Use of Behavior Change Techniques in Digital HIV Prevention Programs for Adolescents and Young People: Systematic Review

Introduction

HIV Infections Among Adolescents and Young People

HIV continues to be a global public health concern, with around 1.3 million new HIV infections occurring in 2022 worldwide [1]. Until 2022, there were around 39 million people living with HIV worldwide. HIV imposes a tremendous health and economic burden on the world. Since the epidemic began, more than 40 million lives have been lost due to HIV-related illnesses [2]. Lifelong antiretroviral treatments require long-term health care services and medications [3]. A study by Schackman et al [4] estimated that the discounted cost treating one patient with HIV, infected at the age of 35 years, for a lifetime in the United States is approximately US $326,500. By preventing 1 HIV infection, US $229,800 could be saved [4].

Despite a decreasing trend [5], adolescents and young people still occupy a significant proportion of new HIV infection. In 2020, adolescents and young people aged 13 years to 24 years accounted for one-fifth of newly diagnosed cases in the United States [6]. The number of new HIV infections among adolescents and young people from 2010 to 2019 were 170,000 and 460,000, respectively worldwide [5]. Meanwhile, the number of HIV-positive cases of young men who have sex with men (MSM) between the age of 13 years and 24 years has significantly increased from 2015 to 2019, indicating an increasing trend in HIV infection among young MSM in the United States [7]. Significant efforts remain to be made to achieve the global target of ending the HIV epidemic by 2030 [8].

Behavioral Interventions for HIV Prevention

The disproportionate burden of HIV among adolescents and young people is demonstrated to be primarily driven by HIV risky behaviors, such as unprotected sexual intercourse [9]. It was reported that the percentage of condomless sex among MSM in the United States increased from 46% in 2012 to 70.5% in 2017 [10]. A study suggested that only half (54.3%) of high school students in the United States reported having used condoms during their last sexual intercourse [11]. Hence, improving their condom-use behaviors is important. Previous literature has suggested that HIV knowledge and condom-use self-efficacy were 2 essential cognitive factors influencing condom use [12-15]. Some studies also showed that condom-use self-efficacy was a significant mediator between HIV knowledge and adolescents’ condom use [16]. Nevertheless, the HIV knowledge and condom-use self-efficacy of youth remains suboptimal. It was found that less than half of adolescents possess comprehensive knowledge about HIV prevention [17]. Another study also documented that more than half of university students reported low levels of condom-use self-efficacy [18]. Therefore, to control and prevent HIV among adolescents and young people, interventions to promote their HIV-related knowledge, condom-use self-efficacy, and condom-use behaviors would be particularly important.

The Internet as a Useful Platform for Health Promotion

In recent decades, increasing use of the internet and smartphones to deliver health interventions has been observed in the literature [19,20]. Numerous advantages of web-based interventions have been identified, such as easy accessibility, availability, and assimilation to participants’ everyday lives. The internet has become the most common platform for young people to search for health-related information [21].

Information on sexual health and sexually transmitted diseases is commonly sought through the internet, especially by ethnic minority groups and minority groups with sexual orientation differences [21,22]. Young MSM are more likely to look for HIV-related information digital compared to the general population [23]. The growing HIV epidemic and the proliferation of the internet use warrants the application of innovative strategies and IT for behavior change interventions. The internet provides an alternative platform for delivering health interventions to the minority population. Accumulated evidence has shown that digital interventions are effective in promoting HIV knowledge and condom use, with varying effect size [24-28].

Behavior Change Techniques in Health Interventions

One factor contributing to the varying effectiveness of digital HIV prevention can be attributed to the difference in the use of behavior change techniques (BCTs). BCTs are the observable, replicable, and irreducible intervention components that are designed to modify the processes that regulate behavior. In 2008, Abraham and Michie [29] developed a taxonomy of BCTs and defined 26 techniques with respective theoretical support, providing a standardized method of classifying intervention content [29,30]. Theoretical frameworks most frequently identified in the use of BCTs include the information-motivation-behavioral skills model, theory of planned behavior, theory of reasoned action, social cognitive theory, control theory, and operant conditioning [29,31]. Furthermore, as suggested by the Transtheoretical Model, individuals experience stages of precontemplation, contemplation, preparation, action, and maintenance to achieve and sustain a behavioral change [32]. People situated at different stages of change have various needs in driving them onto the next stage, including recognizing the pros and cons of behavior and acquiring the skills to act as well as managing the action to sustain changes. The 26 BCTs categorized in the taxonomy can be classified into various hierarchical levels fulfilling the diverse needs a person requires to produce a behavioral change under the Transtheoretical Model.

