Combating misinformation with internet culture: the case of Brazilian public health organizations and their COVID-19 vaccination campaigns

2.1 Misinformation combat by public health organizations on social media

As a result of the COVID-19 “infodemic,” both the production of and research about misinformation, that is, deliberate or unintended false or inaccurate information, has increased (Nascimento et al., 2022). An often-overlooked stakeholder group in research on misinformation is public health organizations, such as health ministries, medical research institutions or health associations. Social media provide public health organizations and their executives who are involved in health communication with an opportunity to reach large subsets of the public and provide them with crucial information, for example, during pandemics and other health crises (Mirbabaie et al., 2020). However, the verification of available information from a variety of official and unofficial sources poses a particular challenge for public health organizations (Rathore et al., 2021). In this vein, public health organizations not only have the mandate to react to health crises (e.g. misinformation combat) but also to inform the public actively and accurately as a trustworthy source (e.g. misinformation prevention) and to include the public in the discourse (Muhammed and Matthew, 2022). One of the concerns of research in this field is the effectiveness of communication strategies that may help organizations counteract and prevent misinformation (Mirbabaie et al., 2022). For example, a study conducted by Vafeiadis et al. (2019) suggested that public health organizations counter the spread of misinformation by discrediting the source of the misinformation instead of dealing with a single piece of misinformation (e.g. a tweet). By doing so, the credibility of the source can be reduced, and the public is less likely to fall victim to the threat. Lewandowski and van der Linden (2021) structured the process of misinformation into three distinct phases: prevention, detection and debunking. Table 1 provides an overview of these phases.

Existing research heavily focuses on the reactive phases of misinformation containment such as detection (Asr and Taboada, 2019; Shu et al., 2017) and debunking (Chan et al., 2017). According to Lewandowsky and van der Linden (2021) and Ecker et al. (2022), the preventive phase of misinformation is largely overlooked. Research that explains the mechanisms of misinformation prevention, however, could contribute to our understanding of reducing the amount of misinformation that is created in the first place. Certain topics or events are highly susceptible to misinformation, one of which is vaccination (Vemprala et al., 2021). To prevent misinformation about vaccinations on social media, public health organizations have employed a variety of strategies. One approach is to use social media platforms to promote accurate information about general health topics. This can include the distribution of educational videos, surveying public attitudes or prebunking misinformation (Tsao et al., 2021).

Apart from exploring strategic measures to apply in each phase, another facet of research in this domain is how misinformation is disseminated on social media platforms such as Twitter. Features such as novelty and efficacy (Song et al., 2023) as well as the salience and trustworthiness of the source (Di Domenico et al., 2021) affect the decision to share misinformation on social media. In political contexts that are characterized through social division such as in the case of Brazil, perceiving the political opponent groups as more susceptible to believing in misinformation can make it more difficult to approach these issues (Yang and Tian, 2021). In their frame analysis of Twitter content about the COVID-19 pandemic, Wicke and Bolognese (2020) found four themes of contextualizing information, that is, frame packages: “communications and reporting,” “community and social compassion,” “politics” and “reacting to the pandemic.” The authors also verified the high recurrence of war metaphors in the analyzed sample, categorizing them as alternative frames such as “storm,” “monster” and “tsunami.” In addition, Park et al. (2020) observed that nonmedical frames such as personal stories about the pandemic were being shared more often than factual medical information.

2.2 Framing health information with internet culture

Framing theory explores how individuals use mental frameworks to make sense of the world around them. According to Goffman (1974), frames or frame packages are process-oriented and strategic resources that are constructed by an individual or a group. They mentally support the categorization, organization, and interpretation of information by providing a contextual background and making meaning out of isolated pieces of information. Frames are not neutral but rather are influenced by our social and cultural experiences, as well as by the context in which we encounter them (Walter and Ophir, 2019). For example, a frame used to interpret an event in Germany may differ from the frame used to interpret the same event in Brazil. Being situated in a competitive environment, frames are often constructed to achieve predetermined outcomes. The focus hereby lies within cognitive processes that interact with frames to produce certain effects (Carragee and Roefs, 2004). In this study, we operationalize framing theory by following established approaches to frame analysis. Table 2 shows an overview of the core concepts of frame analysis.

Framing has been widely researched in both offline and online environments (e.g. Baden, 2018; Ylä-Anttila et al., 2022). In this study, we are particularly interested in frame packages and frame devices that originate from internet culture. The latter refers to practices, values, and beliefs that shape the production, circulation and consumption of information on the internet (Porter, 1997). It is shaped by the technical infrastructure of the internet, including contemporary social media platforms, as well as by social and cultural factors. The internet, and social media specifically, has increased the importance of visual framing (Brennen et al., 2021). Visual frames can be categorized across denotative (what they depict), connotative (what metaphors are used), symbolic-semiotic (how they are composed) and ideological dimensions (what interest they promote) (Rodriguez and Dimitrova, 2011).

