Background: Cancer remains a leading cause of premature mortality worldwide, yet over 40% of cases are preventable through modification of established risk factors. Although national cancer control plans emphasize public awareness, the translation of knowledge into sustained behavioral change faces persistent challenges. Current primary prevention strategies often rely heavily on static educational dissemination paradigms. While digital health platforms provide scalable tools for data collection and personalized feedback, evidence regarding their effectiveness in driving sustained or adaptive behavioral change in real-world community settings remains limited.
Objectives: This prospective pilot study aimed to assess the longitudinal relationship between cancer prevention knowledge and health behaviors, quantify behavioral inertia, explore the structural connectivity between knowledge and behavioral domains, and inform equity-oriented strategies for primary cancer prevention in community settings.
Methods: We conducted a prospective longitudinal study (April–September 2025) in Chongqing, China, utilizing a hybrid intervention model combining a WeChat-based digital platform (for cancer risk assessment and tailored recommendations) and routine community health promotion. A multistage sampling design yielded 787 adults with valid baseline and follow-up data (mean interval: 138 days). Cancer prevention knowledge (based on the China Code Against Cancer) and behaviors (aligned with the 2018 WCRF/AICR recommendations plus smoking) were assessed using structured questionnaires. Cross-lagged panel models evaluated bidirectional knowledge-behavior associations and domain-specific heterogeneity, adjusting for sociodemographic covariates. Exploratory network analyses examined structural relationships, centrality, and bridge nodes, with subgroup analyses by socioeconomic status (SES).
Results: Overall baseline cancer prevention knowledge showed no positive longitudinal effect on subsequent health behaviors. Challenging the conventional "Knowledge-Attitude-Practice" paradigm, a small but statistically significant negative cross-lagged effect was observed (β = −0.06, P = 0.03). Behavioral inertia dominated across most domains, indicating resistance to short-term behavior change. Knowledge significantly promoted physical activity behavior (β = 0.07, P = 0.04), which showed relatively low behavioral inertia (autoregressive β = 0.32). In contrast, no significant effects were observed for smoking, body mass index, alcohol use, or diet, all exhibiting stronger inertia (β: 0.46–0.80). Dietary knowledge negatively predicted subsequent dietary behavior (β = −0.08, P = 0.01). Network analyses revealed stable structures but weak connectivity between knowledge and behavior nodes (bridge expected influence ≈ 0). SES-stratified analyses showed that among low-income participants, dietary knowledge and fast-food intake had bidirectional negative longitudinal associations (both P < 0.05), suggesting constrained food environments may undermine knowledge-driven behavior change and promote compensatory dietary choices.
Conclusions and Policy Implications: These findings demonstrate a fundamental decoupling between cancer prevention knowledge acquisition and behavior change, underscoring that awareness alone is insufficient to overcome behavioral inertia. The observed SES-specific negative loops challenge the efficacy of traditional "one-size-fits-all" static educational strategies. Future digital cancer prevention should evolve from passive information delivery toward context-sensitive support systems, with artificial intelligence (AI) serving as an enabling tool. Rather than simply intensifying education, such systems should employ AI for temporal interpretation of behavioral trajectories and domain-specific prioritization, engaging individuals as dynamic agents within their structural constraints. This shift aligns with IARC’s strategic priorities on reducing cancer inequalities.

Figure 1. Longitudinal and network-based analyses of cancer prevention knowledge and behavior