Web and social media searches highlight menstrual irregularities as a global concern in COVID-19 vaccinations

www.nature.com/scientificreports OPEN Web and social media searches highlight menstrual irregularities as a global concern in COVID‑19 vaccinations Ariel Katz1*, Yoav Tepper2, Ohad Birk3,5 & Alal Eran1,4,5* Delineation of public concerns that prevent vaccine compliance is a major step in generating assurances and enhancing the success of COVID-19 prevention programs. We therefore sought to identify public concerns associated with COVID-19 vaccines, as reflected by web and social media searches, with a focus on menstrual irregularities. We used trajectory analyses of web and social media search data in combination with global COVID-19 data to reveal time-dependent correlations between vaccination rates and the relative volume of vaccine and period related searches. A surge of period and vaccine related Google searches followed the introduction of Covid vaccines around the world, and the commencement of vaccination programs in English speaking countries and across the United States. The relative volume of searches such as “Covid vaccine menstrual irregularities”, “Covid vaccine menstrual period”, “Pfizer vaccine menstruation”, and “Moderna vaccine menstruation” was each significantly correlated with vaccination rates (Spearman r = 0.42–0.88, P = 4.33 × 10–34–1.55 × 10–5), and significantly different before and after the introduction of Covid vaccines (Mann–Whitney P = 2.00 × 10– 21 –7.10 × 10–20). TikTok users were more engaged in period problems in 2021 than ever before. International, national, and state-level correlations between COVID-19 vaccinations and online activity demonstrate a global major concern of vaccine-related menstrual irregularities. Whether it is a potential side effect or an unfounded worry, monitoring of web and social media activity could reveal the public perception of COVID-19 prevention efforts, which could then be directly addressed and translated into insightful public health strategies. Women around the world have reported a link between COVID-19 vaccinations and changes in their menstrual cycle regularity and i ntensity1. One of the early reports was a tweet by Kathryn Clancy, a Professor of Anthropology at the University of Illinois, which was followed by hundreds of responses of women identifying with her concerns, and in many cases reflecting on having similar menstrual irregularities following vaccination. Later reports on menstrual irregularities following vaccination were met with c riticism2, some suggesting that pandemic-related stress might be the cause of irregularities. Antivaxxers later highlighted the worries regarding possible vaccine-induced menstrual irregularities, expanding unverified and often manipulative concerns, and promoting fears of vaccine–associated abortions and infertility3,4. Recent studies have examined the effects of COVID-19 disease5 and vaccines6–8 on menstrual irregularities, suggesting lack of adequate reporting of such irregularities due to women’s reluctance to discuss these matters with their p hysicians9. Yet the public perception of a potential link between COVID-19 vaccines and menstrual irregularities remains largely unknown. To examine the spread and magnitude of this concern we mined web and social media search data and its relation to the specific timing of national and state-level COVID-19 vaccination drives. In doing so, we utilized Google Trends10, a powerful tool for analyzing population behavior and search trends11. Google Trends is capable of summarizing time and location-specific searches, enabling integrative spatiotemporal analyses, including those related to h ealth12. For example, recent infodemiological studies have leveraged Google Trends to model disease p revalence13 and s ymptomatology14, especially in the context of COVID-1915–17. 1 Department of Life Sciences, Ben-Gurion University of the Negev, Beer Sheva, Israel. 2Tel Aviv, Israel. 3Genetics Institute at Soroka Medical Center, Faculty of Health Sciences, Ben-Gurion University of the Negev, Beer Sheva, Israel. 4Computational Health Informatics Program, Boston Children’s Hospital, Boston, MA, USA. 5These authors contributed equally: Ohad Birk and Alal Eran. *email: ; Scientific Reports | (2022) 12:17657 | https://doi.org/10.1038/s41598-022-20844-x 1 Vol.:(0123456789) www.nature.com/scientificreports/ Methods We used Google Trends to examine the relative volume of searches for “Covid vaccine menstrual irregularities”, “Covid vaccine menstruation”, “Covid vaccine menstrual period”, “vaccine and period”, “Pfizer vaccine menstruation”, and “Moderna vaccine menstruation” worldwide, in English speaking countries and across the US between January 2020 and November 2021. Specifically, we used the “search term” functionality of Google Trends with no category. The settings of all analyses are recorded in all results files available at https://bit.ly/ covid_vaccine_period_search_data. We also mined TikTok, a social network serving more than a billion users, 53% of whom are female and 78% of whom are under the age of 2 418. Using Analisa.io (https://analisa.io), we examined the activity of “#periodproblems” from January 2019 to October 2021. We integrated the search data with COVID-19 vaccination data obtained from Our World In Data (https:// ourworldindata.org/covid-vaccinations) and other COVID-19 aggregated statistics, including the number of cases and deaths, obtained from https://github.com/CSSEGISandData/COVID-19/tree/master/csse_covid_19_ data. Our python code that performs and visualizes all analyses is freely available at https://github.com/SgtTe pper/C ovidT rends. Specifically, it mines Google Trends via pytrends, integrates the results with COVID statistics using pandas, computes Spearman correlations and Mann Whitney U tests using SciPy stats, plots co-occurring multimodal data on the same timeline using seaborn, and plots relative search volume by state using plotly. Our statistical analyses focused on assessing the relations between The Google Trends search volume index (SVI) and the normalized vaccination rate (NVR) for each search term of interest. SVI measures the relative popularity of a certain search at specific locations and during specific times. This relative index is normalized from 0 to 100, all with respect to the prespecified time interval and locations. NVR reflects the daily total of vaccinated individuals in the studied location, normalized such that the maximal number in the examined time interval is 100. In other words, it is 100*daily total of vaccinated individuals / maximal daily total of vaccinated individuals in the studied period. We utilized two complementary analytical approaches. First, we used Spearman correlation between SVI and NVR across the entire studied period to assess the monotonic relationships between SVIs of interest and the NVR for Covid in the studied location. Second, we used Mann–Whitney U tests to examine the significance of the difference in the distribution of SV (...truncated)