Framing health communication in the context of internet culture involves considering how to effectively present health information to a wide and diverse audience, for example, on social media (De Choudhury et al., 2014). Clearly, social media have transformed the way that health information is accessed and shared, which has implications for how health communication is framed and presented by public health organizations (Zhao and Zhang, 2017). The COVID-19 pandemic provided us with many opportunities for observing the importance of internet culture for raising public awareness regarding the pandemic. Elements and actors native to internet culture such as memes and digital influencers were important factors in the process of making sense of the pandemic and in the process of community bonding (Pulos, 2020; de Saint Laurent et al., 2021; Beck, 2022). In addition, the relevance of elements from internet culture as part of science communication has been debated for engaging new audiences (Gartley, 2022), with potential positive impacts on vaccination campaigns.

Brazilian internet culture is a vivid example of the entanglements of entertainment, public debates and political activism (Amaral, 2016). Public institutions in the country have adopted this understanding and imbed elements from internet culture elements in their communication. An example is the adaption of the Brazilian funk song “Bum Bum Tam Tam” for promoting the CoronaVac vaccine, which was manufactured in Brazil by the research center Butantã Institute (Barretto Briso, 2021). In January of 2021, when Brazil was starting its vaccination campaigns, thousands of Twitter users mentioned the music video because its pronunciation sounds like “Butantã.” After the spontaneous buzz on social media, the institute invited the song's artist Mc Fioti to rerecord the music video to encourage people to get vaccinated. The viral music video has spawned new versions and memes that circulated on the internet in the first months of 2021, representing an antithesis to the Brazilian federal government's suspicious opinion on COVID-19 vaccines.

To better understand the phenomenon of internet culture as a means against the spread of misinformation, this study investigates the emergence of alternative frames on social media in a highly complex political context. The guiding research questions of this work are as follows:

Methodologically, we combine qualitative frame analysis with social media analytics (SMA) techniques to investigate the framing used by public health organizations in Brazil for COVID-19 vaccination campaigns. The design of the empirical investigation is outlined in the following section.

3.1 Case description

The case at issue deals with Brazilian public health organizations, which faced unprecedented challenges during the COVID-19 pandemic. Brazil's president at the time, Jair Bolsonaro, actively engaged in publicly conveying the government's distrust of vaccines, especially in the first months ensuing the delivery of COVID-19 vaccinations. Furthermore, some of Bolsonaro's political opponents supported vaccination campaigns both as an important public health measure as a political strategy. Concerning public health organizations, this complex political context resulted in a much fragmented and decentralized strategy for communicating vaccination campaigns to the public through social media, especially on Twitter. Consequently, federal institutions were divided between managing the openly proclaimed anti-vaccination position of the president, trying to attenuate and even to reframe it and spreading information regarding the national vaccination campaign, encouraging the population to get vaccinated as soon as possible. Meanwhile, local institutions such as state and city health secretaries defined their communicational approaches by aligning them with the political positioning of their leaders, sometimes confronting Bolsonaro's statements openly (Gramacho and Turgeon, 2021). The polarized political environment paired with the sensitive issue of mass vaccination against COVID-19 makes this case particularly interesting for studying the role of public health organizations and their efforts to combat misinformation.

Further, the public engagement in the debate about COVID-19 vaccination through social media was particularly high, as observed during the broadcast of the Parliamentary Investigation on Bolsonaro administration's negligence concerning the pandemic. The investigation was followed by millions of Brazilians via social media platforms such as Twitter and Twitch, with the former being central to the communication between the audience and the Brazilian parliamentarians. Some Twitter profiles deployed fandom-oriented techniques, such as popular culture references and inside jokes to engage the population in political activism against Bolsonaro's support for anti-vaccination politics (Declerq, 2022).

3.2 Data collection

Against the backdrop of this case, we aim to understand how different Brazilian public health organizations managed to deal with this complex political context in framing the COVID-19 vaccination campaign on Twitter. We purposefully selected a sample of 24 Twitter profiles of Brazilian public health organizations and individuals representing those organizations. The selection criteria for the profiles were the following: (1) profiles of official public health organizations that belong to Brazilian government bodies, (2) profiles of high-level executives (such as the Minister of Health) of such organizations, (3) profiles of public-funded health research institutions working in the fabrication and distribution of COVID-19 vaccines in Brazil such as the Butantã Institute in São Paulo, and Oswaldo Cruz Foundation in Rio de Janeiro, respectively responsible for Coronavac and Oxford-AstraZeneca vaccines, (4) profiles of prominent states' and cities' secretaries of health and (5) nongovernmental health organizations. In addition, the Minister of Health's profile was added to the sample because it constituted an important source of information concerning the vaccination campaign. Table 3 shows an overview of the analyzed profiles.

