Background
Young people (aged 15–24 years) have contributed to a recently reported reduction in the incidence of sexually transmitted infections (STIs); this is perhaps due to their greater awareness of STIs, including human immunodeficiency virus (HIV), and their increased engagement in safer sexual behaviors [
1]. Despite these data, adolescents are still considered a high-risk population for engaging in risky sexual behavior worldwide [
1,
2]. Research on condom use predictors among adolescents has attracted interest from preventers and clinicians in order to reduce risky sexual behavior in this population.
The interventions with greater empirical evidence regarding the prevention of contracting STIs are based on socio-cognitive theories [
3]. These are social-cognitive models that describe the key variables and determinant interrelationships in predicting health behaviors [
4]. Behavior modification through preventive actions for the social and cognitive factors attempts to engage individuals in healthy behaviors, such as consistent condom use. Among the most widely used theoretical models in the field of HIV/AIDS are the theory of planned behavior (TPB) [
5‐
7], socio-cognitive model (SCM) [
8‐
10], and information-motivation-behavioral skills (IMB) model [
11‐
13].
According to the TPB, intention is the best predictor of behavior, and this is determined by attitudes, normative beliefs toward such behaviors, and perceived control. The attitudinal component is defined as the set of beliefs regarding the behavior’s value and its consequences. Attitude toward condom use is more favorable when negative consequences of not using a condom are valued (STIs or unwanted pregnancy) and the benefits of its use (condom use as an exciting element) were valued as being positive [
4]. The normative component represents the perception of what the reference groups do and the personal motivation to act in accordance with the reference group’s expectations. This construct encompasses individuals’ perceptions of how the reference group behaves (descriptive norms), and what is expected of the individual in relation to such behavior (injunctive norms). An individual will use a condom during sex if they perceive that the peer group does (i.e., acting according to group pressure). Perceived control refers to the perceived ability to perform a behavior depending on resources and personal limitations. Ajzen [
5] suggested that perceived control may also directly predict behavior when the behavior is complex or not under volitional control.
Meta-analytic and review studies provide extensive empirical support of the TPB in predicting condom use and other health behaviors across different populations, such as adolescents and college students [
14‐
20]. In a meta-analytic review of 185 studies, Armitage and Conner [
19] observed that the TPB accounted for 39 and 27 % of the variance in intention and behavior, respectively. Godin and Kok [
14] found similar results in a review of 56 studies testing the applicability of the TPB to health-related behaviors (41 and 34 % explained variance for intention and behavior, respectively). More recently, McEachan and colleagues [
18] concluded that the TPB shows a poorer explanation regarding safer sex (variance explained from 13.8 to 15.3 %) compared to other health behaviors, such as physical activity (24) and diet behaviors (21 %). Generally, perceived norms show a weaker contribution to predict intention compared to attitude and perceived control [
14,
15,
18,
19]. Albarracín et al. [
15] meta-analyzed 96 studies predominantly conducted in Europe and United States, and concluded that attitude is the best predictor of condom use intention (
r = 0.58; β = 0.47) followed by perceived control (
r = 0.45, β = 0.20), and perceived norms (
r = 0.39; β = 0.20). However, perceived norms have shown to be more predictive of the behavior in adolescents compared to adults [
18]. More evidence about the contribution of the intention’s predictors is needed, especially in adolescent samples.
The SCM [
8‐
10] is the most commonly used theory in HIV prevention for young people worldwide [
21]. From the social-cognitive perspective, behavior is determined by the reciprocal, dynamic, and continuous interaction between the personal (cognitive component), behavior (behavioral component), and environment (social component) [
22]. The cognitive component refers to the individual’s confidence in their ability to perform a task or achieve a particular goal (i.e., high self-efficacy to exercise over one’s sexual behavior). The behavioral component encompasses individual consequences of performing a behavior and influences the chances of the learner behaving correctly by practice (i.e., skills necessary to enable an individual physically capable of performing a behavior to obtain the desired outcomes). An application to sexual health could be the valuable skills in managing interpersonal situations and protecting themselves against STIs and unplanned pregnancies [
9]. The social component involves aspects of the environment that improve the individual’s ability to successfully undertake specific behaviors. It refers to vicarious learning and social consequences of behavior. People develop behavioral competences by learning directly from their actions or through social modelling (observing how other people belonging to the same physical and social environments behave and/or consequences of their behavior) [
23]. Adolescents whose friends used condoms would be more likely themselves to use condoms. The SCM has served as a theoretical framework to evaluate the efficacy of HIV-risk reduction interventions and to predict condom use in high-risk populations belonging to low-income cities, such as adolescents, college students, and drug users [
9,
24‐
27].
