public health facilities [5]. An estimated 87% of young adults in the country reported that it would be effortless to obtain condoms [5]. could in uence consistent condom use. Also, partner communication could facilitate regular use of condoms; however, the link between partner communication about HIV/STIs and consistent condom use has only been sparsely explored. Also, previous studies have mostly focused on prevalence and correlates of consistent use without probing the reasons for inconsistent condom use. This is a miss opportunity given that asking an open-ended question probing reasons for inconsistent use of condoms among a large sample of non and inconsistent users could produce new insights to enhance our understanding of barriers to consistent condom use. Our study addresses this gap by examining barriers and facilitators of consistent condom use among sexually active adolescents and young adults aged 17-24 years. Our ndings could be critical in developing HIV prevention strategies and programs to mitigate the spread of HIV among young people in South Africa.

Methods
We analysed the data from the adolescents and young women's sexual health survey conducted among students of a tertiary institution in Eastern Cape, South Africa between June and November 2018. The study was conducted among adolescents and young women aged 16-24, given the heightened risk of HIV acquisition in this cohort [1]. Despite the signi cant progress in reducing new HIV infection, young people, especially girls, remain disproportionately at risk of HIV infection [1]. Students were chosen for this study conveniently due to the paucity of funds for a household survey. We conducted the study in a government-owned university in Eastern Cape Province, South Africa. Students were included if they were within the age of 16 and 24 years and a university student at selected campus. Students visiting from other universities were ineligible for this study. Respondents were selected using strati ed sampling.
Gender, year of study, faculty of study were the strati cation unit. We selected a proportionate sample size per the strata's population (See Table 1). More females were included in the study, given that the ratio of male to female students is 2:3. The full details of the study methodology have been presented elsewhere [26,27]. Overall, 833 respondents took part in the main study. However, only 631 respondents who were not married but sexually active in the past one year were included in the study. For this study, a sample size of 631 was deemed appropriate at ±3.96% precision level, a 95% con dence level, 39% prevalence of consistent condom use [25] and a population of 15,000, adjusted for 10% possible attrition. Well trained research assistants who were graduate students administered the study instrument using Open Data Kits (ODK) application installed on android device. The use of ODK granted the respondents privacy to complete the study questionnaire and without distraction from and interference of friends. Participants were recruited over six months in their lecture rooms. They were approached and invited to participate in the study and those who consented completed the instrument either on their android enable phones or on the research assistants' phone in private spaces earmarked for the study. The response rate was high, at 89%.

Ethical considerations
The ethical review committee of the University of Fort Hare approved the study protocol. All Institutional Review Board (IRB) guidelines for conducting human subject research were followed throughout the study. Participation was voluntary, and respondents provided written consent indicating they understood the study's purpose and the use of data obtained. Respondents reserved the right not to answer any questions they were uncomfortable with and to withdraw their participation from the study altogether. The right of respondents to privacy, con dentiality, and anonymity was protected throughout the study.
No personal identifying information was collected.

Measures
This study's main outcome was consistent condom use, de ned as the use of condoms at every sexual encounter over the past year. We asked sexually active respondents if they used condoms regularly at every sexual intercourse in the past year, and dichotomised their responses as "yes" or "no". We probed for reasons for not consistently using condoms with an open-ended question. Responses were coded and grouped under six main themes.

