Assessment of nutrition information online and the #nutrition social network

Introduction
The global rise in non-communicable disease underscores the critical role of nutrition and poor dietary choices in public health, which is exacerbated by misinformation, and an unregulated digital environment. This thesis developed a validated tool to assess the quality of online nutrition information, examined the actors in the online debate, and offers strategies to improve information quality and reduce the propagation of nutrition misinformation.
Methods
Quality assessment criteria were developed and validated through six stages: literature review, framework development, pilot testing, validation against existing tools, reliability testing, and application to nutrition-related URLs shared via X (formerly Twitter). Reposted and non-reposted data were compared for quality differences. Finally, to fully understand the discourse and structure of the online nutrition debate, NodeXL Pro was used for network, semantic, and sentiment analyses of the #nutrition network on X in March 2023.
Results
The final validated Online Quality Assessment Tool included 10 questions, and demonstrated high interrater reliability (k = 0.653, p < 0.001) and moderate internal consistency (α = 0.382). Pilot testing found 3% of articles were poor, 49% satisfactory, and 48% high-quality, with significant differences in quality scores between blogs, news articles, and press releases, χ2(2) = 23.22, p < 0.001. Poor-quality information was more prevalent in personal and company blogs, and reposted articles (n=267, sum of rank, 461.6) showing a significant difference in quality, U = 87475, p = 0.006, compared to non-reposted articles (n=738, sum of rank, 518.0). Social network analysis revealed a dispersed #nutrition network, where a few influential users, predominantly non-experts, shaped the debate. Semantic analysis highlighted themes of health, lifestyle, and diet.
Conclusion
This thesis introduced a novel tool for assessing online nutrition information quality with high validity and reliability. Its use revealed that low-quality articles are more frequently shared. Social network and sentiment analyses illustrated a dispersed, publicly driven #nutrition debate, dominated by health and lifestyle topics primarily propagated via reposting.