Complementary, alternative, and integrative medicine-specific COVID-19 misinformation on social media: A scoping review

Abstract

Background

The sharing of health-related information has become increasingly popular on social media. Unregulated information sharing has led to the spread of misinformation, especially regarding complementary, alternative, and integrative medicine (CAIM). This scoping review synthesized evidence surrounding the spread of CAIM-related misinformation on social media during the COVID-19 pandemic.

Methods

This review was informed by a modified version of the Arksey and O'Malley scoping review framework. AMED, EMBASE, PsycINFO and MEDLINE databases were searched systematically from inception to January 2022. Eligible articles explored COVID-19 misinformation on social media and contained sufficient information on CAIM therapies. Common themes were identified using an inductive thematic analysis approach.

Results

Twenty-eight articles were included. The following themes were synthesized: 1) misinformation prompts unsafe and harmful behaviours, 2) misinformation can be separated into different categories, 3) individuals are capable of identifying and refuting CAIM misinformation, and 4) studies argue governments and social media companies have a responsibility to resolve the spread of COVID-19 misinformation.

Conclusions

Misinformation can spread more easily when shared on social media. Our review suggests that misinformation about COVID-19 related to CAIM that is disseminated online contributes to unsafe health behaviours, however, this may be remedied via public education initiatives and stricter media guidelines. The results of this scoping review are crucial to understanding the behavioural impacts of the spread of COVID-19 misinformation about CAIM therapies, and can inform the development of public health policies to mitigate these issues.

1. Introduction

Social media has grown rapidly over the past decade, transforming from a medium that simply connects individuals from around the world to a platform for the consumption of news, facts, and health advice, but also misinformation. There are an estimated 4.55 billion social media users around the world as of October 2021.¹ The five most commonly used social media platforms among these users include Facebook, YouTube, WhatsApp, Instagram, and Facebook Messenger.² The popularity of different social media platforms varies both by demographic and region. In North America, for instance, Instagram is popular among youth and adolescents, with 72% of teens in the United States actively using the platform.³ In contrast, 69% of American adults predominantly use Facebook.⁴ Similarly, across Europe, the most commonly used social media platforms include Facebook, Twitter, Pinterest, and Instagram.⁵ In China, social media popularity trends differ, with platforms such as WeChat and Sina Weibo being the most frequently used media sites.⁶

Typical uses of social media have been described in the literature, which include socializing, romance, job-seeking, business, and information-seeking.⁷ Social media also allows the public and patients to access health-related information efficiently as information sharing is unrestricted, and therefore, easily accessible. While much of the health-related information on social media is accurate, many social media platforms are inundated with misinformation. For instance, users on various platforms have been shown to espouse unsubstantiated claims about Coronavirus Disease 2019 (COVID-19) vaccines.⁸ Misinformation is an umbrella term that is used to define false or inaccurate information that was created and propagated deliberately, regardless of whether or not there was an intent to deceive.⁹ Examples of COVID-19 misinformation on social media include the promotion of ingesting vitamin E as a way to increase immune system functioning as a means to counteract COVID-19,¹⁰ as well as the rumor that the proliferation of cellular 5G network was responsible for the spread of COVID-19.¹¹ Despite the presence of misinformation, users frequently turn to social media to seek advice and help from other users, and taking this advice into account when making medical decisions.¹² More recently, people have engaged with social media to garner a better understanding of complementary, alternative, and integrative medicine (CAIM). Specifically, the Pew Research Center has identified that 35% of internet users have searched for CAIM information online, making it a rather mainstream health information topic across social media.¹³

CAIM encompasses many different types of treatments that are not generally considered part of conventional medicine and can be better understood by breaking down its components into distinct definitions. The National Center for Complementary and Integrative Health (NCCIH) defines “complementary” medicine as approaches in which the procedure or treatment is used together with conventional medicine, whereas those approaches in which the procedure or treatment is used in place of conventional medicine are termed “alternative” medicine.¹⁴ “Integrative” medicine conjoins and coordinates complementary and conventional health approaches with an emphasis on treating a whole person as opposed to a single organ system.¹⁴ While not considered a component of typical “Western medicine”, CAIM has broad-ranging uses in many different health settings around the world. Most notably, East Asian countries regard certain CAIM therapies as a part of their conventional medical practices, with patients in countries such as China and Japan reporting use of CAIM modalities including acupuncture, moxibustion, herbal medicine, and massage at a prevalence ranging from 27%−44% of the population.¹⁵ Its popularity, however, is not unique to Asian cultures; indeed, CAIM practices are now used by 33% of all adults in the United States.¹⁶ In regions outside of Asia, popular uses of CAIM include acupuncture or homeopathy in Europe,¹⁷ and yoga, meditation, chiropractic or naturopathic medicine in the United States.^18,19

