COVID-19-related misinformation on social media: a systematic review

1The coronavirus disease 2019 (COVID-19) pandemic is spreading around the world with an increasing number of people becoming infected. Naturally the demand for information is high and people want to share news about the pandemic and their experiences. Social media have occupied a central role during the ongoing pandemic and the resulting wave of content related to COVID-19 has been referred to as an infodemic. However, incorrect information about COVID-19 can be dangerous because it may divert people away from taking appropriate actions that would help protect their health and the health of others and could lead them to take actions that may spread the illness or to engage in other problematic behaviours. The World Health Organization (WHO) has already recognized the importance of COVID-19-related misinformation and is participating in an awareness campaign aimed at encouraging people to check information with trusted sources.

2A distinction has been made between misinformation, defined as incorrect or false information that is shared without the intent to harm, and disinformation, defined as incorrect or false information that is shared with the aim of causing harm. However, making this distinction involves assessing the intent of the person spreading the information, which may be problematic., Consequently, in this review we use misinformation as a general term for incorrect or false information, regardless of intent.

3Our review focuses on misinformation that appeared early in the pandemic. During this phase, little was known about the virus, such as how it spread or how infected people could be treated most effectively. There was a shortage of protective equipment in many countries, no vaccines had been developed and it was uncertain how fast an effective vaccine could actually be produced. We believe this high degree of uncertainty during the initial phase may have been conducive to the appearance of a substantial amount of misinformation on social media. A synthesis of the evidence on COVID‐19-related misinformation on social media is needed to provide guidance for the health-care sector and to help in the assessment of guidelines for social media.

4We carried out a review of publications on COVID-19-related misinformation on social media that appeared during the first phase of the pandemic. The review followed guidelines detailed in the preferred reporting items for systematic reviews and meta-analyses and in a measurement tool to assess systematic reviews., The review is registered with the PROSPERO international prospective register of systematic reviews (CRD42020182154).

5We searched the PubMed®, Scopus, Embase®, PsycInfo and Google Scholar databases on 5 May 2020 and 1 June 2020 for articles that included keywords related to social media and COVID-19. The full search strategy is detailed in Table 1 (available at: http://www.who.int/bulletin/volumes/99/6/20-276782). Articles were included in the review if they: (i) focused on COVID-19 and social media; (ii) considered misinformation; and (iii) were primary studies that reported findings. Articles that did not meet these criteria or were in a preprint version were excluded.

6References identified were uploaded to EndNote X9 (Clarivate Analytics, Philadelphia, United States of America) and duplicates were removed. Two authors examined the articles’ titles and abstracts, respectively, to assess their eligibility for inclusion in the study. In a second assessment round, the full texts of the articles selected in the first round were carefully analysed to confirm their eligibility by two independent reviewers. Doubts about eligibility were discussed with the third author until agreement was reached. Finally, the selected articles were divided among the three authors for data extraction and data were abstracted onto a specially standardized spreadsheet. The quality of the evidence in, and the risk of bias of, the articles were classified by two authors using the grading of recommendations assessment, development and evaluation (GRADE) approach.

7In total, we identified 1351 publications. After removing duplicates, we screened 825 titles and abstracts for eligibility and 22 articles finally met the inclusion criteria (Fig. 1).,,– A list of articles whose full text was examined but which were excluded from the review is available in the data repository.

8Table 2 shows the main characteristics of the 22 studies included in the review; details of submission dates, funding and conflict of interests are available from the data repository. Fourteen of the 22 studies were cross-sectional and based on data extracted from social media, whereas eight were based on surveys or focus groups or both. Thirteen studies involved a single social media platform: Twitter (nine studies), Facebook (two studies), WhatsApp (one study) and YouTube (one study); the remaining nine studies involved several social media platforms. The monitoring period of social media ranged from 1 to 123 days. Two studies did not specify the monitoring period. According to GRADE evaluation criteria, 11 of the 22 studies were awarded 1 point, and the remaining 11 were awarded 2 points, which means they were all of low quality. However, the low quality was principally due to the studies being observational, whereas randomized trials, in contrast, provide the highest quality of evidence.

9Six of the 22 studies reported the proportion of social media posts that contained misinformation on COVID-19, including false information and jokes. Four of these six studies reported the proportion on Twitter only, one reported the proportion on Twitter and Weibo and one reported the proportion on Facebook (Table 3). The proportion of misinformation ranged from 0.2% (413/212 846) to 28.8% (194/673) of posts.

10Eleven studies did not categorize the specific type of COVID-19-related misinformation, nine described specific misinformation myths and two categorized the misinformation as sarcasm or humour related to COVID-19 (Table 2). Sarcasm and humour can draw on hyperbole or false claims to make a point but typically the intent is not to misinform. However, if the person receiving the message does not understand it is a joke or sarcasm, they may end up being misinformed.

11Only four studies examined the effects of misinformation: all reported that it led to fear or panic (Table 2). One of the four mentioned that misallocation of resources and stress experienced by medical workers were also possible consequences of misinformation. Another study found that 46.8% (525/1122) of survey respondents were tired of COVID-19 being the main theme across all media.