Evidence suggests that specific BCTs would be effective in improving a range of health behaviors [33-39]. For example, a systematic review of 19 studies on Type 2 diabetes prevention interventions published in 2020 [37] showed that effective interventions used in general 3.7 to 5.6 times more BCTs than noneffective interventions, and BCTs of social support, goal setting, feedback on behavior, and self-monitoring of outcomes of behavior were identified in >90% of effective interventions. A review and meta-regression published in 2022 [39] identified BCTs used in smartphone app interventions, which found that BCTs of action planning and graded tasks had medium-positive associations with increasing physical activity in people with cardiovascular disease. Another review of interventions that promoted walking and cycling [36] found that prompting self-monitoring of behavior and prompting intention formation were the most frequently coded BCTs in the included studies. Other review studies on interventions in promoting health behavior change also support that those interventions that incorporated more BCTs tended to have larger effects compared to interventions that incorporated fewer techniques [35]. One review specifically targeting mobile health interventions [40] found that techniques of personalization, feedback and monitoring, and associations were most commonly used in mobile health interventions, and prompts and cues were the most common BCTs used in effective trials.

The BCT taxonomy provides a standardized and reliable way to extract and code the information from behavior change interventions and describe their components. It also offers a nomenclature system of intervention techniques that could be applied to health promotion programs aiming to drive behavioral changes [29]. The use of well-defined BCTs allows investigators to evaluate the effectiveness of an intervention more systematically and to conduct accurate replications of efficient interventions. However, past reviews on digital HIV prevention tended to focus on the interventions’ effectiveness, rather than their techniques in leading behavioral changes [25,26,41-43]. There are limited reviews on the BCTs used in digital HIV interventions targeting adolescents and young people [44]. As digital HIV interventions may vary in their content and effectiveness, identifying the most effective BCTs used in different HIV prevention studies is important. This can help to delineate potential mechanisms between components and outcomes and inform the development of more targeted and efficient interventions in the future.

This Study

This study aimed to identify the commonly used BCTs in effective digital HIV prevention programs targeting adolescent and young people using a systematic review.

Methods

Search for Studies

A literature search was conducted in 4 databases, including PubMed, Embase, Cochrane Library, and APA PsycINFO, to identify publications written in English and published from January 2008 (the beginning of the popularization of smartphones) to November 2024 [45]. Search keywords were developed in the following 5 domains: setting, intervention, outcome, population, and study design, using Boolean logic. The search terms in the 4 databases are presented in Multimedia Appendix 1. In addition, bibliographies of existing systematic reviews were also reviewed for a comprehensive data search. No protocol or registration of this review was prepared. This review was performed following the PRISMA (Preferred Reporting Items for Systematic Reviews And Meta-Analyses) statement [46]. The PRISMA checklist is provided in Multimedia Appendix 2.

Eligibility Criteria

The eligible criteria for this systematic review, developed based on the population, intervention, comparison, outcomes, and study design framework, are provided in Multimedia Appendix 3. Briefly, randomized controlled trials (RCTs) of HIV prevention in adolescents and young people, implementing digital health interventions aiming to improve key outcomes, such as HIV knowledge, condom-use self-efficacy, and condom use, were considered for inclusion.

Study Selection

Articles were reviewed after a comprehensive search and screening based on the inclusion and exclusion criteria. The screening and selection process was conducted by 2 independent reviewers (TCL and LX) to identify eligible studies. Any disagreements were resolved through discussion and were referred to by a third reviewer (PK-hM) if necessary.