There is a myriad of reasons explaining the popularity of CAIM. For instance, several studies have reported on the positive wellness impacts of CAIM use, a result that appears to drive its uptake.^20,21 In addition, as individuals become increasingly interested in accessing a variety of health services and obtain elevated health literacy, they may begin to see CAIM as an appealing component of their healthcare experience.22, 23, 24 The ability to seek CAIM information online facilitates the spread of information and can be useful in inspiring others to learn more about health-related topics.²⁵ While this poses benefits, social media lacks the scientific regulation and rigor that is typically provided to patients by healthcare practitioners, and the anonymous nature of many online forums facilitates the spread of misinformation on these platforms.²⁵ One health topic that has been rife with misinformation is COVID-19.²⁶ With social media being the primary outlet for individuals seeking COVID-19 health information, many individuals have come across misinformation about CAIM treatments, and some studies reveal that a majority of users believe this misinformation to be factual.²⁷

The World Health Organization (WHO) defines COVID-19 as an infectious disease caused by the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) virus.²⁸ This disease has quickly impacted the lives of many, with 767.4 million confirmed cases and 6.9 million deaths worldwide as of May 2023 since its onset in December 2019.²⁹ Many individuals have used social media sites to obtain health information regarding COVID-19. These online forums are a well-documented medium by which both accurate and inaccurate health information regarding COVID-19 quickly spreads.³⁰ One reason for this is that, due to the nature of social media, anyone can easily share information with a wide audience, unlike other media.³¹ In addition, influential figures are often seen as trustworthy to their large follower base and can exploit this to spread misinformation. Others have posed considerable evidence demonstrating that COVID-19 misinformation spreads on social media as a result of a failure to consider whether or not the information they are consuming is accurate before deciding to share it³²

The aim of this scoping review is to summarize the content of studies exploring COVID-19 misinformation on social media that make mention of CAIM, and what these studies report about these groups of therapies. To our knowledge, this is the first review of this nature capturing information at the intersection of these three topics. As a result of the novel importance of COVID-19 health information in people's lives and the growth of social media as a platform to receive that information, we view this study as highly relevant to gathering a broader understanding of the research in this field as it pertains to CAIM, while also identifying any knowledge gaps. The results of this review could be used to aid decision makers and social media companies in their efforts to combat the spread of misinformation, as well as equip future clinicians and researchers with the ability to protect patients and the general public against the dangers of misinformation.

2. Methods

This review was informed by Arksey and O'Malley's³³ five-stage methodological framework for conducting a scoping study and supplemented by modifications proposed by Levac, Colquhoun, & O'Brien and Daudt, van Mossel, & Scott.^34,35 The framework employed the following stages as follows: 1) identifying the research question; 2) identifying relevant studies; 3) study selection; 4) charting the data; 5) collating, summarizing and reporting the results. The five-stage framework was used to ensure that all scoping review prerequisites were met, which included finding and analysing all relevant and current literature on the topic, summarizing the findings, and recognizing knowledge gaps to be targeted for future research.

Examples of modifications suggested by Levac that were incorporated into this review included clearly articulating the research question being used, using a third member to resolve disagreements on study inclusion, and adopting a team-based, consistent approach to data charting and collating.³⁴ Daudt further expanded upon this by suggesting a multidisciplinary and interprofessional team-based approach to performing a scoping review.³⁵ The PRISMA-ScR checklist for reporting scoping reviews was also used in the reporting of this review.³⁶ This checklist includes 20 mandatory items such as the inclusion of a study rationale, eligibility criteria, and a clear data charting process.³⁶

