The AstraZeneca vaccine crisis and the emergence of the Omicron variant were two critical events that shaped the pandemic management and impacted public discourse. This study aims to examine the key issues and controversies that dominated these two events. With a sample of 3,748,302 tweets in English, French, Portuguese, and Spanish, we provide a comprehensive representation of the public discourse during these crises. During the AstraZeneca COVID-19 vaccination crisis, the term “blood clot” was the most frequently used in all communities. However, unlike the Spanish and Portuguese communities, the English and French communities had a significant presence of tweets discussing the most severe risk category, “death”. Only the Portuguese tweets made direct references to a politician, specifically President Bolsonaro of Brazil. Additionally, the English and Spanish communities had more positive and neutral sentiments compared to the French community, which had the highest percentage of negative sentiment followed by the Portuguese community. Regarding the Omicron crisis, the public discourse focused on data and control measures, with a particular interest in cases and deaths, which had a high presence among the Spanish and Portuguese communities. Overall, the dominant sentiment towards the Omicron variant was negative, although the Spanish community showed a more balanced sentiment between negative, neutral, and positive.
4.1. The AstraZeneca Crisis
Our findings indicate the public may be more concerned about the potential for illness associated with this particular vaccine because there is a higher presence of the term “blood clot” in the four communities associated with the AstraZeneca vaccine compared to the other keywords searched. Although this concern, in the instance of a thrombus, does not correspond to the real risk, topics that people perceive to be a threat to their health tend to attract more public attention, reaching larger implications in their networks. According to the Pharmacovigilance Risk Assessment Committee (PRAC) of the European Centre for Disease Prevention and Control [
30], the possibility of a person vaccinated with AstraZeneca developing a thrombus is highly rare.
Furthermore, if we compare the presence of the term ‘Blood clot’ among the four communities, the most striking case was the tweets in French, where the value is much higher, demonstrating the impact of the topic in this specific community. The term side effects also stand out more in the French-speaking community, with a value of almost double, compared to the other languages. France does not have a long history of anti-vaccine lobbying, as is the case in Great Britain [
31]. In the 1990s, France faced controversies and resistance to vaccines for containing adjuvants [
32], but a big change has occurred in the last 10 years when this country became one of the most resistant to vaccines in the world [
18]. The French health authorities exercised significant pressure on those who were not vaccinated against COVID-19 by restricting access to areas that were at the time off-limits to those who fit into this category. Currently, 78% of the population has received the full course of COVID-19 vaccinations [
33].
The word “deaths” is more frequently used in tweets from the English- and French-speaking communities, which further implies that there is serious concern about this vaccination given that it is associated with the most severe health risks conceivable. In the case of the Hispanic and Portuguese communities, references to deaths were much lower and did not seem to be a focus of interest related to the AstraZeneca vaccine for these users.
The two most frequent words that coincide with all four communities refer to the terms vaccines and Pfizer. Pfizer was the first pharmaceutical company to have its vaccine authorized for marketing in the European Union and the United States. If we look at how each community behaves, we can see that, in addition to Pfizer, other pharmaceutical companies have also been cited. The communities that cited them most in the public discourse on AstraZeneca were the Hispanic community and the Portuguese community. It is remarkable that when AstraZeneca is mentioned, reference is made to other pharmaceutical companies, possibly as a comparison between the vaccines of different companies. The discourse appears to be less polarized and more generalist among the Spanish- and English-speaking communities, with a focus on broader subjects such as immunization, booster shots, and health in general. With respect to the vaccine schedule, Spain has a history of achieving high immunization rates. This is also the case with the COVID-19 vaccination, with a complete schedule of 92.8% for those over 12 and a first booster dose of 55.1% [
34].