Independent variables
We included three main independent variables-partner communication, HIV prevention and condoms self-e cacy -in this study. To measure partner communications, we asked respondents if they had discussed HIV/STIs with their partners, knew their partner's HIV status, and if their partner knew their serostatus. Responses to these questions were coded as yes or no. HIV prevention self-e cacy was measured by asking participants if they feel con dent about their ability to prevent themselves from contracting HIV. Condom self-e cacy was measured with 13 items rated on a 3-point Likert scale with categories 1 (agree), 2 (Neutral) and 3 (Agree). This self-rated construct has been previously validated by Barkley and colleagues [28]. The construct includes questions like "I would feel con dent in my ability to put a condom on myself or my partner, purchase condoms, carry condoms, remove condoms after sex, refuse to have sex with someone not willing to use condoms and insist on condom use every time I have sex. We assess the construct's internal consistency and reliability in our study setting, and the analysis yielded a Cronbach alpha reliability estimate of 0.81. Responses to reverse questions were recoded such that all questions align in one direction with lower scores indicating high condom self-e cacy and higher score indicating low self-e cacy. Thirty-nine scores were obtainable, and 1 to 19 was categorised as higher condom self-e cacy while 20 and above were categorised as low self-e cacy.

Covariates
We included three main categories of covariates, including demographic factors (age and sex), family factors (family structure and family support) and behavioural factors (HIV testing and alcohol use). Age was measured as a continuous variable but later categorised as 16-19 years and 20-24 years. Family structure was coded as single-parent family, two-parents family and foster family. Family nancial support is de ned as the perceived rating of support received from parents. This was a substitute for the measure of parental socioeconomic status. Respondents rated the nancial support they got from their parents as adequate, moderate, insu cient, no support. We only included the current use of alcohol in this study by categorising respondents into current users and non-users. Current users are those who drank alcohol in the past year. Lastly, we measured HIV testing by asking if respondents had ever tested for HIV or not. Lastly, we categorised respondents into six categories based on their sexual behaviour in the past year. These categories include one steady partner only, two steady partners, three or more steady partners, one casual partner only, concurrent main and casual partner, multiple concurrent partners.

Data analysis
We used Stata version 15 to analyse the dataset. We performed descriptive statistics for all variables of interests. Speci cally, we computed mean and standard deviation for age and frequency counts and percentages for all categorical variables. All analyses were strati ed by gender. Also, we used adjusted and unadjusted logistics regression to examine the effect of partner communication, HIV prevention selfe cacy and condom self-e cacy on consistent condom use. We strati ed the adjusted and unadjusted regression analyses by gender to examine the pattern of the effect of partner communication, HIV prevention self-e cacy and condom self-e cacy on consistent condom use among males and females. The analysis was performed at 95% con dence level, and p-values <0.05 were considered statistically signi cant. We coded responses to the open-ended questions on reasons for inconsistent condom use. Codes were reviewed by the research team and similar codes were merged under a common theme. We discussed each theme and used verbatim quotes to support out descriptions. Also, we used simple frequency to indicate how frequent participants mentioned them.