• Step 1: Identifying the Research Question

Our research question is as follows: With respect to studies exploring COVID-19 misinformation on social media, how many of them make mention of CAIM and what do they report about these therapies? In defining social media, we adopted a formalized definition to remove ambiguity when determining study eligibility. Obar and Wildman,³⁷ define social media by the following four characteristics: “1) social media services are (currently) applications that are Web 2.0 Internet-based; 2) the lifeblood of social media is user-generated content; 3) for a site or app designed and maintained by a social media service, individuals and groups create user-specific profiles, and; 4) the development of social networks online by connecting a profile with those of other individuals and/or groups is facilitated by social media services.” Defining CAIM for the purpose of this review required considerable thought, given that we anticipated that certain items of misinformation may promote actions/behaviours that are blatantly harmful to health (e.g., drinking bleach). In defining CAIM, we built on the aforementioned NCCIH definitions of “complementary medicine”, “alternative medicine”, and “integrative medicine”, informed by a recently published operational definition,³⁸ however, we also specifically expanded our definition of “alternative medicine” to include all mentions of such (outside of this operational definition), even if they were proposed by a non-clinician, and even if they are well-documented to be neither safe nor effective. All this being considered, the purpose of this review was not to make a determination as to whether any particular CAIM therapy is (un)safe or (in)effective for the prevention, treatment and/or management of COVID-19 or any other disease/condition; rather, this study aimed to explore the nature of misinformation (as described by the authors of the included studies) associated with all items that fall within our expanded definition of CAIM.

• Step 2: Finding Relevant Studies

A systematic search strategy was developed by JYN and used to search several bibliographic databases, including AMED, EMBASE, PsycINFO and MEDLINE. Keywords related to social media, misinformation, CAIM, and COVID-19 were used in the search strategy. The database searches were designed to retrieve relevant literature that was published from inception to January 14, 2022. A comprehensive search strategy we used can be found in Table 1.

• Step 3: Selecting the Studies

Table 1.

Search Strategy for Studies Investigating Complementary, Alternative, and Integrative Medicine-Specific COVID-19 Misinformation on Social Media, Executed on January 14, 2022.

Database: AMED (Allied and Complementary Medicine) 〈1985 to January 2022〉, Embase 〈1974 to 2022 January 13〉, OVID Medline Epub Ahead of Print, In-Process & Other Non-Indexed Citations, Ovid MEDLINE(R) Daily and Ovid MEDLINE(R) 1946 to Present, APA PsycInfo <1806 to January Week 1 2022>
Search Strategy:
1.exp Social Media/ or exp Social Networking/ or exp Blogging/ (85,984) 2.(((e or electronic) adj3 newsletter*) or ((peer to peer adj5 network*) or p2p) or (digital adj5 platform*) or (forum* adj3 (internet or web* or chat*)) or (rss adj3 (reader* or feed*)) or (social adj3 media*) or (social adj3 medium*) or (social adj3 network*) or (twitter or tweet*) or (viral adj5 market*) or ("web 2.0″ or "web 2″) or (website* or "web site*" or webpage* or "web page*") or altmetric or badoo or blackplanet or blog* or buddypress or buzznet or care2 or classmates or clicktotweet or "content communit*" or couchsurfing or discord or facebook or faces or fiverr or flickr or flixster or fotki or fotolog or foursquare or gaiaonline or geni or girlsaskguys or googl* or groupme or hashtagify or hashtags or "health 2.0." or hi5 or hkgolden or hootsuite or i-phone* or instagram or keyhole or kik or kuaishou or "last.fm" or linkedin or lipstickalley or lithium or livejournal or "medicine 2.0." or meetup or microblog* or myspace or myspace or naijapals or neopets or ning or okcupid or pinterest or podcast* or qq or qzone or "really simple syndicat*" or reddit or sermo or skype or skyrock or smartphone* or snapchat or stumbleupon or tagged or telegram or tiktok or trendsmap or tumblr or tweepsmap or twitter or "user generated content" or weblog* or wechat or weibo or weixin or whatsapp or xing).ti. (194,576) 3.1 or 2 (243,965) 4.(Coronavirus or COVID-19 or COVID19 or Coronavirus or "Coronavirus Disease 2019″ or "Severe Acute Respiratory Syndrome Coronavirus 2″ or "SARS-CoV-2″).mp. (493,559) 5.(misinform* or fals* or misreport* or misstat* or decept* or distort* or inaccura*).mp. (734,224) 6.3 and 4 and 5 (935) 7.limit 6 to english language (920) 8.remove duplicates from 7 (684)

Two authors (SL and another research assistant) were responsible for title and abstract screening. Only primary research articles (e.g., cross-sectional studies, descriptive analyses, questionnaire-based studies, etc.) were considered and included in this scoping review. Review articles were not eligible for inclusion in this review; instead, their reference lists were manually screened to source additional eligible research articles (or protocols, for which the full study might have already been published) that were pertinent to the research question. To be included, an article had to have been accessible through our library system, available via interlibrary loan, or publicly accessible. Research protocols, abstracts, editorials, opinion pieces, commentaries, and any non-English texts were not eligible for inclusion in this scoping review. During title and abstract screening, studies were only selected for inclusion if the information provided in the title or abstract indicated that the study specifically explored COVID-19 misinformation on social media. This meant that a study was considered eligible regardless of whether CAIM was discussed in the article's full text. This decision was made to avoid excluding relevant articles that included a discussion of CAIM within the full-text, but not within the title or abstract. If there was uncertainty about the eligibility of an article based on the title and abstract, the article's full-text was screened.