The tweets in Portuguese cite scientific sources, such as the World Health Organization and Butantan, a Brazilian institute that produces the CoronaVac vaccine using ingredients from other countries, such as China. Following this same logic, the country also produces the AstraZeneca vaccine, but by the Oswaldo Cruz Foundation (Fiocruz) [
35]. In addition, the words Brazil and Bolsonaro stand out in the tweets of the Portuguese-speaking community, a reference to the president of this country Jair Messias Bolsonaro (his term of office ended on 31 December 2022), being the only direct reference to a politician that appears in the whole analysis. Brazil is the country with the second highest number of deaths due to COVID-19, reaching on 25 November 2022, the mark of 669,665 thousand deaths [
36]. The initial management of the health crisis was quite controversial, marked by strong politicization during the pandemic and rejection of science [
37,
38]. The Brazilian president has accumulated controversies, such as saying that COVID-19 was a “gripezinha” (little flu), defending herd immunity and the use of chloroquine as a preventive treatment against COVID-19 [
37], even after the World Health Organization did not indicate this drug to prevent the disease [
39]. Since the president has often claimed in the media that vaccines are ineffective while yet emphasizing their potential adverse effects, he has not received any vaccinations [
37,
40]. Another point to highlight is Bolsonaro’s frequent use of social media to communicate with his public and disseminate his ideas through this channel [
41]. These may be a few of the causes behind this politician’s singular presence in the public discourse during the crisis.
The term “thrombus” stands out among the most often used terms in French, ranking fifth, along with the words ‘effects’ and ‘death’, confirming high-risk perception and consequently vaccine hesitancy. The French-speaking community appears to have quite strong opinions on the AstraZeneca vaccine, according to our findings. The low prevalence of neutral sentiment in this community demonstrates how strongly most individuals are positioned on this topic.
Across communities, sentiments on the AstraZeneca vaccination are typically polarizing. Our findings contrast with those made by Marcec and Likic [
17] and Mahyoob et al. [
28] who looked at tweets written in English about the same vaccine and found that sentiment in the Anglo-Saxon population is primarily negative, while our findings show that sentiment is predominantly positive, followed by negative and neutral sentiments. These two studies’ analyses, nevertheless, comprised a different time period a little earlier than the one used in our study. Concerning the Hispanic community, the sentiment was slightly similar, being more neutral, and followed by positive, and negative ones. The most negative sentiment was found in the French-speaking and Portuguese-speaking communities. In relation to this, it is commonly known that Twitter contains a significant amount of anti-vaccine lobbying. With these regards, efforts to fight anti-vaccination [
42] must (a) closely collaborate with technological platforms to address anonymous anti-vaccine tweets; (b) concentrate efforts on misleading information; and (c) go beyond conventional factual methods, such as identifying, labeling, or eliminating fake news, to address the emotions brought on by personal memories, values, and beliefs.
In light of the findings from our study, it is crucial to provide a more comprehensive understanding of the impact of available information on vaccine acceptance. A previous study revealed strong knowledge and acceptance of COVID-19 vaccination among Italian undergraduates [
43]. This highlights the effectiveness of the information strategy accompanying the national immunization campaign. It is noteworthy to mention that the perception of negative health consequences related to vaccines experienced a significant increase following the precautionary suspension of Vaxzevria, the AstraZeneca vaccine for COVID-19. With these regards, a more robust understanding of the broader impact of information on vaccine acceptance should be achieved, emphasizing the need for continued investigation and targeted interventions to address vaccine hesitancy.
4.2. The Omicron Crisis
According to our findings, risk-related terms, such as ‘infects’ and ‘deaths’, were used more frequently by users according to the word search, rather than ‘mask’ and ‘risks’. Since nobody knew if the Omicron strain could be as lethal as its predecessor, the Delta strain, which had previously predominated until then [
10], the development of this variant happened at a time when there had already been 5 million deaths globally [
36]. When an event poses an unknown risk, as in the case of the coronavirus initially and the omicron variant later, may provoke an emotional reaction to decision-making regarding this, risk perception is more impacted [
13]. This phenomenon has also been reflected in our data since the perception of risk and fear has a strong relationship with the emergence of the new variant. In this sense, people’s interest in focusing on these subjects may be an indication that their perception of risk may be closer to the real risk, which is becoming infected. In this sense, perceiving this strain as a risk to their health could lead people to take attitudes towards prevention [
12], such as getting vaccinated, and using masks, among others.
