Misinformation About COVID-19 Vaccines on Social Media: Rapid Review

Introduction

Background

An unprecedented global effort has been undertaken to develop vaccines that protect against COVID-19. However, there is a grave concern that vaccine hesitancy will be a major obstacle to reaching herd immunity. In 2019, the World Health Organization (WHO) had already named vaccine hesitancy as 1 of 10 threats to global health [1]. Global vaccine distribution equity is also a major challenge. Figures from February 2022 show that 61.9% of the world’s population has received at least one dose of a COVID-19 vaccine, but only 10.6% of people in the Global South have received a dose [2]. Furthermore, the rate of people receiving a COVID-19 vaccine in some high-income countries where vaccines are available and free has dropped [3]. The WHO reiterates that COVID-19 vaccines remain critical and are considered effective against severe disease and death [4].

The reasons behind COVID-19 vaccine hesitancy are complex. Fear of side effects and concerns about the pace at which the vaccines were developed have been cited as primary reasons behind this hesitancy [5]. In addition, misinformation about COVID-19 and vaccines has spread on social media platforms at a rate that has made the WHO coin the phrase infodemic [6]. An infodemic is “too much information including false or misleading information in digital and physical environments during a disease outbreak” [7,8].

A well-known false claim is that the measles, mumps, and rubella (MMR) vaccine can cause autism [9]. The claim has since been empirically refuted many times but is still stated as a major concern for some parents [10]. Motta and Steccula [11] examined American public opinion data on MMR safety collected before and after a retracted 1998 study linking autism to MMR. The researchers detected a statistically significant increase in public concern about MMR safety following the retracted study and the media attention it received. This suggests that misleading vaccine information can impact public confidence in vaccines and cause skepticism about vaccines in general. Since the retracted 1998 study, groups of vaccine deniers or antivaxxers have grown, and claims that vaccines are harmful have spread to almost all vaccines [12]. Pullan and Dey [13] analyzed search patterns in Google Trends during the early stage of the pandemic in 2020 and found that search interest in COVID-19 vaccines had understandably increased, but also found that well-known antivaccine searches such as “autism” and “mercury” also had a growing presence and similar spikes as search patterns for COVID-19 vaccines. These results confirm that the false claim of associations between autism and MMR vaccines has become an argument for all types of vaccines and also possibly a concern when it comes to COVID-19 vaccines. Therefore, we examined whether misinformation on social media in recent times linked autism to COVID-19 vaccines.

Furthermore, vaccine hesitancy based on misinformation seems to be a worldwide phenomenon regardless of the uneven distribution of COVID-19 vaccines [4]. Social media plays a crucial role in disseminating both correct information and misinformation about infectious diseases and vaccines [14]. Wilson and Wiysonge [15] showed, in a global cross-national analysis of geographically coded tweets and vaccination rates from 166 countries, that there was a significant relationship between social media use and vaccine hesitancy. However, there has been a joint effort by several of the largest social media platforms and technology companies to combat the spread of misinformation about COVID-19 [16].

Objective

We aimed to synthesize the existing research on misinformation about COVID-19 vaccines spread on social media platforms and its effects. The secondary aim was to gain insight and gather knowledge about whether misinformation about autism and COVID-19 vaccines is being spread on social media platforms. The following questions guided our inquiry: What is known about misinformation regarding COVID-19 vaccines spread on social media platforms? What is known about the effects of misinformation about COVID-19 vaccines spread on social media platforms? What is known about social media misinformation on COVID-19 vaccines concerning autism spectrum disorder?

Methods

Design

We followed the guidance from Cochrane Rapid Reviews [17]. We chose a rapid review protocol in line with the recommendations by Cochrane; that is, the need “for timely evidence for decision-making purposes including to address urgent and emergent health issues and questions deemed to be of high priority” [17]. The need to address vaccine hesitancy toward COVID-19 vaccines is an emergent health issue. The narrative qualitative synthesis was undertaken with the guidance of the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) 2020 Statement [18] and the Synthesis Without Meta-analysis reporting guideline [19].