Data Extraction and Quality Assessment

For the included studies, data extraction was performed by 2 independent reviewers to collect the following information: studies’ bibliographic information, participant demographics, intervention and control group, models or theoretical frameworks use, intervention intensity, duration, settings, completion rate, and results. The intervention strategies used in each study were reviewed and the BCTs used were identified based on the taxonomy of 26 BCTs developed by Abraham and Michie [29]. The BCTs were further classified into 3 hierarchical levels for individuals with different stages under the transtheoretical model (Figure 1 [29]). In particular, the level 1 BCTs have a provision nature and are used to provide information and resources to participants, mainly from a top-down approach. Examples include providing information, providing instruction, and providing general encouragement. Level 2 BCTs aim to initiate actions by prompting preparation and planning. Examples include prompt intention formation, barrier identification, specific goal setting, and practice. Level 3 BCTs are used in strategies to manage and hence maintain desired health behaviors for a sustainable change during the late action and maintenance stages. Examples are prompt self-monitoring of behaviors, relapse prevention, and time management.

‎

The methodological quality and risks of bias were assessed by the 2 reviewers, using the Joanna Briggs Institute critical appraisal tool for assessment of risk of bias for RCTs [47]. In total, 11 categories, including randomization, concealed allocation, blinding of participants and deliverers, and outcome measurements, were evaluated. An overall percentage score (0% to 100%) was calculated for each study representing the overall study quality. The 2 reviewers scored them independently, and discussion was made with a third investigator in case of any disagreements.

Data Analysis and Results Interpretation

Basic study characteristics of all RCTs included were summarized in table form. Characteristics included bibliographic information, settings, sample sizes, participant demographics, inclusion criteria, measurements used, and study duration. Results of RCTs were summarized and compared while significant improvements on predefined outcome indicators, as well as intervention completion rate were recorded.

Identified BCTs from intervention of selected RCTs were analyzed and compared based on study outcomes. Application of health models and theories in those studies were noted and matched with relative BCTs. Interventions that were developed or derived from previously published interventions were remarked for potential further investigations. The overall frequencies of BCTs used among all studies collected for this systematic review were counted and summarized. In addition, the frequencies of BCTs used in every significant outcome were counted to explore the commonly used BCTs in effective digital HIV prevention programs targeting adolescent and young audience.

Results

Included Studies

On the basis of the literature search, 463 records were identified from 4 databases; an additional 39 records were identified through searching for the bibliographies of previous systematic reviews [43,48-50] on digital HIV interventions. Of these records, 119 (25.7%) duplicated records were removed, and the remaining 383 records were screened for eligibility. Following this, an additional 291 (63%) studies were removed after screening by title and abstract, as they did not meet the inclusion criteria. A further full-text review was conducted on the remaining 92 records, and 51 (11%) of them were excluded for the following reasons: non RCT design (n=12, 24%), irrelevant outcomes (n=16, 31%), not digital interventions (eg, in-person interventions, text messaging interventions, or telephone interventions; n=25, 49%), and out of age range (n=5, 10%). Finally, 34 studies were included in this systematic review. The selection process is shown in the PRISMA flowchart in Figure 2 [51].

‎

Study Characteristics

Table 1 provides an overview of the characteristics of the included studies. Among the 34 studies, 24 (71%) were conducted in the United States, 4 (12%) in African countries (Nigeria [52] and Uganda [53-55]), 5 (15%) in Asian countries (Hong Kong [56], China [57,58], Thailand [59], and Cambodia [60]), and 1 (3%) in Europe (Germany) [61]. Of the 34 studies, 14 (41%) included participants around early or middle adolescence (aged <18 years), and the other 20 (59%) included late adolescents and young people (average age: ≥18 years). In addition, 20 studies (20/34, 59%) targeted at those high-risk populations for HIV infection, involving subgroups such as MSM (n=12, 35%) [56,59,62-71], sexual minority men (N=1) [57], transgender women (n=1, 3%) [59], Black people (n=3, 9%) [67,68,70], low-income [72,73] or homeless individuals [74], refugees (n=1, 3%) [55], female entertainment workers (n=1, 3%) [60] and people with substance use [75,76]. The sample size in the 34 included studies ranged from 40 [62] to 1578 [77].