We have identified the 16 most frequent words in each community. The first place is taken by the word ‘variant’, present in the four communities and cited more than 300 thousand times. There are others related to this term, such as ‘Delta’, ‘virus’, ‘Ba2′, ‘infection’, and ‘cases’. The word ‘vaccine’ stands out as the second most used word in the Portuguese and French-speaking communities, with more than 100 thousand citations. Other terms in the four languages related to the vaccine are also observed, such as ‘dose’, ‘immunity’, ‘vaccinated’, ‘Pfizer’, ‘booster’, and ‘rate’. These references may be connected to the fact that Omicron’s arrival left the public uncertain as to whether vaccines would continue to protect against the new variety, despite the WHO’s repeated statements that vaccines play an important role in preventing serious diseases and deaths [
10].
The Hispanic case is highlighted by several references to preventive measures to avoid the spread of the virus. From the public health point of view, prevention seeks to ensure the protection of diseases, reducing their incidence and prevalence in the population [
44]. Spain was one of the countries that have maintained some preventive measures against COVID-19 for a long period, such as the use of masks indoors when other European countries had already relaxed these rules [
45].
It is worth noting that, in general, the discourse on Omicron is quite generalized in the four communities, with specific mentions to Africa, related to the fact that omicron was identified for the first time in this continent, and also references to the pharmaceutical company Pfizer. In relation to scientific institutions, the only one that was highly cited was the WHO in the Hispanic and Portuguese cases, showing the importance of this international organization in the public discourse during health crises.
Sentiments towards Omicron in the Spanish, English, and French communities are predominantly negative, i.e., the perception that this variant represents a risk to users is high. Our results coincide with those found by [
29], when analyzing tweets in several languages, they found quite negative sentiments related to this variant. On the other hand, in our study, the case of the Portuguese community was dominated by neutral sentiment similar to those found by [
30], showing that 50.5% of the posts were neutral, the other emotions being composed of sentiments such as ‘bad’, ‘good’, ‘terrible’ and ‘great’. Finally, it is important to note that a previous study concluded that vaccination plays a key role in dropping the negativity of people, thus promoting their psychological well-being [
46].
4.3. Limitations
This study has some limitations. Our analysis did not allow us to identify geographically the origin of all users, which limited us in terms of analyzing the data taking into account the socio-health contexts of these countries, except in cases where it was possible to detect specific references to countries in the posts published. However, we consider that this strategy also has its strength in the sense that people, by sharing a language, form a virtual community in which topics of interest are discussed. In addition, although our search has completed four languages covering several countries, it is quite concentrated in some regions, for this reason, we consider that in future research it would be relevant to broaden this horizon to include other languages in order to ensure more diversity. The fact that our paper could only analyze text and not visual content was another limitation of the analysis. We recommend further research on this topic given the importance of visuals during health crises [
47,
48].
The purpose of using automatic techniques is the possibility of obtaining an overall balance of all the data, where it would be impossible to carry out an individual evaluation. This procedure evaluates a concept subject to a certain subjectivity, which is why we have used these more quantitative results obtained previously. Conducting a sentiment study and applying it to independent conclusions could lead to inaccurate results. This methodology, in isolation, may have limitations, which is why this study has been combined with the previous analyses, allowing global conclusions to be drawn from the data analyzed.
Finally, although we have not carried out a detailed analysis of the topics investigated, we want to highlight the fact that working with a large volume of data has allowed us to draw a general overview of the perception of risks related to the events investigated. With these regards, our paper contributes to knowledge in four ways. First, it broadens the scope of crisis communication by revealing social media’s role in the constructs of public discourse. Second, we suggest that social media is the optimal place to explore risk perception during a health crisis. Two case studies were used to support this claim. Third, we stress the cultural diversity amongst global communities, demonstrating the necessity of tailoring crisis communication strategies to social and cultural traits. Fourth, while our study does not delve extensively into the nuanced interpretation of public sentiment, we believe that it lays the foundation for future research and provides valuable insights into the role of social media in shaping public discourse during health crises. By uncovering the constructs and patterns of public discourse, we contribute to the broader field of crisis communication.