Three authors (SL, IM, WP) independently conducted full-text screening to determine article eligibility. SL, IM, and WP completed a pilot screen of a portion of the articles and ensured consistency in their interpretation and use of the inclusion criteria by meeting and cross-referencing their work. During full-text screening, an article was considered eligible if it met all the aforementioned criteria, as well as contained sufficient information pertaining to CAIM. We defined sufficient information as an article containing more than one sentence regarding CAIM in the results section of the article. Disagreement about article eligibility was resolved via discussion with JYN, HC, and DM.

• Step 4: Charting the Data

The following data were collected from each included article: article title, author(s), year of publication, study country, study setting, study design, population type and sample size, type(s) of CAIM used, type(s) of social media used, type(s) of CAIM-specific misinformation, study outcomes and how they were measured, main findings, challenges encountered, and conclusion. SL, IM, and WP charted the data independently and in duplicate; JYN then met with the three authors to resolve any discrepancies. All charted data was then reviewed by HC and DM.

• Step 5: Collating, Summarizing and Reporting the Results

Two types of data were included in this study: charted data and descriptive data. Charted data were summarized by tables, and descriptive data were summarized and reported by thematic analysis. The descriptive data were reviewed by all authors. Then SL, IM, and WP organized findings into thematic groups by identifying and applying codes for the collected information. Additionally, SL, IM, and WP connected the results to the research question and identified gaps in knowledge identified in the review. All authors took part in resolving discrepancies and disagreements.

3. Results

3.1. Search results

Searches retrieved 671 items following the removal of duplicates. Of these abstracts and titles, 527 were eliminated leaving 144 full texts to be considered for further review. Of those, 116 articles were excluded as they lacked sufficient CAIM for inclusion. This left 28 articles for data extraction within the scoping review^31,39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, 57, 58, 59, 60, 61, 62, 63, 64, 65; A PRISMA flow diagram depicting this process is provided as Fig. 1.

Fig. 1.

PRISMA Diagram.

3.2. Eligible article characteristics

 ​Eligible articles were published from 2020 to 2021 and were conducted by researchers from the United States (n = 8), China (n = 2), Spain (n = 2), Saudi Arabia (n = 2), Canada (n = 1), Jordan (n = 1), Brazil (n = 1), Taiwan (n = 1) and Pakistan (n = 1). Additionally, one study was conducted by researchers from Bangladesh and Canada (n = 1), another was conducted by researchers from the United States and India (n = 1), and yet another was conducted by researchers from Italy and the United Arab Emirates (n = 1). Several studies were conducted by researchers from 3 or more countries (n = 6).

Of the 28 eligible articles, half focused on social media misinformation related to COVID-19 from a specific country (n = 14). Of the remaining eligible articles, thirteen focused on social media content on a global scale (e.g., any English posts made on Twitter) (n = 13). The final eligible article focused on COVID-19 misinformation posts on social media in two countries, which were Taiwan and China (n = 1). A variety of CAIM interventions were mentioned within the literature, however, the most common were garlic consumption (n = 9), vitamin C and/or D (n = 8), alcohol (n = 7), chlorine-based products (e.g., hydroxychloroquine, Miracle Mineral Solution) (n = 6), and chiropractic care (n = 4).

The most discussed social media platforms were Twitter (n = 13), Facebook (n = 7), and WhatsApp (n = 6). Multiple methodologies were used by eligible studies, with the most common being cross-sectional studies (n = 5) and qualitative/descriptive thematic analysis (n = 5). A summary of all eligible article characteristics is outlined in Table 2, including study settings, study objectives, and primary outcomes; a more detailed table is available in Table S1 of Supplementary File 1. The main findings, limitations, and conclusions of these articles are outlined in Table S2 of Supplementary File 1.

Table 2.

Summarized Characteristics of Eligible Articles.