Regarding recruitment methods of the included RCTs, 8 (23%) recruited participants from educational institutions (n=7, 12% from secondary schools [52-54,72,73,80,83,85] and n=1, 3% from a university [61]), 7 (12%) recruited through a combination of digital advertisements and community-based organizations [56,57,59,63,67,68,77,82], 7 (12%) recruited from community centers or treatment clinics [65,66,74-76,79,81,84], 1 (3%) recruited from factories [58], 1 (3%) recruited from refugee settlements [64], 1 (3%) recruited from entertainment venues [60], and 5 (15%) recruited entirely digital via websites, social media, or dating apps [62,64,69-71,78]. Most of the studies focused on participants who were HIV-negative (Table 1).

Intervention Characteristics

An overview of the characteristics of interventions included is listed in Tables 2 and 3. Most of the studies (29/34, 85%) were 2-arm intervention design with 1 experimental group and 1 control group. In total, 4 studies [53-56] had 2 experimental groups together with 1 control group. Only 1 study [84] had 3 experiential groups together with 1 control group. In addition, 6 studies [56,58,62,72,83,85] were one-off interventions; the remaining 28 studies had multiple sessions, and their intervention duration ranged from 2 weeks [76] to 24 months [84].

For the modality type of the 34 digital interventions, 20 (59%) were web-based interventions (eg, conducted through interactive websites [53,54,56,57,59,61,64-68,72,78,81,83,85] or multimedia software [75,76,82]); 7 (21%) were mobile-based interventions (eg, through mobile apps [62,63,70,74,79] or SMS text messaging [55,60]); 5 (15%) were video-related (eg, video games [69,73,80], educational videos [58], and videoconference [71]); 2 (6%) interventions were conducted through social media [52,77], and 1 (3%) through a combination of videoconference, SMS text messaging, and social media [84].

In total, 9 (27%) of the 34 interventions were delivered at in-person sessions under the supervision of research staff [53,54,62,73,75,76,79,80,82], while the remaining interventions (n=25, 73%) were operated and delivered remotely. All 3 interventions using multimedia software required on-site delivery with a relatively small sample size (n=56 to 178) [75,76,82], hinting the limitation of such modality type in large-scale community implementation.

Theories and BCTs

Table 4 provides an overview of theories and BCTs used in all studies. Of the 34 studies, 23 (68%) interventions were theory-based. Cognitive-based theories, including Integrated Behavioral Model [67,68], Information- Motivation- Behavior Skills Model [65,66], Social Cognitive Theory [69,73,82,84], behavioral theory on Dyadic Health model [71], as well as emotion-based approaches, such as Rational Emotive Behavior Therapy theory [55] and Psychological Wise theory [83], were used in these studies. In addition, over half of studies (21/34, 62%) included the BCTs across the 3 hierarchical levels under the Transtheoretical Model (Figure 1), and the other 13 studies included BCTs solely at level 1 and level 2.

The frequency of each BCT used in all studies is shown in Table 5. The most commonly used BCTs were prompt intention formation (34/34, 100%), provide information about behavior-health link (33/34, 97%), provide information on consequences (31/34, 91%), provide instruction (30/34, 88%), provide general encouragement (29/34, 85%), prompt barrier identification (23/34, 68%), provide information about others’ approval (20/34, 59%), and provide opportunities for social comparison (19/34, 56%). Unused BCTs include provide contingent rewards, teach to use prompts or cues, and time management. The proportion of BCTs used under each level was 67%, 40%, and 17% for level 1, level 2, and level 3, respectively. All interventions included BCTs from level 1 and level 2. However, BCTs from level 3 were more frequently used in studies targeting older adolescents (aged ≥18 years) [55,57,62,65-71,74,84] compared to younger adolescents (aged <18 years) [53,54,63,72,73,83,85]. Specifically, level 3 BCTs in studies targeting younger adolescents included follow-up prompts [53-55,57,63,71,84], prompt self-monitoring of behavior [57,72,73,83,85], motivational interviewing [63,71,72,83-85], and prompt review of behavioral goals [57,63,73,84], while those targeting older adolescents included prompt review of behavioral goals [62,65,66,74], prompt self-monitoring of behavior [62,67,68,70], relapse prevention [65,66,71,74], stress management [57,69,71], follow-up prompts [65-67], and motivational interviewing [62,80].