This work follows a SMA approach as suggested by Stieglitz et al. (2018), combining automated data collection with manual coding. After the first step of concretizing the research domain, the second and third steps of the framework, tracking and preparation, will be laid out in this section. Moreover, we followed the SMA ethics guidelines suggested by Jung et al. (2022). The data for this case study were tracked with the standard Twitter search API. For this analysis, we collected all postings from the Twitter profiles of the 24 Brazilian public health organizations and individual representatives (see Table 1) during a three-month period, starting in January 2021, which marked the beginning of the COVID-19 vaccination campaign in Brazil. The tracking resulted in a dataset of 77,527 tweets. Subsequently, we selected the postings containing the keywords “corona” and “COVID,” considering that these filters are also able to select postings containing the terms “coronavirus” and “COVID-19.” We chose not to use vaccination-related keywords because it would have resulted in also selecting tweets about the vaccination campaigns for other diseases, such as poliomyelitis and influenza. This keyword-based filtering resulted in a subsample of 14,666 tweets, excluding retweets and answers on tweets through Tableau's text filter. All tweets that began with “RT” (retweet) or “@” (answer to a tweet) were filtered from the dataset to avoid duplicates. Hereby, all remaining tweets are original tweets. Given that the qualitative assessment of more than 14,000 tweets was impracticable for a manual coding approach, we used Excel's RAND function to randomly select a final sample of 400 tweets, resulting in a confidence level superior to 95% (Taherdoost, 2017). Moreover, we repeated the keyword-based filtering in the subsample of 14,666 tweets selecting postings containing the terms “fake,” “falso,” “falsa” or “mentira” (“fake,” “false” and “lie” in Portuguese) and manually selecting publications regarding the vaccination campaign. This second filtering resulted in the selection of 136 additional tweets related to misinformation about the COVID-19 vaccination campaign. The final sample for the manual coding thus consisted of 536 tweets.

3.3 Frame analysis

For the analysis (step 4 of the SMA framework), we chose to conduct a frame analysis as it constitutes a useful methodological approach for the scrutiny of strategic communication on social media. Frame analysis emerged from framing theory (Goffman, 1974) and focuses on the observation of interaction processes, with an emphasis on the variance between specific points or episodes in conflict interaction (Dewulf et al., 2009). The identification of framing devices—that is, specific linguistic structures, such as metaphors, visual icons, and catchphrases that communicate frames—is therefore important (Gamson and Modigliani, 1989).

Contemporary research studies that employ frame analysis as an empirical method use both deductive and inductive approaches. Deductively conducted frame analyses operationalize framing theory by using a predefined set of frames known from the literature and categorizing the data accordingly (Ylä-Anttila et al., 2022). However, this approach might not result in the identification of unknown framing devices and frame packages and might restrict the scope of the analysis. Consequently, as Walter and Ophir (2019) suggested, inductive approaches to frame analysis are suitable for identifying patterns of framing not based on theoretical relationships but on the construction of a context-specific data corpus (e.g. Allen et al., 2021). In this case, larger topical themes of framing activity (frame packages) are derived based on inductive grouping of linguistic or platform-specific elements (frame devices). Qin (2015) argued that framing devices for Twitter are specific, since the composition of tweets differs from classic news articles, which are the subjects of earlier frame analyses. Elements such as hashtags, mentions, attached links, images and other items were found to be framing devices in previous frame analyses of tweets (Qin, 2015; Xiong et al., 2019). They differ from classic framing devices in legacy news, which contain headlines, subheadings, photo captions, logos, statistics, charts and other elements (Tankard, 2001). Refer to Table 2 for an overview of the core concepts of frame analysis.

The coding process was carried out by two researchers who were native speakers in two phases. In the first phase, the two researchers approached the data openly to inductively determine frame packages in the posts of Brazilian public health organizations and classify them (Walter and Ophir, 2019). First, we were particularly interested in the way Brazilian public health organizations framed content related to vaccination campaigns and misinformation about them. Second, we were interested in the role of Brazilian internet culture in this process, which might differ from preexisting frames developed in previously analyzed contexts (Wicke and Bolognese, 2020). To determine the frame packages, we considered the assumed goals of the health organizations according to the context of the COVID-19 pandemic in Brazil, presupposing that the main objective of their communication was to educate and engage the public regarding this subject. Therefore, the frames packages were classified according to what the coders assumed as the intention behind the postings, such as informing about the vaccination schedules in one city, debunking misinformation or inviting the adoption of preventive measures. In the second phase, the sample was analyzed again addressing specific framing devices. Social media framing devices were deductively classified according to Qin (2015) and Tankard (2001). Emerging frame devices were inductively categorized if needed. The frequency of frame packages and framing devices was quantified to be able to rank them by occurrence. The combination of quantitative and qualitative approaches in this frame analysis aims to provide multiple perspectives on the complexities of the case (Touri and Koteyko, 2015).