Search Strategy and Selection Criteria

With the help and expertise of an information retrieval specialist, we used the following search string in this rapid review: (“misinformation” OR “disinformation” OR “information”) AND (“social media” OR “Facebook” OR “Twitter” OR “Instagram” OR “WhatsApp” OR “Telegram” OR “Tumblr” OR “Pinterest” OR “YouTube” OR “VKontakte” OR “Snapchat” OR “TikTok” OR “Weibo” OR “WeChat” OR “Reddit”) AND (“covid*” OR “corona*” OR “pandemic” OR “Sars-CoV-2” AND “vaccine*” OR “vaccination*”).

No date or language limitations were used. The full search strategy of the information retrieval specialist is available in Multimedia Appendix 1.

Publications were excluded if the studies were not original empirical research, if studies examined vaccines in general and not COVID-19 vaccines, if studies did not examine social media misinformation, and if data were gathered before the COVID-19 vaccine Pfizer-BioNTech phase 3 clinical trial [20].

Data Collection Process and Extraction

This review was registered with the PROSPERO international register of systematic reviews (CRD42021277524). Systematic searches in the PubMed, PsycINFO, ERIC, EMBASE, Cochrane Library, and Cochrane COVID-19 Study Register databases were conducted by an information retrieval specialist on September 9, 2021. Duplicates were identified and removed by IS and EG. We used Rayyan [21] as the screening tool. Rayyan is a web application and mobile app for systematic reviews. It eases the process of the initial screening of abstracts and titles and helps researchers save time when they share and compare include-exclude decisions. All titles and abstracts were screened by IS and ANH independently. In the initial search, no date restriction was set. However, during the piloting of the title and abstract screening, IS and ANH discussed the fact that there were studies that explored misinformation about COVID-19 vaccines at a very early stage in the pandemic, before any COVID-19 vaccines were a reality. We decided that we needed a threshold date as to when we believed we found misinformation about the actual COVID-19 vaccines to be relevant, as misinformation at a very early stage would be about a potential vaccine. Therefore, we decided to include studies that were conducted during and after the Pfizer-BioNTech phase 3 clinical trial, because then the news about an actual vaccine was starting to spread around the world and thus starting to become a reality. We chose the Pfizer-BioNTech vaccine because it was the first COVID-19 vaccine to be approved by the WHO [22].

Of the 319 titles and abstracts screened, IS and ANH disagreed on 35. The disagreements were resolved through discussions between the 2 reviewers and if an extra opinion was needed, EG was consulted. Of these articles, 1 article was in German, 2 were in Spanish, and the rest were in English. IS can understand German and EG speaks Spanish. IS and EG performed a further assessment of the eligibility of the full-text records and conducted a pilot exercise using the same 10 full-text articles to calibrate and test the review form. After the screening, both reviewers assessed the articles that the other had excluded. ANH assisted with conflicts and discussed doubts surrounding the included or excluded articles. The data extraction from the included articles involved 2 reviewers (IS and EG), where IS extracted data using a piloted form and EG checked for the correctness and completeness of the extracted data. Data from the included articles were extracted based on design and study population, type of misinformation, effect of misinformation, misinformation about autism, ethical considerations, and social media channels. The agreed evidence was then synthesized narratively.

To synthesize the knowledge gathered about the types of misinformation, a thematic analysis was performed [23]. After the data extraction, IS gathered the data on the content of the misinformation. The data extracts on misinformation were then coded by ANH. IS and ANH searched for themes based on the codes and agreed upon 3 final themes of misinformation: conspiracies, medical misinformation, and vaccine development. EG approved the themes. Multimedia Appendix 2 provides an overview of the thematic analysis that was undertaken.

