We conducted a three-round Delphi study to develop a theory-based educational model for breast cancer prevention in adolescents based on expert consensus.¹⁸ The primary objective of the Delphi process was content validation (i.e., assessing the relevance and appropriateness of proposed model components), rather than item prioritization, ranking, or weighting. The Delphi approach was chosen because it enables the structured synthesis of expert judgment when empirical evidence is limited or fragmented, making it suitable for formulating intervention model components in adolescent health.^18–20 Reporting followed the Conducting and Reporting of Delphi Studies (CREDES) guidelines (https://www.equator-network.org/reporting-guidelines/), which emphasize clarity of objectives, panel selection, iterative procedures, a priori consensus definition, controlled feedback mechanisms, stability indicators, and termination rules. ²¹ A completed CREDES checklist is provided as Supplementary File S1.

The study targeted school-based and primary health care educational components for adolescent girls aged 15–19 years in the Regency of Majene, Indonesia. The Delphi panel included oncology professionals (physicians and nurses), health promotion academics, health service practitioners, representatives of breast cancer survivor communities, and representatives of relevant community and religious organizations involved in adolescent breast cancer prevention.

Experts were recruited purposively through professional networks and reference institutions to ensure disciplinary heterogeneity and stakeholder role diversity. Inclusion criteria were: (1) relevant professional experience of at least 2 years; (2) ability to complete all Delphi rounds; and (3) absence of conflicts of interest related to the study objectives. There is no fixed numerical standard for Delphi panel size; adequacy is primarily determined by the quality of expertise, relevance, diversity, and retention across rounds. In health-related Delphi studies, heterogeneous panels commonly include 7–12 experts to balance diversity and response burden, while anticipating attrition between the rounds. Attrition rates of approximately 10–30% are considered acceptable in health Delphi research. Based on these considerations, 15 experts were recruited in Round 1, resulting in 11 experts completing Rounds 2 and 3, thereby maintaining panel adequacy. We acknowledge that small panels can inflate I-CVI estimates; therefore, modified kappa (k) was used to adjust for chance agreement and to support cautious interpretation of item-level content validity.*

Round 1 used a semi-structured questionnaire (online and offline) to generate candidate items related to educational content, SCT-based behavior change constructs, and delivery strategies. Two researchers independently reviewed the responses and synthesized the initial item pool, grouping items into three domains: (1) self-care, (2) SCT enablers, and (3) delivery strategies. Discrepancies were resolved through discussion, and items were editorially refined for clarity and contextual relevance to adolescents.

In Rounds 2 and 3, panelists evaluated the relevance of each item using a 4-point Likert scale (1 = not relevant to 4 = very relevant), consistent with recommended practice for content validity index (CVI) assessment in health Delphi studies.²² After Round 2, controlled, aggregated feedback (median, IQR, percentage agreement, I-CVI, and anonymized comments) was provided to support re-evaluation in Round 3, while maintaining panelist anonymity.^21,23

The primary consensus criterion, defined a priori, was I-CVI ≥0.80 in Round 3 (i.e., ≥80% of experts rated the item as 3 or 4), a commonly accepted threshold for good content validity.²² In addition to I-CVI, S-CVI/Ave and modified kappa (k*) were calculated to adjust for chance agreement.²² Items meeting the I-CVI threshold but not achieving the strict kappa criterion (k ≥0.74) were retained in the model with interpretive caution and were not classified as “final-strict” consensus items. Inter-round stability was assessed using changes in median and IQR from Round 2 to Round 3, while Kendall’s W was examined as an additional indicator of inter-rater agreement.^24,25 Kendall’s W was used descriptively to support assessment of rating convergence rather than as a formal criterion for item retention. The Delphi process was terminated after Round 3 when item ratings demonstrated stability and minimal likelihood of further change.²⁵

Data analysis was conducted using R (version 4.x) and Python (version 3.11). For each item and round, the number of raters (N), percentage agreement (score ≥3), I-CVI, Wilson 95% confidence intervals, modified kappa (k*), median, and IQR were calculated. Given the small number of raters, the Wilson 95% confidence intervals were relatively wide, which should be considered when interpreting item-level validity estimates. Missing data were not imputed; denominators reflected the number of experts providing ratings in each round. The extent of missing ratings across rounds is summarized in the Results section to improve transparency. The analysis code and de-identified dataset are available upon reasonable request.