4.1 Frame packages used by Brazilian public health organizations

We identified 14 emergent frame packages. The most frequently addressed frame package in the sample was case numbers,” followed by “fact checking,” “vaccination campaign information” and “public policies.” The three least recurrent frame packages were “vaccine information,” “personal stories” and “debunking fake news.” An overview of all identified frame packages is provided in Table 4.

An analysis of the most liked and retweeted posts among the sample revealed that their engagement metrics tend to be higher in the frame packages addressing the pandemic statistical data and giving instructions on how and when to get vaccinated. We also found that 195 (36.4%) of the postings provided some numeric or statistical information. The most liked frame packages, on average, were “public policies,” “vaccination campaign information,” “vaccination coverage report” and “case numbers.” The most retweeted frame packages were similar: “public policies,” vaccination campaign information,” “case numbers” and “vaccination coverage report.” The most liked posts varied between 6,946 and 232 likes, and the most shared posts ranged between 1,104 and 31 retweets.

The content of the most engaging frame packages in both samples emphasized positive feelings and efficacy, such as tranquilizing news about public policies, debunking upsetting misinformation concerning vaccines and clear instructions on how to get vaccinated. We also observed that postings classified in the framing packages “personal stories” and “vaccine information,” both appear among the most engaging tweets, despite occurring only 5 and 11 times in the sample. The same applies to the high recurrence of numbers and statistical data, which can also be interpreted as an effective strategy to inform the population about the status of the vaccination campaign.

4.2 Framing devices used by Brazilian public health organizations

Aside from Twitter-specific frame devices that were previously identified by the literature, we discovered additional salient frame devices relevant in this case: “infographics,” “pop culture references,” “recognized authority” and “internet-native symbolism.” Images and hashtags are the most frequent framing devices followed by links, numbers and statistics. Their quantitative distribution is shown in Table 5.

The salience of images within the tweet sample suggests that framing information visually is central for the vaccination campaigns of Brazilian public health organizations. Some organizations shared screenshots of fake news pieces that circulated social media for debunking purposes. Memes and pop-culture references were used to appeal to specific target groups and fan communities. Infographics were used as a design technique to show numeric and statistical information. Photographs were shared to provide an insider perspective and to convey accessibility and transparency. Those images (and videos) mostly showed journalistic images of hospitals, vaccine bottles, vaccination locations and politicians and the organizations' employees inspecting hospitals or attending official events.

One image shows the Fiocruz Foundation in Rio de Janeiro. In the middle, an inspection of a hospital is depicted, working as a proof that this event took place (denotative dimension). Another image shows an infographic of the vaccination coverage taxes in Porto Alegre city. It shows on the left the number of first doses administered in the city, and, on the right, the number of people who have already received two doses of the vaccine. In the center, we can read the total of administered doses and below that is the number of booster doses. The image title can be translated as “Vaccinometer” and is a wordplay with “Velocimeter.” The emphasis on numbers (symbolic-semiotic dimension) and the velocity metaphor (connotative dimension) are deployed to convey the efficiency of the vaccination campaign. Other images posted by public health institutions include montages with screenshots of misinformation or fake news that are being debunked. In one case, the Sao Paulo state government uses a screenshot to reference a specific video containing misinformation. The translation of the accompanying text reads as follows: “The video that shows an empty syringe in vaccine administration in Ceará is fake. True information can only be found through official channels. São Paulo State Government.” This approach to visual framing is also an example of addressing the denotative dimension; however, the reality depicted is not an event or public health measure, but a fragment of the fake news piece itself.

