Public discourse on knowledge, attitudes, and practices toward dengue and chikungunya in China: a multi-platform social media infoveillance study, 2022–2026

Dengue and chikungunya are expanding mosquito-borne arboviral diseases. In 2025, Guangdong (Foshan), China, experienced the largest domestic chikungunya outbreak on record. Real-world Chinese public discourse remains poorly characterised, and domestic KAP survey evidence is sparse and localised. We used social-media infoveillance to characterise Chinese public discourse on dengue and chikungunya across five platforms (2022–2026). We conducted an observational, cross-sectional infoveillance study. Public Chinese-language posts, comments, and danmaku mentioning dengue, chikungunya, or their vector were collected from Bilibili, Weibo, and Zhihu (API-based) and, because of platform anti-automation controls, Douyin and Xiaohongshu (manually extracted; n = 97, reported descriptively). After deduplication and relevance filtering, 4,707 records (4,610 API-collected) formed the analytic sample. We applied rule-based disease classification, a lexicon-based multi-label KAP-aligned thematic classifier (six themes), SnowNLP sentiment, a regex-screen plus adjudicated misinformation workflow, and a stratified qualitative subsample of long-form posts (n = 53). Classifier reliability was quantified against blinded human coding (N = 250). Multi-platform incremental value was assessed by active-month coverage. Chikungunya-focused content (38.5%) exceeded dengue-focused (25.9%); discourse spiked in July–August 2025 (n = 1,166 and 987), coinciding with the Guangdong outbreak. Transmission-route (33.8%) and preventive-practice (17.8%) themes dominated, whereas vaccine/treatment knowledge was thinnest (7.5%). Sentiment (n = 3,571) was near-balanced negative 40.9% / positive 40.6% / neutral 18.4%; disease focus was a significant but weak predictor (χ2 = 24.2, p < 0.001, Cramér's V = 0.058). The automated KAP classifier showed weak-to-moderate reliability (macro κ = 0.24; preventive-practice κ = 0.09, CI including zero; disease κ = 0.93). Misinformation was rare (3.47 per 1,000 records); the person-to-person myth yielded zero assertions, conspiracy-origin had the highest yield (27.8%), and corrective content rose during the outbreak while misinformation did not. Daily social-media volume peaked on 28 July 2025, ~ 7–9 days after the official case peak (19 July 2025), providing no advance warning. Multi-platform coverage: the three API platforms' active-month union was 51 months, but Weibo alone would have missed 31. Chinese public discourse on dengue and chikungunya is strongly outbreak-driven, prevention-oriented, and shows an affective valence near-balanced between negative vigilance and positive reassurance. The 2025 Guangdong outbreak exposed persistent discourse gaps, especially around vaccine/treatment reality and disease differentiation, that warrant proactive, video-first, multi-platform health communication. Social-media infoveillance is a useful complementary outbreak-intelligence component, but in this outbreak it tracked rather than preceded the clinical peak; multi-platform design materially outperformed any single platform. What is already known about this subject • Dengue is a long-established mosquito-borne disease in southern China, but chikungunya had not caused a substantial domestic outbreak on the mainland until the 2025 Guangdong/Foshan event, the largest chikungunya outbreak ever recorded in mainland China. • Surveys show that public knowledge of arboviral diseases is uneven, yet real-world public discourse on Chinese social media has not been systematically characterised for either dengue or chikungunya. • Social-media infoveillance is an established early-warning and health-communication tool, but most Chinese-platform studies focus on a single platform or a single disease. What this study adds • A five-platform (Bilibili, Weibo, Zhihu, Douyin, Xiaohongshu), 4,707-record infoveillance dataset spanning 2022–2026 shows that public attention shifted sharply toward chikungunya during the 2025 outbreak (38.5% of disease-focused posts vs 25.9% for dengue), with a clear monthly spike in July–August 2025 that mirrors the official Guangdong epidemic curve. • A KAP-framed discourse analysis shows that the vaccine/treatment theme is the thinnest discourse strand (7.6%). This fits the absence of a domestically available chikungunya vaccine and exposes a discourse gap (a potential communication blind spot) for risk communication. • Sentiment was near-balanced (negative 40.9% vs positive 40.6%), and short-video platforms (Bilibili science videos exceeding one million views) functioned as the dominant knowledge carrier, so outbreak communication should be video-first and proactively address the "no vaccine / symptomatic treatment only" message. • Multi-platform, real-world evidence: 4,707 cleaned Chinese social-media records (2022–2026) provide the first cross-platform characterisation of public discourse on knowledge, attitudes, and practices toward dengue and chikungunya in China. • Outbreak signal validity: The July–August 2025 chikungunya attention spike on social media tracks the official Guangdong/Foshan epidemic curve, supporting social media as a near-real-time outbreak-surveillance signal. • Identified discourse gap: Vaccine/treatment discourse was the weakest theme (7.6%), reflecting the lack of a domestic chikungunya vaccine and pointing to a clear health-communication target. • Platform role: Short-video content (e.g., Bilibili explainer videos with >1 million views) was the main knowledge channel, arguing for video-first, proactive public-health messaging during arboviral outbreaks. • Methodological transparency: The study openly reports platform anti-automation constraints (Douyin/Xiaohongshu collected via manual extraction as exploratory supplements) and follows ROSES + STROBE reporting.