Results and BCTs per Improved Outcome

Three key outcome measurements were identified among the included 34 studies, including (1) HIV knowledge (n=16, 47%), (2) condom-use self-efficacy (n=15, 44%), and (3) frequency of condom use (n=26, 76%). Over half (20/34, 59%) of the interventions had an intervention completion rate higher than or equal to 85%. It is estimated that the high completion rate might be a result of school-based and clinic-based recruitment. Follow-up measurements were commonly carried out immediately after intervention, at 3 and 6-month follow-ups. Variations existed depending on the nature and duration of studies (Tables 2 and 3).

In total, 11 (69%) of the 16 studies that included HIV knowledge as an outcome reported a substantial improvement in HIV knowledge. In addition, 7 (47%) of 15 studies that included condom-use self-efficacy as an outcome reported a statistically significant increase in condom-use self-efficacy. Moreover, 14 (54%) of 26 studies that included condom use as an outcome reported a statistically significant increase in condom use frequency. Table 5 summarizes the frequencies of BCTs linked to each outcome indicators in studies with significant improvements. Prompt intention formation (100%), provide information about behavioral-health link (97%), provide information on consequences (97%), provide instruction (97%), and provide general encouragement (91%) were the most commonly used BCTs among all the 3 improved target outcomes. Prompt practice (5/7, 71%) and prompt self-monitoring of behavior (4/7, 57%) were more frequently used in studies to drive improvement in condom-use self-efficacy compared to other outcome indicators, while a more diverse and even use of the 26 BCTs across the 3 levels was observed in studies with an increase in condom use frequency.

Quality Assessment

The quality assessment of included studies based on the Joanna Briggs Institute tool is provided in Table 6. All 34 studies included were randomized. Among these, 10 (29%) studies reported the use of concealed allocation [52,57-59,63,65,66,71,81,84], and 31 (91%) studies demonstrated baseline similarity between intervention and control groups [52-54,56-77,79-83,85]. Blinding was less frequently reported, with 5 (15%) studies documenting participant blinding [52,63,65,66,84], 5 (15%) studies documenting outcome assessor blinding [52,55,56,60,82], and 7 (21%) studies documenting blinding of deliverers [60,63,65,66,72,81,84]. Most studies (31/34, 91%) ensured that the groups were identical except for the intervention [52-54,56,57,59-70,72-85]. All studies reported identical outcome measurements, and 32 (94%) studies used reliable outcome measurements, with 2 (6%) studies lacking this information [55,60]. Follow-up analysis was conducted in 23 (68%) studies [52-54,56,59,60,62,65,67,69,70,72-81,84,86], and 31 (91%) studies performed analysis according to correct groupings [52-56,58-71,73-82,84,85]. The quality score of the included studies ranged from 45% [55,83] to 91% [52,65], with 24 (71%) studies receiving the score <70%, and 10 (29%) studies receiving a quality score >70%.

Discussion

Use of BCTs and Theories

From our understanding, this study is the first review on the use of BCTs in digital HIV interventions among adolescents and young people. The findings showed that among the 3 levels of BCTs, BCTs at level 1 are the most frequently used with an average frequency of 67%. BCTs at level 1 help to deepen participants’ understanding of HIV and the role of condoms in preventing sexually transmitted diseases. By enhancing and consolidating important knowledge, higher motivation will be generated to encourage change. This is especially useful for people at the precontemplation and contemplation stages. In contrast, level 2 BCTs mark the starting point of turning internal ideas into actual behavioral change, or in other words mobilizing participants to enter the preparation and action stage from precontemplation stage, according to the transtheoretical model. Level 2 BCTs had an average frequency of 40% in this study, which is less commonly used among the 34 studies compared to level 1 BCTs. Among all, level 3 BCTs were at the highest hierarchical level as they are the most complicated to implement though potentially most impactful in causing long-term change. Level 3 BCTs were the least used among all 3 levels with an average frequency of only 17%. The use of BCTs from 3 hierarchical levels in this systematic review matches with the rule of thumb for at-risk populations under the transtheoretical model. Specifically, level 1 BCTs were commonly used for the largest group of recipients from precontemplation (40%) and contemplation (40%) stage, while level 2 BCTs were less frequently used for populations at the preparation stage (20%) [32]. It is also important to note that the majority of the included studies (23/34, 68%) were observed to be supported by or constructed based on behavior change models or theories. No substantial difference was observed in the number of BCTs used under each theory in this review. In other words, the impact of theories on the pattern of BCTs used seems to be minimal.