In combination with other framing devices, we also found instances of public health organizations using “pop culture references.” One of them depicts the Brazilian vaccination mascot, Zé Gotinha—translated as “Little Drop Joe”—which has been in existence since the 1990s (Pôrto and Ponte, 2003). The mascot was used by pop singer Olivia Rodrigo on the cover of her album Sour, which was released in 2021. The image was shared by the Rio de Janeiro City Secretary of Health. It also contains the vaccination schedule for teenagers aged 13 and above, and in the post in which it was shared, we could read a wordplay with “Good 4 U,” the name of the singer's most well-known song, and a reference to the name of the Olivia Rodrigo's fandom, “Livies”:

Good 4 U that the most anticipated moment has arrived for the teenagers aged 13: the covid-19 vaccine! “Livies” or not, take your colorful stickers and glue them on your planner, so you don’t forget! It’s at the top of the charts and Droplet Joe confirms: getting vaccinated is good for you! #JustComeOn

Other posts had references to pop culture products, such as video games and Brazilian popular TV shows, or addressed “internet-native symbols,” such as memes. One compares the COVID-19 vaccine's second dose to a new stage in the Super Mario World video game. The text says: “You beat the stage on the video game, but are you stagnated in the vaccination? Tsc tsc … Get up and go to get the second dose!”

In the example in Figure 1 (left), we can read “Wake up girl!” (In Portuguese: “Acorda menina!”), which is known in Brazil as the popular TV frontwoman Ana Maria Braga'“s catchphrase, used here to draw attention to the affirmative “every vaccine is good.” The text outside the box says, “Getting Vaccinated Sampa” (“Sampa” is an informal abbreviation of “São Paulo”). Inside the box, we read, “Wake up, girl! Getting the COVID-19 vaccine protects you against severe forms of the disease. #AllVaccinesAreGood.”

The second example in Figure 1 (right) is a reference to a Brazilian popular meme, a photo of two young women subtitled as “Se juntas já causam, imagina juntas” (translation: “If we already rock together, we also imagine together”). In the context of the public health organization's tweet, it is used to address the importance of getting the first and the second doses of the COVID-19 vaccine.

This analysis showed that “pop culture references” and “internet-native symbols” were deployed strategically by Brazilian health organizations to approach specific groups that had certain identification or affective bonds with those references, especially fans. As fans are interpretive communities with shared practices and cultural references, fandoms are attractive to political coopting. Further, these characteristics make it possible to use the same shared practices and references for community bonding. Considering this, “pop culture references” and “internet-native symbols” such as memes provide public health organizations with powerful framing devices, for example, to deal with a context of serious social division.

This same logic applies to individuals who can be classified as social media influencers, which we grouped under the framing device of “internet-native symbols.” One public health organization sought the support of a local social media influencer to support the COVID-19 vaccination campaign. In a post from the São Paulo City Secretary of Health, the institution shared a photograph of the famous Brazilian fashion social media influencer and writer Bruna Vieira, who has more than 394,000 followers on Twitter. She is depicted showing her vaccination card to the camera and smiling.

The translation of the accompanying text reads as follows: “Used to put her feelings on the paper, @brunavieira, 27, let the emotions flow as getting vaccinated in São Paulo this Wednesday (4). Keep your eyes on our social media feed and, when your time arrives, do it like her: get vaccinated! #VaccineSampa #AllVaccinesAreGood”

These shared experiences supported a collective narrative of the crisis and, combined with other internet-native symbols such as memes, were central in the shaping of public opinion around subjects such as social distancing and vaccination. Social media influencers have their own fandoms and can appeal to already existing communities, but the personal experiences of people who are not famous can also be very engaging, as exemplified in this analysis by the frame package “personal stories.” When a social media influencer shows support for a public health measure such as vaccination campaigns, which in the Brazilian context is also interpreted as a political opinion, the fandom tends to follow this tendency. Appealing to community-shared references and fan communities, the “pop culture references” and internet-native symbols' framing devices were related to the connotative dimension of visual framing. Memes, catchphrases, pop culture elements and digital influencers were used to add symbolic meaning to these images, mostly to position the public health organization as an actor who earns membership within an internet-native audience or community.

Within the more general strategic dimension of misinformation prevention, we found that public health organizations frequently used the framing device “recognized authority.” This included external scientists and individuals holding high-level positions in the public health organizations themselves. The approach here was not to appeal to fandom communities but to use these individuals' authority to validate the vaccination campaign. The engagement metrics of these postings were not very impressive, ranging between 0 and 66 retweets and 0 to 304 likes. Apart from visual framing devices, public health organizations frequently used hashtags. An analysis of their distribution revealed a lack of organization and unicity in the overall strategy of Brazilian public health organizations to promote the COVID-19 vaccination campaign with hashtags. We found 197 hashtags in our sample. This number is comparatively high as each local government created its own hashtag. Of those hashtags, 111 were mentioned only once. In some cases, we also highlighted the variations of spelling, such as “#Covid19,” “#Covid-19” and “#Covid_19,” which makes it more difficult for the regular Twitter user to receive all information about the pandemic from one search query. Local hashtags can be important framing devices in helping users find information regarding the vaccination campaign in their places of residence, but without a global hashtag indexing the entire Brazilian national campaign, it remains difficult to discover a specific local hashtag. In the Brazilian context, local hashtags also have a political aspect as framing devices: they indicate the local governments that are supportive or critical of President Bolsonaro.

