Social media to monitor prevalent diseases: Hay fever and Twitter activity in Germany.

Abstract

To the Editor, Allergic rhinitis (AR), also known as hay fever, is the most common allergic disease worldwide, affecting up to 30% of the global population.1 The analysis of publicly available, population-based data can be beneficial for novel insights and monitoring of the disease burden in populations.2 One source for these data is social media, which are gaining increased interest in medicine and public health. Twitter is among the most popular social media sites in Germany, with a market share of about 20%.3 Tweet counts were shown to correlate with local pollen counts in some countries.4 The aim of our study was to provide insight into the German AR landscape on Twitter and to identify influential regional climate factors for the future development of tailored awareness and prevention campaigns. A total of 43,965 tweets in German language containing the keyword “heuschnupfen” (hay fever) from 2018 to 2021 were found by querying the Twitter Academic API using the query string “heuschnupfen lang:de.” The keyword is searched against the tokenized tweet body (i.e., split by punctuation or spaces) and includes hashtags. The year range was chosen based on the available pollen data in Bavaria, which were kindly provided by the Center Allergy and Environment (ZAUM) and used as proxies for pollen counts in Germany. The tweets originated from the German-speaking countries such as Germany, Austria, and Switzerland; however, our analysis focused on Germany because it is the most populous country of the three by a large degree. Manual content analysis of 500 random tweets revealed that 392 tweets (78.4%) were presumably from allergic individuals. When aggregating AR-related tweet counts by month, the data showed a seasonal pattern, with peaks in the European spring months (March–May; Figure 1B). This pattern was consistent across all analyzed years (Figure S1). The mean AR-related tweet count over the 4 years rose from 223 per month in January to a peak of 1914 per month in April. The number of tweets per month correlated moderately with local total pollen counts and strongly with local birch pollen counts (Figure 2 and Table S1; for total pollen r = .59, p = .046; for birch pollen r = .7, p = .011). The number of tweets did not correlate with temperature and precipitation in Germany (data not shown). Previous studies found positive correlations between AR symptom severity with local temperatures and precipitation.5 Our data set unfortunately does not provide any more insight into this relationship, likely because it is too small to accurately reflect this complex relationship. The 10 accounts with the highest tweet volume included news sites, healthcare professionals, disease advocacies, and research institutions (Figure 2). These accounts published only a small fraction of the total number of tweets (between 101 [0.2%] and 2940 [6.7%] tweets per account during the analyzed period; percentage of total analyzed tweets), suggesting a diversified landscape of AR-related tweets with lack of central authority. The data are limited by the fact that social media data are subject to selection bias. For example, the Internet is primarily used by younger individuals for health-related information in Germany.6 In addition, tweet counts were only correlated to Bavarian pollen counts. However, since Bavaria is the biggest federal state by area and extends to about half of Germany's overall longitude and one third of Germany's latitude, this limitation is weakened to a certain degree. Indeed, the mean peak date in the years 2020–2022 for birch pollen between Munich in the south and Berlin in the north differed by only 1 day (n.s., data not shown). We assimilated monthly Twitter data and think that the differences in pollen season within Germany are much less than this resolution. Our analysis did not include other keywords beyond “hay fever,” which excludes AR-related posts with, perhaps, more unspecific keywords such as symptoms. In summary, our study highlights the relevance of social media for prevalent diseases like AR. Pollen count and, thus, disease burden correlated well with post volume. Tweet volume for hay fever can, therefore, be used to estimate disease burden. Furthermore, the study offers an overview of the relatively fragmented landscape of AR-related content in German language on Twitter. Social media platforms like Twitter have the possibility for not only monitoring public disease burden but also improving awareness about common diseases like AR and minimizing misinformation. S. S.—conceptualization, methodology, formal analysis, data curation, visualization, writing—original draft; H. W.—methodology, investigation, writing—review and editing; J. B.—supervision, resources, validation, writing—review and editing; T. B.—supervision, writing—review and editing; A. Z.—supervision, writing—review and editing. Data for this publication was obtained through the Twitter Academic API. Open Access funding enabled and organized by Projekt DEAL. The authors declare no conflicts of interest. The data that support the findings of this study are available from the corresponding author upon reasonable request. Appendix S1 Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.