First author (year) [ref[	Study setting Study design (Sample Size)	Misinformation discussed about CAIM	Social Media Platforms	Primary Outcome
Almomani (2020)39	Jordan Questionnaire-based study (n = 2000)	Prevention/treatment	Social media in general	Proportion of participants various misinformation as real.
Alotiby (2021)40	Saudi Arabia Descriptive cross-sectional study (n = 1300)	Prevention/treatment	Social media channels (WhatsApp, Twitter, Snapchat, Instagram)	Percentage reduction Awareness videos decreased usage of natural product to treat COVID-19
Alshareef (2021)41	Saudi Arabia Descriptive cross-sectional study (n = 1249)	Prevention/treatment	Mainly WhatsApp, Twitter, & Snapchat. Facebook, Instagram, & Telegram to a lesser degree	Examine prevalence in HCWs and NHCWs spreading COVID-19 natural remedies
Al-Zaman (2021)42	Bangladesh, Canada Mixed methods content analysis (n = 1406)	Prevention	Facebook	Examine percentage of participants mistaking misinformation as fact.
Axen (2020)43	Global (Sweden, Norway, Canada, Denmark, France, Australia, USA, UK Qualitative cross-sectional study (n = 99)	Prevention	Facebook, webpage, YouTube, email, Twitter, other	Examine chiropractic claims involving COVID-19 in discord with scientific literature
Bapaye (2021)44	USA, India Quantitative questionnaire-based cross-sectional study (n = 1137)	Treatment	WhatsApp	Examine misinformation vulnerability of user groups
Chary (2021)45	USA Content analysis of phone calls and tweets (NR)	Prevention/treatment	Twitter (as a proxy for social media in general)	Analyze association of misinformation prompted household cleaner exposure and calls to poison control
Chen (2021)46	China Qualitative descriptive analysis (n = 547)	Prevention/treatment	Chinese social media platforms (e.g., WeChat, Weibo)	Types attention grabbing misinformation on social media
Forati (2021)47	USA Qualitative geospatial analysis (n = 37,587)	Prevention/treatment	Twitter	Association between COVID-19 misinformation on social media and local incidence rate
Galhardi (2020)48	Brazil Quantitative empirical study (n = 154)	Prevention/treatment	WhatsApp, Facebook, Instagram	Types of fake news related to COVID-19 disseminated on social media
Herrera-Peco (2020)49	Spain Observational, retrospective, time-limited study (n = 5040)	Prevention	Twitter	Interactions #yonomevacuno hashtag and spread of antivaccination messages
Islam (2020)50	Bangladesh, Australia, Thailand, & Japan Observational mixed methods retrospective study (n = 2311)	Prevention/treatment	Facebook, Twitter	Impact of misinformation on public health in different countries
Islam (2021)51	Bangladesh, Australia, Thailand, & Japan Mixed methods study (n = 637)	Prevention/treatment	"Google, Google Fact Check, Facebook, YouTube, Twitter, websites of fact-check agencies, websites of television and newspaper"	The verifiability of COVID-19 vaccine related information spread online
Jamison (2020)52	USA Descriptive study (thematic analysis) (n = 2000)	Prevention/treatment	Twitter	Proportion of unreliable tweets among vaccine proponents and vaccine opponents
Kavaluru (2021)53	USA Descriptive study (thematic analysis) (n = 400)	Prevention/treatment	Twitter	Tweets mentioning the use of smoking/vaping to protect/treat COVID-19
Kawchuk (2020)54	Canada, Denmark, Switzerland Descriptive study (thematic analysis) (n = 1118)	Prevention	Twitter	Prevalence of tweets that claim that spinal manipulation therapy will cure COVID-19
Kawchuk (2020)55	Canada, Denmark, Australia Descriptive study (internet analytics) (n = 4773)	Prevention/treatment	Websites, social media pages	Identifying College of Chiropractors of British Columbia (CCBC_ registrants with websites or social media pages containing words not permitted by the CCBC guidelines
Lobato (2020)56	USA Qualitative exploratory study (n = 404)	Prevention/treatment	Social media in general	Factors that affect an individual's willingness to share COVID-19 misinformation
Moreno-Castro (2021)57	Spain Exploratory study (n = 2353)	Prevention/treatment	WhatsApp	Date of receipt, format of message length of message, type of person sending the message, identity of person sending message narrative voice and sex, keys of the message, proposal/recommendation,
Naeem (2021)31	Pakistan Qualitative (content analysis) (n = 1225)	Prevention/treatment	Fact-checkers, Myth-busters, and COVID-19 dashboards, social media	Pattern identification in the fake news and source of misinformation
Rovetta (2020)58	Italy, UAE Qualitative content analysis (internet analytics) (NR)	Treatment	Google, Instagram	The searching behaviours of social media users and the extent of misinformation circulating on Instagram and Google around the world.
Singh (2020)59	USA Qualitative (thematic analysis) (n = 21,417,552)	Prevention/treatment	Twitter	Determining the number of tweets that are discussing more prevalent myths.
Swetland (2021)60	USA Qualitative (thematic analysis) (n = 358)	Prevention/treatment	Twitter	Tweets addressing common management issues surrounding COVID-19 as well as their source and content.
Vijaykumar (2021)61	UK, USA, Brazil Cross-sectional (survey) (n = 726)	Prevention/treatment	WhatsApp	Types of social correction behaviors (SCBs) and health and technological factors that shape the performance of these behaviors.
Vraga (2021)62	USA Experimental (survey) (n = 1596, wave 1; n = 1122, wave 2)	Prevention	Facebook	Researchers tested whether the effects of correction of misinformation endure over 1 week.
Wagner (2020)64	Canada Qualitative (content analysis) (n = 28)	Prevention	Instagram	The frequency and importance of posts about "immune boosting" and its portrayal as beneficial or if the concept was critiqued.
Wang (2021)63	Taiwan Qualitative (algorithm development and content study) (n = 114 124)	Prevention	LINE	Reported messages containing COVID-19-related suspicious content (misinformation)
Zhang (2021)65	China Exploratory (n = 2754)	Prevention/Treatment	Q&A platforms, video-sharing platforms, chat platforms, news-sharing, health care platforms	Daily number of COVID-19-related posts on Chinese social media and changes in post numbers over time periods.