Assessment of Risk of Bias

The risk of bias was graded according to the Joanna Briggs Institute (JBI) Critical Appraisal tool “Checklist for Analytical Cross-sectional Studies” [24] by 1 experienced reviewer (DSQ). The evaluation was based on answers to 8 questions (yes, no, or not applicable). The studies were classified as having low (>70%), moderate (40%-70%), or high (<40%) risk of bias. A complete overview of the assessment can be found in Multimedia Appendix 3 [25-69].

Assessment of the Quality of the Evidence

One experienced reviewer (RW) assigned certainty of evidence ratings based on recommendations by the Grading of Recommendations Assessment, Development and Evaluation (GRADE) working group [70]. The included studies that looked at associations were given a narrative GRADE score related to the outcome “Association between social media misinformation and vaccine hesitancy.” The level of quality of evidence was classified as very low, low, moderate, or high. A complete overview of the assessment can be found in Multimedia Appendix 4 [25-69].

Data Synthesis

Narrative synthesis was undertaken with the guidance of the PRISMA 2020 Statement [18] and Synthesis Without Meta-analysis reporting guideline [19]. In the synthesis, findings from our included studies were grouped according to study design, population, social media sample, types of social media, types of misinformation reported, misinformation about autism, the reported effect of the misinformation on vaccine hesitancy, and the assessments of risk of bias and quality of evidence. When synthesizing the findings narratively, studies with a low risk of bias or high quality of evidence will be highlighted on several occasions.

Results

Study Selection and Risk of Bias

We identified 45 relevant studies (Figure 1). The list of excluded articles during the full-text review and the reasons for exclusion are reported in Multimedia Appendix 5. The risk of bias in 53% (24/45) of the included studies was classified as low, according to the JBI Critical Appraisal tool [25-42,47-52]. In total, 18% (8/45) of the studies showed a moderate risk of bias [43-46,53-56]. Finally, 27% (12/45) of the included studies showed a high risk of bias [57-68]. Of the 45 studies, in 1 (2%) study [69], none of the questions in the JBI tool were applicable.

We grouped the studies into 2 major categories according to data sampling. One group gathered data through surveys, interviews, or focus groups (Table 1). The other group gathered data from social media platforms (Table 2). The largest total population sample in the first group of 22 studies (Table 1) came from Europe, with 27,975 respondents in total. All respondents were described as adults or >18 years, except for 2 studies in the United States where the participants were aged ≥65 years [28,34]. Another exception was 1 study from Slovenia, where participants aged ≥15 years were included [35]. Gender has not been a focal point in any of the 45 included studies.

Data were extracted from social media platforms in 23 of the included studies. These studies formed the second group (Table 2).

The 12 studies that were assessed to have a high risk of bias were found in the second group (Table 2), whereas the studies in Table 1 had a low or moderate risk of bias according to the JBI tool.

Many of the studies did not name social media platforms in the first group (Table 1) but rather discussed social media platforms in general. However, some studies did specify which social media platforms they were assessing. Figure 2 summarizes the types of social media platforms specified in the 45 included studies.

‎View this figure

‎View this figure

Thematic Analysis

Figure 3 provides an overview of the 3 overarching themes of misinformation identified from the thematic analysis, and examples from the data extraction and codes that laid ground for the final 3 themes are seen in the inner circles of the figure.

Of the 45 included studies, 18 (40%) studies reported misinformation across all 3 categories [27-29,35-37, 45,48,49,51,55,57, 58,60-62,67,68], 9 (20%) studies reported only on conspiracies [26,30,33,42,44,46,53,54,56], 6 (13%) studies were concerned specifically with medical misinformation [25,34,38,43,63,64], and 12 (27%) studies reported on COVID-19 vaccine misinformation or antivaccine discourse without going into further detail [31,35,39-41,47,52,59,65,66,69].

