Introduction
Condom use is effective in preventing the transmission of human immunodeficiency virus (HIV) and sexually transmitted infections (STIs) [
1]. However, the proportion of condomless sex among college students remains high [
2]. Currently, condomless sexual transmission remains the main route of HIV transmission among college students [
3]. The number of newly diagnosed people living with HIV among college students in China was increasing at a rate of 30–50% per year [
4]. In addition, a survey of 35,383 unmarried female college students in China found that the proportion of unintended pregnancy was 31.8%, of which 53.5% experienced two or more pregnancies [
5]. The current situation has reminded us that condomless sex among college students is still an issue that deserves attention. With greater self-control and adaptability, college students had the opportunity to change their behavior. Therefore, understanding factors associated with condomless sex at this stage could help college students to break their risky sexual behavior habits in time.
Condom use at the last intercourse is one of the most common measures used to assess condom use [
6]. A large number of studies have shown that condomless sex is influenced by a wide range of factors, including social demographic characteristics, behaviors, substance use, psychological factors and so on [
7‐
11]. For example, condom use at the first sex [
12‐
14], experience of intimate partner violence (IPV) [
8,
15], and self-efficacy of condom use [
10,
16,
17] were significant predictors of condomless sex. Previous studies have usually used logistic regression to access the risk factors for condomless sex, but the effect of different risk factors varies. When many factors coexist, it is important to identify the salient factors and to help individuals recognize or identify their risks based on these factors. Identifying students at risk of condomless sex not only helps to prevent condomless sex at an early stage, but also protects their sexual health. The nomogram is a graphical tool based on a regression model, which can distinguish the differences between each factor [
18]. In the nomogram, each factor corresponds to different numerical points, which could be calculated to obtain the total score of each factor for each individual in relation to the risk of the event, reflecting their personal risk. Nowadays, the nomogram has been widely used to predict a variety of clinical outcomes [
19‐
23], which formed the basis for the formulating treatment cases and patient management. However, it has rarely been applied school health settings. The nomogram can quantify the risk of individual events without the need for complex formula calculations, which is easily accepted by health educators and facilitates better health education. Therefore, a cross-sectional survey was conducted to explore the significant predictors of condomless sex, and then nomogram to predict the individual possibility of condomless sex at last intercourse was developed and validated among college students in Zhuhai, China.
Discussions
In this cross-sectional survey, the proportion of condomless sex at last intercourse reported in our study (18.2%) was lower than that of the students with similar demographics from other universities in 15 provinces and cities in China (27.4%) [
35], and lower than that of the students from the Philippines (83.0%), Indonesia (61.3%), Laos (62.3%), Thailand (58.2%), Singapore (57.8%), Myanmar (48.3%) and Cambodia (42.6%), but higher than that of Malaysia (8.0%) and Vietnam (13.6%) [
2]. The different proportions of condomless sex at last intercourse among college students could be attributed to differences in study settings, populations, health education policies, and so on [
10,
36]. Efforts were still needed to make to spread knowledge about safe sexual health among college students to facilitate the implementation of the Healthy China 2030.
Our results showed that students who had heterosexual intercourse were less likely to have condomless sex, which we need to explain in detail. Because the whole population in this study had sex in the last year, individuals who had heterosexual intercourse included those who had only heterosexual intercourse and those who had both heterosexual and homosexual intercourse. Students who had not had heterosexual intercourse refer to those who had only had homosexual intercourse. Therefore, it could be concluded from our results that students who had only heterosexual intercourse and those who had both heterosexual and homosexual intercourse were less likely to have condomless sex than those who had only homosexual intercourse, which is similar to what was reported in the study by Cathy Maulsby [
37]. This may be because many students still believed that the purpose of condom use was contraception, which increased the likelihood of condomless sex among students who had had homosexual intercourse [
37]. It is worth noting that having anal sex was also a risk factor for condomless sex in our study. Given the two factors above, we could further conclude that having had homosexual anal sex was associated with a higher likelihood of having had condomless sex. This finding reminded the universities that it was men who had sex with men who were the focus of the students.
Consistent with existing studies [
12‐
14], our study showed that condom use at first sex was a protective factor for condomless sex. There may be a cognitive link between sexual behavior and condom use at first sex, with students choosing to use condoms at subsequent sex based on the habits of their first sexual experience [
12‐
14]. In addition, condom use at first sex could lead to a reduction in the risk of STIs, and this feedback helped students to use condoms the next time [
38]. Previous studies in different populations have suggested that victims of IPV had a higher risk of condomless sex [
8,
15]. In addition, our study contributed to the existing literature by adding that having more types of IPV was a risk factor for condomless sex among college students, which may be explained by an accumulated effect of physical, psychological, and sexual victimization. It has been reported that IPV may be caused by the power of oppression [
39]. The more types of IPV experienced and the more aspects were oppressed and hurt, the more disadvantaged it was to negotiate safe sex, which was prone to condomless sex [
8].