CNR: not reported.

3.4. Findings from thematic analysis

A total of four unique themes emerged from our analysis of the included studies, which are detailed in the text to follow. A figure representing the summary of findings from our thematic analysis is shown in Fig. 2.

Theme 1

Misinformation Prompts Unsafe and Harmful Behaviours

Fig. 2.

Summary of Findings from Thematic Analysis.

Over one-third of the included studies indicated that misinformation on social media can lead to potential harm to individuals or the public more broadly. Two subthemes were developed from the collected studies related to this theme: potential harms to the individual, and potential harms to the community.

Subtheme 1.1

Potential Harms to the Individual

There were a number of studies that identified misinformation that had the potential to cause danger to people's health and well-being.^{45,47,49,50,52,53,58,60} One study, based on data from Massachusetts and Rhode Island, concluded that there was a 34.4% increase in calls to regional poison control centres compared to the previous 8-year period.⁴⁵ This suggested that misinformation about potential treatments for COVID-19 led individuals to consume poisonous substances.⁴⁵ Another study reached a similar conclusion after then United States President Donald Trump posted a tweet suggesting hydroxychloroquine could be a helpful cure for COVID-19, which was followed by 30 cases of disinfectant poisoning in New York City.⁵⁸ In this study, the authors found that when Mr. Trump suggested a potentially harmful disinfectant as an alternative therapy for COVID-19, it led to a spike in Google searches and discussion of the disinfectant as a potential cure for COVID-19 in the following days. Other studies found that misinformation online supported various other treatments that are potentially dangerous to personal health, such as drinking urine, smoking/vaping nicotine, or ingesting bleach as a means to eliminate the virus causing COVID-19.^50,52,53,60 In addition to unsafe treatments, social media misinformation also promoted unsafe behavior. Similar to the discussion of unsafe treatments, these unsafe behaviours were meant to help readers evade the virus. For instance, in an analysis of over 5000 Twitter users participating in anti-vaccine discussion, one study's authors concluded that COVID-19 vaccine hoaxes must be considered when combatting misinformation.⁴⁹

Subtheme 1.2

Potential Harms to Public Health

One article identified threats to public health as a result of the dissemination of misinformation on social media.⁴⁷ This article discussed that five of the top ten states in the United States for misinformation spreading via tweets experienced a surge of COVID-19 cases following the study period, suggesting that misinformation might have been linked with behaviours that were dangerous for public health in the spread of COVID-19. However, it should be noted that this did not necessarily mean that misinformation caused the surge in COVID-19 cases.