‎View this figure

Effects of Social Media Misinformation

We identified 19 studies that made assumptions regarding the effects of social media misinformation on vaccine hesitancy (Table 3). The evaluation of the certainty of evidence of these 19 studies that measured the “Association between social media misinformation and COVID-19 vaccine hesitancy” was classified as moderate or low to moderate according to GRADE in 2 cases [38,65]. For the rest of the studies, the certainty of evidence according to GRADE was considered low or very low.

The 2 studies with a higher certainty of evidence had an experimental design. Rotolo et al [65] aimed to develop and distribute infographics that addressed COVID-19 vaccine hesitancy and misinformation. Although their infographics reached thousands of people, they were unable to determine the impact on vaccine hesitancy. Sharevski and Gover [38] analyzed the perceived accuracy of COVID-19 vaccine–related tweets when they were moderated by smart device technology that Twitter applies to COVID-19 misinformation. The results from the 304 participants suggested that vaccine-hesitant users ignored warnings as long as the tweets aligned with their personal beliefs.

Allington et al [26] analyzed findings from a web-based survey conducted with a sample of 4343 adults in the United Kingdom. They found a positive correlation between trust in social media and vaccine hesitancy and the strongest link was found for YouTube and Facebook. Conspiracy suspicions about COVID-19 and general vaccine attitudes appeared to be uniquely predictive, jointly explaining 35% of the variance. Boucher et al [60] analyzed 636,516 English and French tweets. A total of 23.4% (n=146,191) of the conversations on Twitter during the study period could be directly attributed to vaccine hesitancy. A British study by Liu and Liu [30] of 5114 adults found that social media dependence and high levels of conspiracy mentality were most likely to be associated with web-based discouragement of vaccination. In a study of 4571 Norwegian adults, individuals who preferred social media platforms had nearly 2-fold (ie, 1.64) odds of being hesitant toward COVID-19 vaccination compared with those preferring source-verified media platforms [31]. In addition, those who held the belief of the superiority of natural immunity over vaccination were more vaccine hesitant (odds ratio 2.663, 95% CI 2.350-3.028; P<.001). Petravić et al [35] asked 12,042 Slovenian residents about their attitudes toward COVID-19 vaccines. Those who trusted alternative media sources and alternative explanations on social media were more vaccine hesitant. A total of 11 studies [27-29,33,34,36,39-43] discussed social media misinformation, vaccine uptake, and vaccine intentions.

Discussion

Principal Findings

The 45 included studies about misinformation on social media platforms about COVID-19 vaccines suggest that there should be great concern about the volume of misinformation being spread, and the association between COVID-19 vaccine misinformation and vaccine hesitancy. To our knowledge, this is the first review to analyze social media misinformation about COVID-19 vaccines. We identified 3 overall categories of misinformation, namely, medical misinformation, conspiracies, and distrust in vaccine development; however, the 3 categories are connected and sometimes overlapping, as distrust in vaccine development might be founded in conspiratorial beliefs about hidden power structures and corrupt elites. The included studies were predominantly from Europe and the United States, and therefore, there is a lack of information, especially from African and South American countries. Twitter was the most studied platform, with Facebook and YouTube being in the second and third place, respectively.

Fear of side effects is a major concern when it comes to vaccine hesitancy, and as this review shows, this concern can easily turn into medical misinformation and exaggerations of side effects. To synthesize what is known about social media misinformation about COVID-19 vaccines from the included studies, a thematic analysis was undertaken. The coded extract of data that made up the theme medical misinformation contained misinformation about side effects such as infertility, chronic illness, changes in DNA, physical deformities, and mental illness. Only one study mentioned autism as an adverse side effect of COVID-19 vaccines [63]. Knowing that the side effects of the vaccines are a major concern [5], medical misinformation has the potential to do a lot of harm.