After Round 3 consensus was established, and all final-strict items were synthesized using thematic analysis to facilitate interpretation and clarify the overall structure of the educational model. This process generated main themes and subthemes reflecting the core components of adolescent breast cancer prevention education, including self-care behaviors, breast cancer literacy, educational delivery strategies, social support, and stakeholder roles. A summary of the thematic synthesis is presented in Table 2. For transparency, items meeting I-CVI ≥0.80 but not k ≥0.74 were retained as consensus components but were not classified as “final-strict” in the thematic synthesis.

Table 2 summarizes the final-strict consensus results from Round 3 and presents the thematic structure of the adolescent breast cancer prevention educational model. Experts agreed that the model should prioritize preventive self-care behaviors, including breast awareness, and familiarity with normal breast changes, regular breast self-examination–related skills framed as awareness rather than a formal screening tool (not as a mortality-reducing screening tool), healthy lifestyle–based risk reduction (physical activity, balanced diet, weight control, and avoidance of smoking and alcohol), and self-regulation to sustain healthy behaviors. The second theme emphasizes the need for adequate breast cancer literacy among adolescents, covering breast anatomy, disease burden, risk factors, warning signs and symptoms, and screening/early detection concepts as future-oriented knowledge rather than age-indicated clinical practice and consequences of delayed detection. From an implementation perspective, experts emphasized that educational strategies should be interactive, skill-based, and youth-centered, with emphasis on direct, skills-based breast awareness/self-check demonstrations, peer learning, and school-based delivery supported by digital media. Experts also identified cultural and psychological barriers, such as stigma, shame, and taboos surrounding breast health discussions, that must be addressed within interventions. Consistent with Social Cognitive Theory, the model underscores multi-stakeholder support, involving teachers, school health units, primary health care providers, parents, peers, and breast cancer survivors, to strengthen self-efficacy, behavioral modeling, and program sustainability.

Based on the thematic synthesis of final strict Round 3 items, the expert panel reached consensus on an integrated educational model framework grounded in Self-Care Theory and Social Cognitive Theory (SCT), as illustrated in Figure 2. The model comprises four interrelated pillars: (1) preventive self-care behaviors, including breast awareness, breast self-check as a breast awareness skill (rather than age-indicated screening), risk reduction through healthy lifestyles, and self-regulation to sustain healthy behaviors; (2) breast cancer literacy, encompassing breast anatomy, disease burden, risk factors, warning signs and symptoms, screening options, and consequences of delayed detection; (3) youth-centered delivery strategies, featuring interactive, skill-based learning through breast awareness/self-check demonstrations and practice peer and role-model approaches, and utilization of school and digital channels, complemented by reinforcement and reminder sessions; and (4) SCT-based enablers and multi-stakeholder support, emphasizing self-efficacy, behavioral modeling, reinforcement, and social support from teachers and school health units, primary health care providers, parents, peers, and breast cancer survivors. The model explicitly accounts for cultural and psychological barriers, including stigma, shame, taboos, and low perceived vulnerability, to guide context-sensitive adaptation of educational strategies for adolescents.

Second, the breast cancer literacy domain endorsed by the panel was comprehensive, encompassing anatomy and physiology, disease burden, risk factors, warning signs and symptoms, and screening/early detection concepts framed as future-oriented knowledge (given that adolescents are not candidates for routine breast cancer screening), and the consequences of delayed detection. This is consistent with evidence indicating that low breast cancer and breast awareness/self-check literacy among adolescents constitutes a major barrier to preventive behavior adoption. Recent systematic reviews of breast cancer knowledge among secondary school students show that many adolescents hold misconceptions regarding risk factors, warning signs, and self-check/BSE procedures, underscoring the need for education that integrates clinical and practical knowledge.⁷ Compared with traditional curricula that tend to emphasize disease facts alone, this model balances “what breast cancer is” with “how it can be prevented and detected early”, while avoiding overmedicalization by clarifying that adolescents are not candidates for routine breast cancer screening.