From a critical review of the theoretical models, Fisher and Fisher [
11‐
13,
28,
29] developed the IMB model, which included relevant social and psychological constructs [
7,
30]. Although the IMB model has a broad application in health promotion, it was initially developed to reduce sexual risk behaviors and HIV infection [
12]. The IMB model has demonstrated its appropriateness as a framework for predicting sexual risk, primarily with the adult population, although also with adolescents [
12,
29,
31,
32]. Most of these studies were conducted in low- and middle-income countries [
33‐
35]. From this approach, information, motivation, and behavioral skills are determinants of HIV-prevention behavior. The information component includes what HIV is, the risk involved in having unprotected sex, and protection methods. Knowledge is a prerequisite in this explanatory model of sexual behavior [
28]. The motivational component refers to the motivation to engage in safe sex and avoid risky behaviors. This includes attitudes toward condom use, perception of social support, and intentions to engage in safe sex. The behavioral skills component comprises the individual’s ability to carry a behavior, and the perceived self-efficacy in the practice of the behavior [
9,
29,
30]. Perceived self-efficacy refers to the perception of control over motivation, thought processes, emotional states, and behavior [
9].
A review of the intervention strategies based on the IMB model for health behavior change concluded that it is applicable for promoting healthy behaviors in chronic patients (e.g., heart disease self-care, diabetes, etc.) and for interventions focusing on reducing risk by promotion of consistent condom use [
36]. The IMB model accounted for the explained condom use variance of 19 % in students from Cape Town, South Africa [
37]. In a cross-sectional study involving 3,183 college students aged 16 to 25 years in China [
38], consistent condom use was significantly predicted by behavioral skills (β = 0.75), although it was not directly related to information and motivation.
These socio-cognitive models—TPB, SCM, and IMB model—specify variables that can determine whether someone will engage in a healthy behavior, such as using condoms during sex. Some of these variables are common in the TPB, SCM, and IMB model; however, their conceptualization may differ across the models [
39]. Each of these theories assumes the relevance of having a positive attitude toward the behavior and peer influence to use condoms during sex. Perceived norms emphasize people’s perception of social normative pressures or other’s beliefs that they should or should not perform a particular behavior in the TPB. However, in the SCM, it refers to vicarious learning and social consequences of behavior. Self-efficacy is a predictor of intention in the TPB, is part of the behavioral factor in the SCM, and is key in the behavioral skills in the IMB model. Self-efficacy refers to the belief in one’s ability to succeed in specific situations. Compared to the TPB and SCM, the IMB model does not emphasize confidence that the action can be undertaken in adverse conditions [
8]. Intention is the best predictor of condom use in the TPB, while it is part of the motivational component in the IMB model. Compared to the TPB, the informational component is present in the IMB model and cognitive factors of the SCM, noting the need to have sufficient knowledge about methods of protection to use condoms during sex.
Extensive empirical evidence supports the adequacy of the TPB, SCM, and IMB model in predicting healthy behaviors in general [
7,
9,
29], particularly condom use in adolescents and college students [
9,
12,
25,
35,
40‐
42]. However, no studies thus far have compared the suitability of these three models for condom use prediction in adolescents. According to Noar and Zimmerman [
43], several reasons justify the comparison of these health behavior theories. First, this allows us to identify the theory most accurate to predicting health behaviors. Second, it examines the similarity and difference across models. Third, it provides relevant information about the relationships among theoretical constructs, and identifies which ones are more relevant to predict the behavior and must be included in health promotion interventions. Although the empirical comparison of health behavior theories is essential in research to move forward, studies with this aim are rarely found in the scientific literature [
39,
43].
The present study aims to investigate which socio-cognitive models—TPB, SCM, or IMB—best predict the frequency of condom use (FCU) in adolescents [
1,
2]. Determining the weight of each variable in predicting the FCU is of great value when designing interventions that promote consistent condom use. This study’s first objective was to determine the applicability and suitability of these theories of health behavior in predicting the FCU among Spanish adolescents. The applicability and suitability of each model was determinate by the model fit indices to predict the primary outcome. A good-fitting model (in terms of the statistical indices of CFI, TLI, and RMSEA) indicated a better applicability and suitability of the model for its purpose. The second objective was to analyze the relationships among the constructs postulated by these models and their contribution to the prediction of the FCU. Based on previous research, we hypothesized that three models will be suitable to predict FCU among Spanish adolescents. According to the theoretical models, the variables involved in predicting condom use will present a high and significant relationship.