BCTs Linked to Improved Outcomes

Findings of the study found that level 1 BCTs with a provision nature were the most commonly used in bringing significant improvement in HIV knowledge, indicating that providing information and instructions can be an effective method to increase knowledge of participants. Some level 2 BCTs were used for consolidating information, such as prompt practice. Level 3 BCTs were rarely used regarding this outcome, indicating that higher level BCTs concerning larger degrees of self-initiated participation were not considered necessary in enhancing HIV knowledge. In contrast, a higher percentage use of level 2 BCTs was observed in studies with significant increase in condom-use self-efficacy. As suggested by the self-efficacy theory [89,90], self-efficacy can be determined by past experience (performance accomplishment), modeling by others (vicarious experience), coaching and evaluative feedback (social persuasion), as well as physiological and emotional states. Level 2 BCTs, such as prompt practice, plan social support, and motivational interviewing, were more frequently used together with level 1 BCTs, including providing feedback on performance and model the behavior, to provide and initiate experience, and modeling and coaching feedback to participants to effectively enhance one’s condom-use self-efficacy.

Finally, studies with a significant increase in condom use frequencies used level 3 BCTs the most among all 3 outcome indicators, suggesting the importance of managing an action for a sustained behavioral change. Involved level 3 BCTs include use follow-up prompts, relapse prevention, and stress management, which are rarely or have never been used in achieving the other 2 outcome indicators. In addition, provide information about others’ approval, provide feedback on performance (level 1), and plan social support (level 2) were used more frequently in increasing condom use, showing the importance of perceived social norms on driving actual behavior change. Findings of this paper match with those from relevant systematic reviews of digital health interventions of other health conditions, which show that provide encouragement, provide information about behavior-health links, provide instructions, and provide information on consequences were the most frequently used BCTs [91,92].

Advantages of Digital Platforms for Delivering BCTs in HIV Prevention

Digital platforms offer unique features that allow the application of certain BCTs in HIV prevention interventions. Web-based interventions can reach a wider audience, including hard-to-reach populations (eg, MSM), and offer cost-effective solutions compared to traditional face-to-face interventions [93]. In addition, the interactive and engaging nature of digital platforms helps to deliver BCTs more effectively, making them a powerful tool in HIV prevention efforts. For instance, customizable avatars allow participants to choose characters that share their race and cultural background, which makes it easier for participants to empathize and imagine the cases in reality. This customizable function facilitates the use of BCTs, such as prompt barrier identification and providing opportunities for social comparisons. Anonymous interactions on forums and video-uploading platforms enable participants to express genuine concerns and thoughts freely. These interactions incorporate BCTs like providing information about others’ approval and facilitating planning of social change [94]. Such anonymity ensures that the youth feel safe and supported, which is particularly important for sensitive topics like HIV prevention. Tailor-made content based on risk assessment results and preset goals ensures that messages are accurate and relatable to targeted populations, which can enhance the relevance and impact of the interventions. The process of information through the central route is deepened with the use of follow-up prompts and providing feedback on performance [95], ensuring that participants fully understand and engage with the information provided. Finally, unlocking graded modules, for example, quizzes or chapters with elevating difficulty, helps participants gradually build a solid foundation of knowledge and behavioral change. BCTs like setting graded tasks and providing performance feedback are effectively integrated into digital interventions, ensuring continuous engagement and learning in HIV prevention.

Public Health Contribution and Implications

The youth demographic is particularly vulnerable due to the high level of engagement in risky sex and barriers to accessing traditional health care services. The focus on digital interventions is significant in this underserved and hard-to-reach population. The internet offers a wider yet more complex platform for health education. Its distinctive attributes, such as rapidly evolving trends, a multitude of information sources, and diverse user demographics, can make it challenging to effectively use digital health interventions. To navigate this evolving landscape of health education, it is beneficial to examine successful strategies that incorporate commonly used BCTs in digital interventions.