12Sixteen of the 22 studies proposed one or several ways of tackling COVID-19-related misinformation. The most popular measure, mentioned in eight studies, was promoting and disseminating trustworthy information.,,,,,– Seven studies suggested addressing, containing or debunking misinformation:,,,,,, misinformation could be replaced by facts and accurate information, or health authorities could debunk myths and help answer people’s queries. Four studies mentioned increasing the health literacy of social media users:,,, they highlighted the need to educate social media users on how to determine what information is reliable and to encourage them to assume personal responsibility for not circulating false information. Three studies proposed that social media should be supervised by an authority or government:,, misinformation could be addressed by the government providing more comprehensive reports on the current epidemiological situation. Three studies suggested introducing policies or regulations for social media,,, and two mentioned the need for more research., Six studies did not suggest any solutions.,,,,,

13Studies done during the first phase of the COVID-19 pandemic found that between 0.2% and 28.8% of social media posts about COVID-19 could be classified as misinformation. The large variability observed in the proportion may have been due to differences in social media samples, methods or the definition of misinformation. Studies on social media carried out during previous pandemics also reported a large variation in the proportion of posts identified as misinformation: 4.5% of posts on Twitter about H1N1 influenza, compared with 23.8% of content posted on YouTube about Zika virus disease, and 55.5% of posts on Twitter about Ebola virus disease.

14The studies identified several COVID-19-related myths that were spread through social media but provided no clear evidence of the effects of this misinformation. However, a few studies reported that misinformation led to fear and panic and to people becoming tired of hearing about COVID-19. There is evidence that misinformation can evoke negative emotions, which could, in turn, further contribute to its spread.

15Although misinformation is not a new phenomenon, today it can spread rapidly on social media and potentially reach more than half the world’s population. The studies in our review proposed six main ways of tackling COVID-19-related misinformation: (i) disseminating trustworthy information; (ii) addressing, containing or debunking misinformation; (iii) increasing social media users’ health literacy; (iv) officially supervising media in general; (v) introducing policies and regulations for social media; and (vi) increasing research on the topic. These suggestions have been included in published proposals for managing infodemics.,

16Recently, WHO launched social media chatbots in Rakuten Viber and WhatsApp to provide accurate information about COVID-19., Several studies confirm that health professionals and public health authorities could assist by debunking misinformation and providing true information.,,,,,, Correspondingly, WHO has created a specific webpage for correcting misinformation about the disease. However, although messages that debunk misinformation on social media may have the desired effect, it has been observed that such messages can also contribute to the persistence of misinformation.

17As social media users can easily lose track of what information can be trusted, teaching users how to identify reliable information is important.,,, One way of educating users about what information is trustworthy is to mark misleading posts as such. In addition, nudging (i.e. prompting or encouraging) people to think about the accuracy of a social media post has also been proposed. However, some form of self-regulatory behaviour may already exist on social media, whereby a collective intelligence acts to identify and stop misinformation by not forwarding it to others.

18Officially supervising media in general and introducing regulations for social media are sensitive topics because both measures can conflict with freedom of the press and the principle of free speech. Nevertheless, they may be considered during a pandemic. In fact, several large social media companies have introduced policies on controlling false or manipulated information.–

19A good strategy for tackling COVID-19-related misinformation could employ several or all of these proposed measures along with any new approaches that might appear. In addition, as misinformation appears to spread faster on some social media than on others, platform-specific strategies could be developed. However, further research is needed., Investigations into the effectiveness of different approaches to countering misinformation will provide valuable knowledge that could help governments fight misinformation in future pandemics or health emergencies.

20Evidence on the proportion of COVID-19-related misinformation on different social media platforms is insufficient. Moreover, little is known about the relative importance of the different reasons why people propagate misinformation. Accounts that are not verified by social media platforms as authentic seem to spread more misinformation than verified accounts., However, we do not know if social media users respond differently to these different types of account. Nor do we have much understanding about the impact of misinformation spread by bots.

21Studies that examine the longitudinal development of misinformation and the effect of that development is needed. Although information posted on social media may encourage specific behaviours, it is difficult to attribute people’s actions solely to social media postings because other factors may have an equal or even more important influence on determining when people decide to act.

22In addition, the most effective strategies for tackling COVID-19-related misinformation are currently not known. Although there are many ongoing attempts to correct misinformation, we were unable to identify any study that examined the effects of these attempts, such as whether they enabled people to be better informed or helped them feel safer.

23The study has several limitations. Our review considered only peer-reviewed articles that were published during the first few months of the COVID-19 pandemic. We did not explore the grey literature and we excluded a considerable number of non-peer-reviewed preprints. We also excluded one article that met the inclusion criteria because the full text was not available and it was not possible to obtain a copy from the authors. The low quality of the articles included in the study is an important limitation. Moreover, there was a high risk of bias because data were collected over a short time period and because several studies used only part of the collected data in their analyses. We could not conduct a meta-analysis, because of the small number of studies that reported the effects of COVID-19-related misinformation and the nature of these studies.

24In conclusion, our review found that COVID-19-related misinformation on social media is an important issue, both in terms of the amount of misinformation in circulation and the consequences for people’s behaviour and health. Despite rapidly growing scientific interest in the topic of misinformation, few studies have examined the scope of the problem, including why misinformation is spread, its impact and how best to tackle it. The impact of COVID-19-related misinformation could be reduced by: (i) social media users, who should avoid spreading it; (ii) social media platforms, which should identify it, label it as misinformation or remove it; and (iii) public health authorities and health providers, who should increase their presence and COVID-19-related activities on social media. Our review investigated only the initial phase of the pandemic; future developments are likely to result in new types of misinformation.

25As more countries experience additional surges in severe acute respiratory syndrome coronavirus 2 infection rates, social media will come to play an increasingly important role in disseminating accurate information. The knowledge acquired in our review of COVID-19-related misinformation may help health-care organizations prepare their responses to subsequent phases in the COVID-19 infodemic and to future infodemics in general.