Furthermore, despite the states' and cities' individual hashtags, the most mentioned ones were very generic and did not focus specifically on the vaccination campaign. They included general terms such as “COVID-19,” “coronavirus,” “fake news,” “health” and “bulletin”; institutional and organization’ names, such as SUS or Conass; and Brazilian states such as Minas Gerais (MG) or Rio de Janeiro (RJ). The hashtag #SUS, a reference to the Brazilian Universal Healthcare System (in Portuguese, “Sistema Único de Saúde”), is important due to its centrality in political disputes around Bolsonaro's positionings regarding universal public healthcare. The president's critics used the hashtag to show support for universal healthcare in Brazil and to criticize Bolsonaro's statements against it. Moreover, public health organizations and Bolsonaro's supporters used the hashtag to attribute the success of the vaccination campaign to the alleged good management of SUS in Bolsonaro's government. This usage of the hashtag #SUS is comparable to tactics used by fandoms, which promote massive postings of specific hashtags to make the subject of fan interest more visible. In this sense, public health organizations did not use local hashtags as effectively as some fandom communities did.

5.1 The role of internet culture for misinformation combat in Brazil

The findings of our analysis with regards to RQ1 highlight that Brazilian internet culture and the complex political situation in the context of the COVID-19 pandemic are represented in public health organizations' use of framing devices. The presence of “internet-native symbolism” and “pop culture references” in the analyzed sample is surprising considering existing knowledge about misinformation prevention (Allcott et al., 2019). However, it is not a surprise when considering the history and context of internet culture in Brazil. As Amaral (2016) showed in her discussion of the Brazilian culture of memes on Twitter, Brazilian internet culture is quite unique when viewed from a perspective typical for the Global North. Cultural differences, for instance, can be observed in relation to how humor is used, the appropriation of global media franchises (or celebrity cultures) produced outside of Brazil to its social-political context and the constitution of a Brazilian internet culture that reveals contradictions and issues on gender, generations, age, social classes and races. The intersectionality between issues in these categories has also appeared in Vieira's (2021) analysis of memes in the Brazilian Twittersphere. In terms of the political situation, we observed some aspects of how Brazilian political disaggregation affects its public organizations' misinformation prevention strategies.

Tweets relying on “internet-native symbols” and “pop culture references” can be considered good examples of the alignment of public vaccination campaigns with internet culture (Amaral, 2016). Those postings, however, did not necessarily show higher engagement metrics than other postings. The tweets containing images with more likes and retweets often provided information about the number of hospital beds and vaccine doses available to the population, the COVID-19 contamination taxes, and the pandemic death rates. However, the post mentioning social media influencer Bruna Vieira is the 5th most liked (1,135), and the reference to Olivia Rodrigo is the 21st (248 likes). These numbers indicate that these framing devices might reach better engagement metrics if used more often or combined with other frame packages, for example, those referring to the “hard” facts and numbers of the crisis. The presence of “internet-native symbols” and social media influencers represents another framing device for misinformation prevention, as the presence of popular individuals can overcome the popularity shortcomings of institutions during crisis situations (Mirbabaie et al., 2020). The high engagement metrics of posts that emphasize numerical and statistical data indicate that these categories of informational evidence have high potential for preventing and combating misinformation targeting the public and not only highly involved individuals (Vafeiadis and Xiao, 2021).

5.2 Toward a frontier of political contextualization in misinformation research

Brazil has a historical tradition of successful vaccination campaigns. Nevertheless, the country's population vaccination rates are falling in parallel with increasing anti-vaccination movements inside and outside the country (Langlois, 2021). Yet, despite this pessimistic scenario, the Brazilian population is still very supportive of vaccines, which is visible in the good numbers of the COVID-19 vaccination campaign (Henrique, 2022). The main difficulty faced by the Brazilian people regarding this matter during the pandemic was the delay of Bolsonaro's government in buying vaccines, combined with his constant public exhibition of distrust of them (Ferrante et al., 2021). With regards to RQ2, the public health organizations analyzed in this paper found a very receptive public regarding their publications addressing the COVID-19 vaccination campaign. We verified this interest in the high numbers of retweets and likes in the “public policies,” “vaccination campaign information” and “vaccination coverage report” frame packages. The engagement metrics also confirm that Brazilian Twitter users rely on the public health organizations' profiles as sources of information about the pandemic status and know when and where they could get vaccinated. In turn, the public health organizations seemed to rely heavily on the Brazilian population's trust in vaccinations since they did not address the ubiquitous problem of misinformation very often at that time.