Descriptive ndings
Participants mean age was 21.23 years (SD ±1.70). As presented in Table 2, most respondents were aged 20-24 years (82.6%), received inadequate family nancial support (63.7%), have ever tested for HIV (70.2%), felt con dent in their ability to prevent themselves from contracting HIV/STIs (71.8%). Less than half of the respondents were from a single-parent family (44.9%). More females (76.4%) had tested for HIV compared to males (60.4%). Over half of the respondents were current alcohol users (56.3%), including 46.4% of females and 52.1% of males. Slightly over half (52.8%) of the respondents had ever discussed HIV/STIs with their partners. However, partner communication about HIV/STIs is higher among more females (55.4%) than males (48.6%). Only two-fth of both male and female respondents knew their partner's HIV status. However, close to half of the respondents (48.0%) a rmed that their partner knew their status. A higher proportion of females (52.9%) a rmed that their partner knew their status than males (40.4%). Lastly, about half of the male and female respondents had a high condom selfe cacy score.  Table 3, there was no signi cant age difference in condom use consistency proportion. However, consistent condom use prevalence was higher among respondents from a foster family (45.7%) relative to single-parent families (36.8%). This difference is even more pronounced in the strati ed analysis for females, where over half (52.9%) of those from a foster family reported using condoms consistently compared to only 36.2% of those from a single-parent family. Even though a higher proportion of males from a two-parent family (43.4%) than those from a foster family (32.4%) reported consistent condom use, the difference was not statistically signi cant. Respondents who received adequate family nancial support (53.3%) reported a higher rate of consistent condom use than those who received inadequate nancial support (31.3%). A similar pattern was observed among males and females in the strati ed analysis. Individuals who had tested for HIV (44.0%) reported a higher rate of consistent condom use than those who had never tested for HIV (28.2%). The pattern holds among males and females in the strati ed analysis. The prevalence of consistent condom use was lower among current alcohol users (34.5%) than non-users (44.9%), and this result was consistent for both male and females in the strati ed analysis.
Young people who had discussed HIV/STIs with their partners (42.0%) tend to use condoms more consistently than those who never did (36.2%). This is especially true among males, where 46.2% of those who had discussed HIV/STIs with their partners reported consistent condom use compared to only 32.5% of those who never did. There was, however, no difference in the proportion of consistent condom users by the discussion of HIV/STIs with sexual partner among females. A higher proportion of those who knew their partner's status (42.1%) than those who did not (37.4%) reported using condoms consistently, but the difference was not statistically signi cant.
Similarly, those whose partner knew their status (45.2%) were signi cantly more likely to use condoms than those whose partner was unaware of their status (33.8%). Over half of males whose partner knew their status used condoms consistently compared to only 31.5% of those whose partners were unaware of their status. Whereas, the rate was slightly lower among females with 42.7% of those whose partner knew their HIV status, reporting consistent condom use compared to 35.7% of those whose partners were unaware. Young people who reported having self-e cacy for HIV prevention (45.3%) were signi cantly more likely to use condom consistently than those who reported no con dence in their ability to prevent HIV (24.2%). The pattern persists among males and females. Also, young people with high condom selfe cacy (49.4%) were signi cantly more likely to use condoms consistently than those with low condom self-e cacy (29.5%). This was true among males as well as females in the strati ed analysis.

Multivariable ndings
The results of the adjusted and unadjusted logistic regression are presented in Tables 4. In the unadjusted model, age, family structure and gender were not signi cantly associated with consistent condom use. This pattern persists in the adjusted model.
Alcohol use was associated with lower odds of consistent condom use in the unadjusted model. The direction and magnitude of the association persist in the adjusted model. Also, adequate family nancial support was associated with higher odds of consistent condom use in the unadjusted model. The direction and magnitude of effect remained in the adjusted model, indicating that adequate family nancial support could facilitate consistent condom use. Similarly, having tested for HIV was associated with a higher likelihood of consistent condom use in the unadjusted regression model. However, while the direction of effect persists after adjusting for relevant covariates, the magnitude of effect reduced signi cantly, and the con dence interval crosses one.
On partner communication, discussion of HIV/STIs with partners was not signi cantly associated with a higher likelihood of consistent condom use in both unadjusted and adjusted models. Similarly, knowing one's partner's status was not signi cantly associated with consistent condom use for all respondents in both the unadjusted and adjusted models. However, revealing one's HIV status to one's partner was signi cantly associated with higher odds of consistent condom use. The magnitude and direction of effect remained in the adjusted model, providing clear and robust evidence of the association of informing partner of one's HIV status with higher odds of consistent condom use.
The unadjusted model also shows that self-e cacy for HIV prevention was associated with a higher probability of consistent condom use. The direction and magnitude of effect persist in the adjusted model, suggesting that self-e cacy for HIV prevention was signi cantly associated with a higher likelihood of consistent condom use. Similarly, low self-e cacy for condom use was associated with lower odds of consistent condom use for all in the unadjusted model. The direction and magnitude of effect persist in the adjusted model for all participants.  (Table 5). This group of non-users believed that condoms reduce or hinder their ability to derive maximum pleasure from sex. Lack of availability of condoms was the main reason for inconsistent condom use for about one-third of non-users. They explained that sex is often unplanned and sometimes happens so fast that they often forget to use condoms. Also, unplanned sex means condoms are unavailable and sex could not wait until they are able to access condoms.
Also, trust was among the main reason for inconsistent condom use for about one-fth of those who did not use condoms consistently. HIV testing played no role in trusting their partner, rather engaging in committed relationship and delity to and perceived delity of their partner motivated them to engage in condomless sex. Only a few of them had tested for HIV and considered themselves to carry no risk of contracting HIV from their partner.
Another one-tenth did not use condoms consistently because they relied on hormonal contraceptives.
Preventing pregnancy was the primary concern of this group of inconsistent condom users and hormonal contraceptives help them realise this goal. They appear to be more concerned about preventing pregnancy and less concerned about preventing HIV/STIs. A few of them (7.0%) however blamed their partner for refusing to use condoms.