When we examined the types of reported misinformation, we also found that a lot of misinformation is grounded in conspiracy theories. Some of these conspiracy theories have become infamous, such as the belief that there are secret societies and hidden power structures run by corrupt elites. These elites are believed to be networking with big pharmaceutical companies to make money or to depopulate the world. There are also conspiracy theories about racially motivated depopulation. For example, we found 3 studies from the United States that mentioned the fear of racist motives by official health authorities as a reason for vaccine hesitancy [45,61,67]. Some of this fear has historical roots in the United States, as one of these studies [67], for instance, brought up the Tuskegee Syphilis study. This was a clinical study (1932-1972) in which the United States Public Health Service used African Americans to observe untreated syphilis and therefore denied them treatment [71]. This exemplifies that a lack of trust in public health institutions might have deep historical roots in some countries and cultures. Other issues to be aware of are religious concerns and vaccine hesitancy. We found several studies that reported on misinformation about the content and development of vaccines and in some studies [29,45,55], we found very explicit language (eg, “pigs” and “cells from aborted children”). Such wording can cause worry in some religious communities.

The second objective of this review was to examine the effects of social media misinformation about COVID-19 vaccines. The 19 studies identified in Table 3 interpreted the results as associations among social media use, misinformation, and vaccine hesitancy. According to the JBI and GRADE evaluations, there is a need for more robust designs to become more certain regarding the actual effect of social media misinformation on vaccine hesitancy. Only 1 study, an intervention study regarding the impact of addressing misinformation on Twitter users, was assessed to have a low risk of bias and moderate quality of evidence [38]. In addition, 4 studies reported significance levels of associations, but the effect size was not reported [31,32,35,50]. Other studies in this review showed that social media platforms did not necessarily spread misinformation to a great extent, perhaps reflecting that the effort made by some social media platforms to halt misinformation has worked. Chan et al [47] examined 48 COVID-19 vaccine–related videos on YouTube in December 2020 and found only 2 videos (4.2%) that made nonfactual claims. Hernández-García et al [54] also examined YouTube videos during February 2021 and found that only 2 out of 110 videos contained COVID-19 vaccine hoaxes or conspiracy theories. Pascual-Ferrá et al [64] examined social media data from Facebook, Instagram, Reddit, and YouTube and did not find evidence of the dominance of misinformation. However, what is being spread and discussed in closed groups is another question that needs to be examined further. Another valid approach would also be to examine comment sections. Although antivaccine content has been prevented from surfacing in searches, this does not prevent people from commenting about their beliefs or posting other types of information in the comment section. It is controversial to deplatform people [72] and might even do harm, as these people might be seen as someone speaking against the establishment, which are, in essence, some of the core beliefs of some conspiracy theorists.

Surprisingly, there was a dearth of studies examining misinformation about autism and COVID-19 vaccines. Considering the history of misinformation about vaccines and autism over the past 2 decades, more research should focus on this topic. One could also speculate whether this would have played out differently if COVID-19 vaccines were more targeted toward younger children. Future research should also aim to examine social media platforms such as TikTok, which is a very popular platform worldwide, and is often used by people who are younger than, for instance, the average Twitter user [73]. The low inclusion of some social media platforms such as TikTok or Telegram is a limitation, as certain parts of the population and particular communities are not included.

When addressing vaccine hesitancy, one should be careful before labeling all vaccine-hesitant people as antivaxxers or misinformed people. The primary concerns from people who say that they are vaccine hesitant are the safety of the vaccines and the rapid pace of their development [7]. However, being hesitant and skeptical does not mean that these people are unwilling to take the vaccines but rather that they have some concerns that should be adequately addressed to convince them of the safety and efficacy of the vaccines. To understand a complex issue such as vaccine hesitancy, knowledge about sociodemographic conditions and cultural awareness is key. In addition, countries with a more undemocratic regime will suffer from a lack of trust in official authorities, which may damage an official vaccine campaign. People’s trust in the government varies between countries and cultures. Although some countries have a tradition for mandatory vaccination, this is less acceptable in other countries.