Third, the expert panel emphasized the importance of interactive, skill-based learning strategies, including direct breast awareness/BSE demonstrations, structured practice, feedback mechanisms, peer modeling, and digital support. These recommendations are consistent with recent intervention trials demonstrating that hands-on practice and simulation or hybrid approaches are more effective in improving BSE-related skills and confidence than traditional lecture-based methods.²⁹ Digital and virtual approaches have also been shown to enhance breast awareness /BSE skills and self-efficacy by providing private learning spaces and repeatable instructional materials, which are particularly suitable for adolescents. ³⁰ An important distinction of this model is that these delivery strategies are positioned as structural components of the theoretical model, rather than merely technical methods, thereby enabling clearer theoretical tracking of behavior change mechanisms.

Fourth, the explicit inclusion of cultural and psychological barriers such as taboo, shame, and stigma related to discussing breast health represents a key strength of this model compared with many international interventions. Qualitative studies and sociocultural reviews from South Asia and other low- and middle-income countries indicate that stigma and norms of modesty often delay early detection learning, limit adolescents’ willingness to ask questions, and reduce self-efficacy for performing breast awareness/BSE. By explicitly addressing these barriers, the model enables the design of culturally adaptive educational strategies, such as the use of female facilitators, culturally sensitive language, and framing breast health as self-care rather than a sexual issue. This is particularly important because much of the previous literature implicitly assumes cultural neutrality, despite strong evidence that sociocultural context shapes intervention acceptability and effectiveness.

Fifth, consensus regarding SCT-based enablers and multi-stakeholder social support indicates that the panel did not view preventive behavior as a purely individual process. SCT emphasizes reciprocal determinism, whereby behavior emerges from interactions among personal factors (knowledge, outcome expectations, self-efficacy), behavioral practices (breast awareness/BSE and lifestyle behaviors), and the environment (parental support, schools, health workers, and peers).³¹ Theory-based intervention studies in breast cancer prevention consistently show that self-efficacy enhancement and social reinforcement are key mediators for sustaining preventive health behaviors and appropriate help-seeking.³² This model extends existing evidence by explicitly structuring the roles of teachers and school health units, primary health care providers, parents, peers, and breast cancer survivors as role models, with messaging adapted to adolescents to avoid fear-based or overly clinical narratives while maintaining credibility, an approach that has been rarely formalized in adolescent-focused studies. When involving survivors, narratives should be developmentally appropriate and empowerment-focused, avoiding fear-based or overly clinical messaging to minimize anxiety and overmedicalization among adolescents.

From a methodological perspective, the Delphi technique is widely used in the development of health education curricula and cancer-related indicators to ensure content relevance and cross-professional applicability, particularly when empirical intervention evidence remains heterogeneous.³³ Supported by a multidisciplinary expert panel, this model demonstrates strong content validity and enhanced potential for program adoption. However, consistent with the limitations of Delphi-based research, expert consensus does not guarantee real-world effectiveness. Accordingly, future research should focus on implementation and effectiveness testing in school and primary health care settings, assessing impacts on breast cancer literacy, self-efficacy, breast awareness/BSE skills and appropriate help-seeking, and healthy lifestyles using quasi-experimental or cluster-randomized designs. Given that lifestyle risk factors differ in the strength of evidence for long-term breast cancer risk modification, future evaluations should consider which behaviors are most plausibly mediating life-course risk reduction when initiated in adolescence. Existing literature further indicates that the success of adolescent cancer education is strongly influenced by program duration, repetition, and environmental support, underscoring the importance of medium- to long-term evaluations to assess behavioral sustainability.²⁶