Discussion
In this study, three socio-cognitive models for predicting the FCU in Spanish adolescents were evaluated. The results revealed that the TPB best contributed to the prediction of the FCU in this study, at least when compared with the other two models. Furthermore, condom use intention, motivational and cognitive-attitudinal factors (i.e., HIV and STI knowledge, attitude towards condom use, subjective norms), and behavioral skills were involved in predicting FCU in Spanish adolescents.
The three models are comprised of similar constructs, although they offer a different approach to predict health behaviors, including condom use. Compared to the TPB, the SCM and IMB model consider knowledge as a cognitive factor. The SCM has a specific behavioral component, including skills to succeed in a particular situation, such as condom use negotiation. The IMB model also includes the behavioral skills component as a direct predictor of the behavior. It is important to note that the premise that skills, abilities, and environment constructs can moderate the relationship between intention-behavior was not included in the original TPB, yet it is recognized in the lasting formulation of a reasoned action approach, which is the integrative model of behavioral prediction [
52,
53].
In the analysis of the contribution of each construct for the prediction of the FCU, intention in the TPB proved to be good predictors of the behavior. The correlation between condom use intention and condom use behavior was moderate (0.61). This is higher than values in previous studies: Albarracín et al. [
15] recorded 0.57 and Sheppard et al. [
20] 0.53; however, this was slightly lower than Van den Putte et al. at 0.62. Several studies indicate that the association between condom use intention and the use of this method of protection tends to be higher when past condom use is evaluated, rather than when a prospective measure is evaluated [
15]. This happens because the intention to engage in a behavior is closely related to past behaviors [
54]. Based on this premise, the intention-behavior correlations may be overestimated in this study since a retrospective measure of condom use was undertaken.
Compared to subjective norms and perceived control, attitude toward condom use was the best predictor of intention to use condoms. This result indicates that the FCU in Spanish adolescents mainly depends on their attitude toward this method of protection, rather than their peers’ FCU or self-efficacy to manage obstacles in using this method. Meta-analytic studies agree with our study on the importance of attitude compared to subjective norms in the prediction of condom use [
14,
19].
In this study, self-efficacy and perceived control were interchangeable constructs [
5,
52]. In the estimated model for the TPB, the correlation between perceived control/self-efficacy and condom use intention was non-statistically significant, suggesting that a higher perceived/self-efficacy control is not associated with condom use intention. In a critical review of 11 peer-reviewed studies, Protogerou et al. [
55] found a small correlation (0.04) between perceived control and condom use in the prediction of condom use in South African university students. Consistently, the meta-analysis conducted by Albarracín and colleagues [
15] concluded that perceived behavioral control does not contribute significantly to condom use, which is in contrast to the TPB. These unexpected results can be explained by our use of a brief self-efficacy scale focus on skills to manage obstacles to use condom. Another possible explanation is that self-efficacy and perceived control constructs are not equivalent [
10,
56]. This study demonstrates that the perception of being an effective person to manage condom use obstacles does not contribute to the prediction of condom use. It is important to note that although specific self-efficacy does not contribute to the prediction of condom use in this study, some evidence exists of its mediation role on the efficacy of sexual-risk reduction intervention for adolescents [
41,
57,
58], and prediction of condom use indirectly via intention [
37].
Cognitive factors, including knowledge and self-efficacy, contributed to the explanation of the FCU to a greater extent compared to behavioral factors in the SCM. While the importance of having the skills to use condoms and condom negotiation with one’s partner is recognized, it appears that self-efficacy to manage obstacles for condom use is a key predictor of condom use frequency. Notably, knowledge about STIs and condom use had the lowest standardized weight in the SCM (−0.01) and the IMB model (0.10), and were non-significant in the paths. This result does not empirically support the hypothesis that young people with more knowledge about STIs and condoms undertake greater condom use. However, this is not a definite conclusion—many factors could have contributed to this low explanatory contribution; for example, the instruments and scales we used could have been too broad in focus, meaning that they did not properly evaluate specific knowledge about condom use [
16]. Therefore, we recalculated the SCM and IMB model using only the subscale “Knowledge about condom use” as a measure of knowledge. However, the goodness of fit for both models worsened significantly, possibly due to the small number of items that comprise this factor. Testing the applicability of the IMB model, knowledge about HIV has been shown to be influential in the prediction of condom use among students from South Africa [
37]; however, most studies have suggested that knowledge is not a significant direct determinant of the behavior [
35,
59].
Competing interests
The authors declare that they have no competing interests.
Authors’ contributions
JPE obtained the funding and contributed to the study concept and design. AM collected data and wrote a draft. AGR conducted the data analysis, interpreted data, and provided statistical support. MO and RB collected data and provided guidance and supervision at every stage. All authors edited the article and approved the final manuscript.