The findings of this review offer valuable guidance and insights for the design and implementation of future HIV prevention programs aimed at adolescents and young people. The BCTs identified in this review serve as a valuable reference for further studies, enabling researchers to select appropriate BCTs for intervention development and improvement. The most used BCTs identified in this review included prompting intention formation, providing information about behavior-health links, providing information on consequences, providing instructions, and providing general encouragement, which is important in fostering awareness and motivation for behavior change. Interventions showing improvements in HIV knowledge mostly included BCTs with a provision focus, such as providing information on consequences, behavior-health links, and encouragement. This suggests that educational components are essential for building foundational knowledge and promoting initial engagement and readiness for behavior change. In this review, BCTs aimed at initiating actions (eg, prompts for intention formation and barrier identification) were commonly observed in studies reporting a significant increase in condom-use self-efficacy, which provides evidence for bridging the gap between awareness and action. While to improve actual condom use, BCTs focused on managing and sustaining behavior (eg, use of follow-up prompts and relapse prevention) may need to be incorporated into the interventions. Considering the diverse range of outcomes to be achieved, hierarchical levels of BCTs should also be taken into account to design tailored and effective interventions. It is recommended that public health practitioners use behavior change theories as a foundation, using BCTs at different levels to select intervention activities and deliver pertinent messages for maximizing program effectiveness.

In addition, future studies can delve into identifying intrinsic and extrinsic motivational techniques within the current BCT taxonomy. Exploring the effectiveness of emotive techniques in driving behavior change in digital interventions could be another promising avenue for investigation. Moreover, the digital nature of the included interventions may allow for the scalability and adaptability of these findings to other topics beyond HIV prevention, particularly those that share similar characteristics. By leveraging the knowledge gained from this review, we can enhance the effectiveness of HIV prevention programs for adolescents and young people and facilitate border implementation.

Limitations

There were some limitations of the study. First, some BCTs may not be easily observable and thus might have limited the identification of the exact BCTs used. Four BCTs, namely provide contingent rewards, teach to use prompts or cues, agree behavioral contract, and time management were never used in the included studies. Among these, teach to use prompts or cues and time management can be unobservable techniques that are hidden in message contents, rather than being described as intervention activities. The limited description of intervention messages can conceal certain BCTs, leading to bias in analysis and estimation of BCT effects. Also, definitions of some BCTs, such as provide general encouragement and prompt intention formation, may be too general and ambiguous for having a meaningful analysis. Some BCTs, such as prompt intention formation (34/34, 100%), provide information about behavioral-health link (33/34, 97), provide information on consequences (33/34, 97%), and provide instruction (33/34, 97%), were used in almost every study. Their high frequency can be due to their broad definitions, making them almost necessary for all behavioral change interventions.

Second, emotive techniques are proven to be highly effective in affecting one’s judgment and decision-making [96] but are not included in the 26-item taxonomy of BCTs. Fear appeal is an emotive technique used in the intervention of one of the included studies but failed to be categorized into any of the BCTs. The omission of important emotion-driven behavior change approaches in the BCTs may have limited the applicability of the taxonomy in other settings.

Third, counting the frequencies of BCTs used may not fully represent the associations between BCTs and significant improvements. Effect size of individual BCT is difficult to ascertain as multiple BCTs are used in one study. Due to the lack of effect size calculation, it is hard to determine whether the high frequency of use of a BCT is due to its effectiveness or its simplicity in application. It also limited meta-analytic evaluations. Hence, it is difficult to fully attribute the significant improvements in results to the use of a specific BCT.

Finally, this review included study samples aged 10 to 30 years, consisting of early adolescents, middle adolescents, and late adolescents or young people. The age difference of targeted populations may influence the effectiveness of the BCTs provided, as the risk levels of HIV infection and challenges in enacting certain BCTs may differ among early adolescents and young people. Future studies can consider age-specific tailoring to address the potential differences.

Conclusions

This systematic review has identified the most commonly used BCTs used in digital HIV prevention programs targeting adolescents and young people. The findings can serve as important references for future interventions to provide more effective approaches for HIV prevention on the internet.