The issue of trust is also an important issue to be considered. “Fake news” became a buzzword in the last decade and the term was used not only to actually coin false news but also to spread distrust to news agencies and official actors, accusing them of spreading falsehoods. Vosoughi et al [74] aimed to understand how false news spread and examined a set of rumors (n=126,000) spread by 3 million people on Twitter from 2006 to 2017. The results showed that false news spread much faster and reached a larger audience than real news. Social media has contributed to a far more complex information landscape than before and has created new challenges when it comes to building trust in official actors. These are issues that need to be addressed and analyzed in future studies of misinformation about vaccines.

Limitations

We did not include gray literature or preprints in this review. The rapid pace at which the pandemic is moving makes preprint research particularly relevant. However, although peer review is not a guarantee of quality, we decided not to include gray literature or preprints and limited eligible articles to peer-reviewed manuscripts. We did not contact researchers with potential projects on this subject matter. Furthermore, the searched databases were selected based on the topic at hand. There will always be a chance that other, more specified or general databases would capture other studies.

A limitation of the evidence included in this review is that, in our assessment, all but one received a low score on the assessment of quality of evidence. However, it is a challenge in the process of assessing quality of evidence and risk of bias, that the included studies have a range of different designs, each with its strengths and weaknesses. We applied 2 tools in this regard: 1 from the JBI [24] and 1 from the GRADE Working Group [70]. Neither tool provides a complete picture of the included studies, but they may help the reader in obtaining a broader view of the included studies.

Furthermore, there is a poor correlation between self-reported social media use and actual use [75]. A high proportion of these studies extracted data from Twitter because Twitter has opened up access for researchers to extract data from its platform, making it more accessible compared with other social media platforms. The Twitter sample may not be representative of a random sample of the population, as its users tend to range in age from 25 to 34 years and are predominantly from the United States [76,77]. Furthermore, we did not assess the potential presence of social media bots (automated accounts) spreading incorrect information in these studies. We also did not discuss how social media algorithms partake in creating echo chambers [78]. These are well-known challenges in researching data gathered from social media [79]. Only 1 study included in this review was from an African country [32]. The study was from Uganda and included 600 participants. There were no studies from Middle or South American countries or Pacific Island countries and Australia. The studies included in this review focused mostly on high-income countries, thus making the conclusions and generalizations weaker in terms of applying them to Global South nations.

Conclusions

This review suggests that there should be great concern about the volume of misinformation being spread and the association between COVID-19 vaccine misinformation and vaccine hesitancy. Many studies have shown that there is a link between misinformation on social media and COVID-19 vaccine hesitancy. However, there is a need to examine this effect using a more robust experimental design to assess this effect. It is possible to conduct more experimental studies in an ethical manner in a laboratory setting; for instance, a study to see whether people are able to distinguish between false and true information and how they do so. Such a study would, of course, have to be based on informed consent and be approved by an ethics committee. It is also possible to improve observational studies that extract data from social media by gathering more representative data (eg, including data from several social media platforms, different audiences, several languages, and covering longer periods). There are many types of misinformation that are spread on social media platforms, and to prevent these myths from taking hold, health authorities should openly address and discuss these false claims with both cultural and religious awareness in mind. This review showed that a greater variation in studies is needed when it comes to both social media platforms and geographic location. We only found one study that mentioned misinformation about autism and COVID-19 vaccines, but taking the history of autism and the antivaxx community into account, we believe that this an issue that should be given attention in future research.

Although some major tech companies have taken steps to prevent misinformation, more action is needed to stop this infodemic. One valid approach proposed for infodemic management is first information monitoring (infoveillance); second, to enhance and build eHealth literacy and science literacy capacity; third, to encourage quality improvement processes such as fact-checking and peer review; and finally, to encourage accurate and timely knowledge translation [80].

Misinformation about COVID-19 vaccines is still thriving on social media platforms. However, this undertaking represents a balance between people’s right to speak their minds and strategies to counter the spread of misinformation.