The frame package “debunking misinformation and fake news” is the least frequent in the analyzed sample and did not appear among the tweets with higher engagement metrics, although public health organizations had high levels of trustworthiness that usually helps to debunk misinformation (Allcott et al., 2019). The spread of misinformation is an increasing problem in Brazil (Amaral et al., 2022). Combined with the lack of a national communication strategy for the COVID-19 vaccination campaign in Brazil, the absence of a more assertive approach to debunking misinformation is a relevant strategic weakness if we consider the tactics used by fake news spreaders (Vafeiadis et al., 2019). Unlike the health organizations analyzed in this paper, these groups often build large networks of multiple profiles that share the same hashtags and texts (Ross et al., 2019).

Hence, we observed that in the Brazilian political context, frame packages that focus on communicating governmental public health measures and their results, such as “public policies” and “vaccination coverage report,” were also deployed by the federal government-related health organizations as strategic framings for attenuating President Bolsonaro's anti-vaccine public statements. By focusing their communication on divulgating the vaccination campaign's success, public health organizations conveyed that the federal government complied with the people's desires. Moreover, despite the president's public statements, the government-related health organizations' communication on social media made visible efforts to show Bolsonaro's supposed support of the vaccination campaign, attributing its success to his leadership. By contrast, public health organizations associated with the opposite political alignment used the divulgation of the vaccination campaigns' evolution to demonstrate their success in facing Bolsonaro's anti-vaccination positioning. In both cases, the framing device “numbers and statistics,” including the use of infographics, has been identified as crucial. This finding seems contradictory but also demonstrates that misinformation research must consider the broader local political context and its complexities instead of only focusing on the content of social media communication (Bisiada, 2022).

Our analysis of frame packages and framing devices emphasizes that the political disputes in Brazil shape public health organizations' communication on Twitter. This particularly applies to the appropriations and interpretations by opposite political groups, making it hard to classify these frame packages as aligned with one or another political group. Concerning future directions in misinformation research, this aspect emphasizes the need to contextualize information sources that stand in a partisan relationship to each other. Moreover, as this case showed, the political context in which public health organizations operate can have a large impact on how to approach misinformation combat.

5.3 Contributions to theory

The salient strategic element we identified in the Brazilian vaccination campaigns on social media was the use of visual framing techniques. From existing literature, we know that the importance of visually framing accurate information in contexts such as vaccination campaigns increases, given that misinformation is also being framed visually (Brennen et al., 2021). Concerning the denotative approach to visual framing in this case, numbers might be perceived as “hard” and objective facts. Therefore, the framing device of “infographics” can be considered an important element for misinformation prevention and combat. Numeric and statistical representations combined with an appealing visual language make it possible to combine these data with symbolic-semiotic elements and their associated meanings, such as the speed and efficiency metaphor in the “vacinometer” infographic (Rodriguez and Dimitrova, 2011). Whereas the denotative dimension is mostly addressed for showing the veracity of the facts and events mentioning or addressing specific fake news pieces, connotative elements such as community-shared references and memes are central for targeting highly engaged audiences, positioning a public health organization as an insider in Brazilian digital culture. The use of “infographics” constitutes a symbolic-semiotic strategy for the visual representation of statistics and numerical data. Considering the ideological dimension of visual framing, the images shared by the analyzed institutions focused on showing accessibility and efficiency while managing the vaccination campaign.

Further, the framing devices “pop culture references” and “internet-native symbolism” show that there is an alternative to the “war” framing previously introduced by Wicke and Bolognese (2020). The analyzed public health organizations focused their publications on the framing categories I and IV (“Communications and Reporting” and “Reacting to the Pandemic”). Largely avoiding categories II and III (“Community and Social Compassion” and “Politics”) makes sense considering the complex Brazilian political context and the dilemma public health organizations found themselves in. The “Community and Social Compassion” frame has also not been deployed for addressing misinformation, pointing to the severe Brazilian social division in this context.

Even though some “pop culture references” were previous research, our categorization enables a better understanding of their intersection with fandom communities and internet culture. We encourage future research to dive deeper into the entanglements of organizational and fan communication studies as we also observed a lack of unified strategies around the use of hashtags, for instance. The findings of this study help to develop conceptual dimensions of misinformation combat that extend existing theory. In Figure 2, we introduce a conceptual framework of visual framing for misinformation combat. It includes the key findings of this study and integrates them with existing literature on visual framing and misinformation literature (Vafeiadis and Xiao, 2021; Rodriguez and Dimitrova, 2011). The concepts that were developed in this study are highlighted in gray.

