Recurrent waves of infection are influenced by a combination of factors, including seasonal effects, the emergence of new pathogen strains, and fluctuations in human behaviour. Social networks play a key role in shaping individual perceptions of infection risk. However, capturing the complex influence of social media on individual attitudes toward public health measures—such as social distancing—remains a significant modelling challenge. In this study, we investigate how social network topology impacts opinion dynamics related to risk during a pandemic. Building on a previously calibrated and validated agent-based model of the COVID-19 pandemic, we incorporated a social network layer to simulate the spread of risk perception. This enabled us to compare the effects of different network structures on the emergence of recurrent waves. Our simulations indicate that networks exhibiting both scale-free and small-world properties most closely reproduce real-world infection dynamics.