Theme 2

Misinformation can be Separated into Different Categories

Many of the studies further specified the types of misinformation reviewed by subgrouping them into different categories.^{31,43,49,54,57,60,63,64} Three studies aimed to illustrate the most common trends in misinformation and categorized their data accordingly. For example, Kawchuk et al. separated a sample of 1118 tweets into the two most common tweet themes based on the frequency of words mentioned.⁵⁴ Other studies took a similar approach, breaking down misinformation into the most commonly mentioned misinformation archetypes such as false claims, conspiracy theories, and pseudoscientific health therapies.^31,63 Some studies aimed to categorize the false claims based on the specific CAIM therapy the misinformation recommended, giving the number or percentage of tweets that mentioned therapies including herbs, zinc, vitamin C, natural drugs, synthetic drugs, sunbathing, or healthy eating.^49,57,60 Finally, other studies categorized the misinformation based on descriptive characteristics of the source of misinformation itself. One study, for example, mentioned that while some misinformation sources referenced unspecified research, others questioned undisputed scientific phenomena, while others contained no argument at all.⁴³ Another study similarly attempted to categorize the misinformation by describing the content, stating that misinformation referred to either medical authorities, evidence and research, or COVID-19 itself.⁶⁴

Theme 3

Individuals are Capable of Identifying and Refuting CAIM Misinformation

Subtheme 3.1

Individuals are Able to Identify CAIM Misinformation with Assistance from Educational Programming

Two studies indicated that individuals could improve their recognition of misinformation through exposure to educational programming.^40,62 For example, one study found that 69.5% of participants were better able to refute CAIM misinformation after exposure to awareness videos curated by the Saudi Arabian Ministry of Health.⁴⁰ Another study found that when individuals were exposed to misinformation correction strategies developed by the WHO, they had reduced misperceptions about natural remedies compared to a control group.⁶²

Subtheme 3.2

Social Media Users Can Often Identify and Refute Misinformation Without Assistance from Educational Programming

Several studies found that the general population was generally capable of correctly identifying COVID-19 rumours that were potentially harmful.32, 34, 35 For instance, Almomani³⁹ found that most respondents found rumours regarding the prevention of COVID-19 to be fundamentally incorrect to them. In addition to refuting potentially harmful rumours, individuals also infrequently recirculated rumours about unproven COVID-19 natural health remedies on social media, demonstrated by the finding that only 35% of Saudi Arabian social media users had reported frequent sharing of natural remedies for the prevention of COVID-19 online.⁴¹ Unsurprisingly, this figure was even lower among healthcare professionals, of whom only 32% reported frequent recirculation of natural health remedies for COVID-19.⁴¹ Beyond simply correctly identifying misinformation, some individuals would take it upon themselves to publicly refute misinformation concerning the efficacy of alternative therapies.⁴² For example, individuals would oppose suggestions to consume ethanol to treat/prevent COVID-19 by citing literature on the harms taking ethanol may cause to the body.⁴²

Theme 4

Studies Argue Governments and Social Media Companies Have a Responsibility to Resolve the Spread of COVID-19 Misinformation

Many studies provided recommendations for future action in response to the spread of COVID-19 misinformation on social media. These recommendations were directed toward government agencies and social media companies.^{31,40,46,48,50,51,53,54,58,60,62,65}

Subtheme 4.1

The Spreading of Educational Material in Response to Misinformation

A number of studies suggested that government bodies or social media companies should respond to the spread of COVID-19 misinformation with educational material.31, 40, 46, 48, 51, 62 Some studies suggested misinformation should be corrected through the sharing of posts by government bodies or social media platforms themselves.^40,46,48,62 One study sought to “evaluate the effect of media on raising the level of health awareness of Saudi Arabian populations regarding the medical misinformation about the use of natural remedies against COVID-19″ using the Ministry of Health's awareness videos.³³ Among 1300 participants, 78.9% accessed this material through social media, of which 69.5% halted the use of natural remedies as protective measures against COVID-19.³³ The study concluded that public health authorities should use social media platforms to increase public health awareness and share educational videos. Two studies also proposed direct community engagement initiatives to limit misinformation and make the public aware of strategies to identify false information.^31,51

Subtheme 4.2

Monitoring Misinformation to Inform Future Action

Six studies focused on the importance of misinformation surveillance and monitoring as critical responsibilities for social media companies and government agencies.^{40,49,50,53,54,60} All six studies asserted that monitoring these occurrences will assist in mitigating the spread of misinformation. One study sought to test how people interacted with different information circulating through social media and online platforms and mentioned the necessity of the improvement of existing monitoring systems.³⁹ A similar conclusion was reached by Islam et al. ,⁵⁰ who suggested that health agencies should track real-time misinformation spread related to COVID-19 in order to debunk it. Another study found that such public surveillance programs were crucial for preventing misinformation.⁴⁹ Not only was monitoring misinformation found to be necessary, but some studies also suggested that this could be achieved by artificial intelligence (AI) .^53,60 Specifically, machine-learning models could be trained to fact-check misinformation and correct it through AI-generated responses.^53,60 Finally, Internet analytics has been suggested as another strategy to engage in surveillance, as it can be used to monitor compliance with jurisdictional guidelines and regulations based on social media activity and could be of assistance to regulatory bodies when keeping pace with Internet activity.⁵⁵