5.4 Practical implications

Our findings show a broad repertoire of elements that can be deployed by stakeholders of misinformation combat such as public health organizations. We support a view that organizations who have limited authority in certain social settings can mitigate this problem by employing authoritative proxies on their behalf. For example, a public health institution which has no authority in a certain fandom community can make up for this through a collaboration with an influential individual or organization from that community. This strategy can also be extended by the employment of local celebrities or cultural products that target specific communities.

Considering the work on misinformation prevention by Lewandowsky and van der Linden (2021), our findings show that framing health information with internet culture can serve the purpose of shifting the playing field from political narratives to an alternative framing or “encoding” of the information. This allows organizations to detach information from political ideologies and make it accessible to internet-native audiences. For example, the politically laden topic of vaccination can be communicated more effectively to certain communities if reframed in a non-political narrative. As the case of Brazilian health organizations shows, symbols and narratives coming from internet culture can help organizations to better address sensitive issues for the benefit of the people.

Finally, we suggest that organizations with a stake in misinformation combat approach their actions in alignment with the phases of misinformation combat. Most importantly, organizations should be aware that measures can be taken to prevent misinformation before they spread. This requires a proactive engagement with topics and platforms which constitute the breeding ground from misinformation. The framework provided in Figure 2 provides an overview of phases and objectives that characterize the pursuit of strategic misinformation combat.

6.1 Summary

In this paper, we analyzed a sample of Twitter posts published by Brazilian public health organizations in the context of the COVID-19 vaccination campaign in 2021. Specifically, we applied qualitative and quantitative frame analysis techniques to identify the most frequently used frame packages and framing devices in public health communication in Brazil. A total of 14 frame packages characterize the strategic activities carried out by the investigated organizations and their representatives. Alongside already known framing devices, the investigation of this case unearthed novel framing devices that relate to Brazilian internet culture, including “infographics,” “pop culture references” and “internet-native symbolism.” Moreover, this case shows that the political context shapes the choice of framing techniques for misinformation combat. The dilemma in which many Brazilian public health organizations found themselves (e.g. governmental vs. oppositional positions toward vaccination) affected their ability to openly address misinformation. Finally, the analysis of this case emphasizes that research on the proactive phase of misinformation prevention deserves more attention as it is currently given.

6.2 Theoretical and practical contributions

This study contributes to existing knowledge in a threefold manner. First, the identification of visual framing devices relating to internet culture adds to our understanding of hitherto little addressed organizational framing. The case of Brazilian public health organizations provides a novel perspective on knowledge by offering a notion of cultural peculiarities (e.g. internet-native symbolism, pop culture references) for misinformation prevention and combat. Based on these findings, we introduce a conceptual framework that integrates theoretical dimensions of visual framing, framing devices that are borrowed from internet culture, and the distinct phases of misinformation combat (prevention, detection, debunking). Second, this study emphasizes the need for a political frontier of misinformation research that does not relate to the partisanship of the spreaders but that relates to the political context of the analysis (e.g. dilemmas of public organizations with a commitment to provide accurate information to citizens). Third, the findings of this study inform decision-makers and public health organizations about framing devices that are alternatives to prevalent communication practices and can guide strategies to promote accurate information during crises and/or prevent misinformation.

6.3 Limitations and further research

This study has limitations as the chosen manual coding approach restricts the ability to draw much larger subsamples. The results of this case study do not allow statistical generalization to other countries or cultural contexts. If readers wish to generalize from this case, we caution them to make their own judgment calls about the risks of doing so. Consequently, we invite scholars to form inter-disciplinary teams that cover expertise in both computational data analysis and cultural contextualization of misinformation phenomena. This will allow research teams to draw and analyze samples from a more diverse range of countries and to make theoretical contributions that are not confined by generalization to organizations from the Global North. Moreover, although very prominent in Brazil, Twitter does not cover the Brazilian social media and internet sphere in its entirety. Certain societal or political groups might be underrepresented and some overrepresented in the sample.

We suggest further research to put more emphasis on exploring the possibilities of misinformation prevention, for example, how organizations can strategically use prebunking mechanisms to sensitize social media users for misinformation. Furthermore, we encourage researchers to further explore the use of internet-native symbolism and (local) internet culture for misinformation prevention in future studies. Here, investigating how framing information with internet culture affects user engagement and how it can reduce the susceptibility to misinformation would be a valuable research endeavor.

In March of 2023, the newly elected government in Brazil launched an official campaign to fight misinformation (“Brasil Contra Fake”). We invite all misinformation scholars to take Brazil as an example and continue their efforts to support this fight.