Our findings were in line with several previous studies conducted among college students, which revealed that higher levels of condom use self-efficacy promoted condom use at last intercourse among college students [
10,
17]. Condom use self-efficacy has been considered a crucial part of many theories of condom use, including the Theory of Reasoned Action (TRA), Social Cognitive Theory (SCT) and the Health Belief Model (HBM) [
28]. In addition, condom use self-efficacy was one of the most important predictors of condom expectancy and actual use [
16]. Our results revealed that the higher the positive attitude towards condom use, the higher the likelihood of condom use. This finding was also reported in a systematic review of 96 studies [
40]. In other words, if an individual feels that the negative consequences of not using a condom (STIs or unintended pregnancy) are unbearable, or if an individual feels that there are benefits to using a condom, then the individual’s attitude towards condom use is positive [
41]. However, it was interesting to note that the norms on condom use were not included in the nomogram in our study, which needs further investigation. The previous studies [
40,
41] mentioned that norms on condom use contributed less to the prediction of condom use compared to attitude and self-efficacy, which may partly explain the underlying reasons.
The findings above reminded us to pay more attention to students who have only had homosexual intercourse or anal sex, especially those who have sex with men. In addition, comprehensive sexuality education should be popularized at an early age to promote the use of condoms at first sex so that students can develop the habit. For students who are victims of IPV, professional and effective services could be provided to protect their physical and mental health and reduce their fear of safe sex. Most importantly, it is necessary to establish positive attitudes towards condom use and improve condom use self-efficacy among college students.
Our study developed an effective and convenient tool to calculate the individualized possibility of condomless sex among college students, which showed high accuracy with an AUROC of 0.83 and good discrimination in predicting condomless sex. The nomogram was a visual graph based on a multivariable logistic regression model, where each variable corresponded to a specific point, and the total score of each variable was calculated to quantify the risk of events in the population. This method is effective and convenient, and can convert intangible risks into visual and quantifiable scores. There were various factors associated with condomless sex, and many college students were unable to recognize or identify their risks and therefore did not take any protective measures. We minimized the influence of multicollinearity using LASSO regression, identifying the predictors from multiple perspectives and presenting them in the nomogram. In our nomogram, each college student could calculate a total score based on their actual risk factors for condomless sex and assess their possibility of having condomless sex, which was a quantitative way of acknowledging their risks. Only by understanding their risks can students better guide their practical actions.
Students played an essential role in education, as recipients and practitioners of sexuality education. Each student was responsible for their own health and had to take responsibility for their own consciousness, behavior and consequences [
42]. If students were aware of their risks, they could receive individualized education to adapt their behavior, change existing risk factors and reduce the subsequent risks. Our nomogram standardized the scores of the risk factors so that students could recognize the contribution of each factor to the risk of condomless sex, find out the most important risk factors, and receive individualized education on behavior modification and habit change to reduce the influence of the risk factors in the right order.
In addition, our nomogram could be a source of information for individualized education. Only by accurately understanding students’ individualized information can the best personalized teaching strategies be developed for the students. Our nomogram which collected information from students, was used to quantify the individual possibility of condomless sex among college students, which could provide specific recommendations for schools to develop learning plans and training programs. In addition, our study also provided the risk subgroups. For universities and colleges, hierarchical management could be applied to the existing management system to correct the risk factors. For students in the low-risk group, regular education could continue, and for the moderate-risk and high-risk groups, more frequent and intensive education and interventions could be implemented. In the long term, the nomogram could also be used to assess the trend in students’ risks. Universities and colleges could use a series of cross-sectional or cohort studies to assess the risk of condomless sex among freshmen and then implement interventions based on hierarchical management. Continuous assessments could be conducted every 6 months or annually among different subgroups to tailor targeted interventions to each subgroup of students.
Some limitations of this study should be noted. First, the cross-sectional design does not allow conclusions to be drawn about causality [
23,
34]. Therefore, prospective studies are needed to verify the predictors of condom use at last sex. Second, the nomogram model was constructed among college students at six universities in Zhuhai City, Guangdong Province, and it needs further external validation among college students in other regions. Although probability proportional to size sampling was used to determine the sample size of each university, in practice there was a gap between the projected proportion and the actual proportion, possibly because the actual number of students drawn from some universities was much larger than the projected number, resulting in a disproportionality and affecting the representation of the sample size. Third, because we relied on self-reported condom use at last sex, we cannot rule out reporting bias, but the anonymity of the questionnaire may reduce the impact of reporting bias.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.