Discussion
In a high HIV prevalent setting, consistent condom use promotion is one of the tools to combat the spread of HIV, especially among adolescent and young adults (aged 15-24) known to have the highest rate of new infections. Our study examined the prevalence, barriers and facilitators of consistent condom use among sexually active adolescents and young adults. We found that only two-fths of our study participants consistently used condoms in the past year, which is in line with previous studies [29,30] among young people in South Africa. However, a study conducted reported a lower prevalence among young women in rural settings [31]. In contrast, a study among recently circumcised males and another conducted among young people in Cape Town reported a slightly higher consistent condom use prevalence [6,32].
Surprisingly, this study's nding is similar to previous studies in other settings with a relatively low HIV prevalence [25,33] compared with South Africa. Even though the need to prevent STIs and unintended pregnancy is universal, we expected that the prevalence of consistent condoms use among young people would be higher in South Africa than those reported among a similar population [25] in a low prevalent setting, like Nigeria. Contrary to a previous study in South Africa [30], we did not observe any gender differences in condom use consistency. Also, the factors associated with consistent condom use are similar for both males and females, underscoring the absence of differences in condom behaviours of young men and women.
This study highlighted factors that facilitate consistent condom use, some of which have been reported in previous studies [25,[30][31][32]. We found that family nancial support, disclosure of status to partner, HIV testing, and self-e cacy to prevent HIV status positively impact consistent condom use. While the in uence and importance of condom self-e cacy are well established in the literature, our study highlights the considerable importance of family nancial support, partner communication, HIV testing uptake and self-e cacy to prevent HIV/STIs. Previous studies have not explored the link between family nancial support and condom use consistency. Our plausible explanation for why young people who received adequate nancial support use condoms more consistently than those who did not is that they may have su cient resources to purchase condoms than their counterparts. They could, for instance, purchase and keep their preferred condoms brands in their rooms in anticipation of sex, unlike their counterparts who may have to ration their scarce resources and advance purchase of condoms may not be among their priorities. It is worth noting that condoms are distributed on campus, but this happens infrequently and may not go far enough in solving condoms unavailability problem.
Previous studies have highlighted the importance of partner communication on protective sexual behaviours [25,34]. We know from these studies that partners who discussed the need to use condoms and prevent unintended pregnancy are more likely to engage in protective sex. Our study further adds to the emerging evidence of the importance of partner communication for HIV/STIs and unintended pregnancy prevention. Partner communication on HIV/STIs means that couples would risk and resort to the appropriate preventive methods.
Consistent with the literature [25,30], our study shows that HIV testing encourages protective behaviours.
Those who have tested are more likely to use condoms consistently than those who never tested for HIV. The difference observed could be due to the positive impact of the counselling before and after HIV testing. Also, uptake of HIV testing is a positive health behaviour, suggesting that most people who undertook HIV testing are more concerned about preventing HIV than those who had never been tested.
However, no studies have examined the in uence of HIV prevention self-e cacy. Expression of high selfe cacy for HIV prevention re ects that young people are not only aware of methods of preventing HIV/STIs but also feel con dent in their ability to protect themselves. Therefore, it is not surprising to see that those who expressed con dence in their ability to prevent HIV/STIs are more likely to use condoms consistently.
Based on our ndings, the desire for maximum pleasure, preference for sex without condoms, unavailability of condoms, partners' objections, alcohol use, and hormonal contraception use constitute barriers to consistent condom use. Like previous studies, our study demonstrates that young people who believe that condoms reduce sex pleasure are unlikely to use it regularly, if at all. It appears that the desire for maximum fun trumps the need to prevent HIV/STIs for many young people. Also, in line with previous studies, condom unavailability was reported as part of the factors hindering condom use consistency [25,30]. For many, sexual intercourse is unplanned, and in such situations, condoms are not always available.
Rather than wait to source for condoms, some young people often take the risk, ignoring the possibility of contracting HIV/STIs.
Trust has also been documented in the literature as one of the barriers to condom use consistency among young people. In this study, young people failed to use condoms consistently because they trust their partners to remain faithful in the relationship. The trust is based on perceived delity in a relationship. Since some already tested for HIV and were on hormonal contraception, they do not see the need to use condoms. For others, the trust could be described as a "blind trust" given that it was implied despite having never tested for HIV. In these relationships, unprotected sex is equated with trust and initiating a conversation about condoms could be misconstrued as distrust [34]. Our result indicated that those who failed to use condoms at all or inconsistently because they were on hormonal contraception suggest that some young people are more concerned with pregnancy prevention than HIV/STI prevention. Our ndings on the link between alcohol use and condom use consistency further add to the literature on the adverse effect of alcohol on high-risk sexual behaviours. Excessive use of alcohol could make young people inebriated, therefore, lacking agency to negotiate condom use.
In light of our study ndings, there is a need to invest in multi-prongs interventions that begin with early and comprehensive sex education. This sex education should continue beyond high school and should happen in other settings outside of school to reach out of school adolescents and young adults. In addition to sexuality education, the government should continue to make condoms and other contraceptives freely available through the health department. The government initiative to make different varieties and avours of condoms available is a positive development that could impact condom use consistency, given the complaint about pleasure concerns. Interventions need to be innovative and convenient while addressing these identi ed barriers to reach these adolescents and young adults. Counselling of students has historically been the standard intervention in safe sex education [36]. However, many young people appear not to use condoms despite counselling. To address this need, innovation on behavioural interventions are critical. Social marketing campaigns on HIV are necessary to target these groups, and therefore should be implemented more rigorously. Integrating interventions into the routine standard of care in South Africa is necessary to promote behaviour change.

Study strengths and limitations
This study adds to the literature, highlighting the barriers and facilitators of condom use consistency in a high HIV prevalent setting. The use of the mixed-methods approach is a strength of this study.
Nevertheless, our study is not without some limitations. Our study population is not representative of adolescents and young adults in the country, given that they are more educated, limiting the generalisability of our ndings. However, our ndings are consistent with previous studies among young people living in rural and urban settings [29,30,32], suggesting that our results may apply to other young people in the country. Even though we ensure privacy, anonymity and con dentiality throughout the study and with our use of ODK, we could not completely rule out social desirability bias, which could lead to over-estimation of the prevalence of consistent condom use.

Conclusion
Inconsistent condom use remains a challenge among unmarried sexually active young people in South Africa. Alcohol use, trust, spontaneous sex, use of hormonal contraception, partner's objection, pleasure concerns constitute barriers to consistent condom use. High condom self-e cacy, partner communication, HIV testing and self-e cacy for HIV prevention could facilitate consistent condom use. Education of young people on the need for partner communication about HIV, HIV testing uptake, and condom use should happen more rigorously in school and through the media to improve their consistent use of condoms. Also, interventions focusing on increasing self-e cacy for condom use and HIV prevention and risk reduction activities such as provisioning of condoms are crucial to enhancing condom use and plausibly tackling the HIV epidemic among young men and women in South Africa.