Subtheme 4.3

Collaboration Between Government Organizations and Social Media Platforms

Three studies proposed a collaborative approach between government agencies and social media platforms to address the spread of COVID-19 misinformation.51, 53, 58 Zhang et al.⁶⁵ stated that targeted measures must be taken by both governing bodies and social media companies to counter and control the infodemic. Multi-faceted approaches with efforts taken by both governing agencies and social media companies can assist in diminishing the impacts of misinformation spreading as well as limit the spread of misinformation itself.⁵⁸ Swetland et al.⁶⁰ highlighted that while the US government could amend The Communications Act to require social media companies to monitor and remove inaccurate and harmful information, the impact of this on the spread of misinformation would be unclear.

4. Discussion

The purpose of the present scoping review was to summarize the content of studies exploring COVID-19 misinformation on social media that make mention of CAIM, and what these studies reported about these groups of therapies. This review identified 28 eligible articles, published between 2020 and 2021. The literature identified through this scoping review provides insight into the scope of information available pertaining to COVID-19 misinformation on social media in the context of CAIM. Such interventions included a wide range of food and drinks, off-label drugs, and chiropractic procedures as alternatives to typical COVID-19 interventions. To our knowledge, this is the first review of this nature to capture information at the intersection of CAIM, COVID-19, and the spread of misinformation on social media. It is hoped that this scoping review will provide both practitioners and researchers with a comprehensive overview of the research conducted at these topic intersections.

4.1. Non representative study samples have resulted in considerable generalizability issues

Numerous studies obtained results from a study sample that was not representative of the true population that was being studied. This raises generalizability concerns regarding the studies in this review. For instance, Twitter appeared to be the primary source of social media content in several studies.^49,53,54 This may have misrepresented social media discourse to be more reflective of public Twitter users rather than the population at large, decreasing the generalizability of the findings. Moreover, some studies were conducted based on participant self-reporting which is a method of reporting known to be susceptible to self-biases.39, 40, 41^,44,56,63 There are also concerns about the type of misinformation being spread by private users on these platforms, as this subset of information was inaccessible to researchers and is thus excluded from the data.^42,50 Other studies had questionable sample characteristics, such as one in which conclusions were based on a sample of only 26 participants,³⁹ and another which had an imbalance in the number of participants in their comparison groups.⁴¹ In both cases, explanations pertaining to why the sample characteristics were acceptable for the study parameters were either not provided or not written in ways that provided sufficient details to justify the rationale of their decision. These generalizability issues raised questions surrounding the findings of these studies and whether they reflect the true prevalence of CAIM misinformation pertaining to COVID-19 on social media.

4.2. Comparative literature

The results of this review revealed that COVID-19 misinformation on social media is prompting individuals to engage in potentially harmful behaviours. This included behaviours that caused direct harm to individuals, such as the ingestion of dangerous chemicals. It was also found that the discussion of misinformation on social media might also be damaging at the public health level.⁴⁷ These findings correlate well with other research studies that suggest COVID-19 misinformation can prompt individuals to put their own health at risk. Broadly, it has been illustrated that health misinformation leads to a general distrust in healthcare guidelines and that such distrust leads people to make unfavorable decisions for both personal and population health.66, 67, 68 Studies have also asserted that the panic caused by the spread of misinformation and conspiracy theories can lead to an increased death toll in the face of emergent health crises.⁶⁹ Other research has provided specific examples of misinformed actors in various communities encouraging dangerous behaviours. For instance, one study found that an Imam of a Mosque in Bangladesh was advising religious followers to visit their mosque in spite of social-distancing guidelines, promoting the misinformed idea that religious faith can save followers from COVID-19.⁷⁰

Researchers frequently separated misinformation into categories to reflect different characteristics of the misinformation. Common practices identified in this review included separating misinformation by trends^31,54 or by the specific type of CAIM therapy suggested.^49,57,60 This general pattern has been seen in the discussion of misinformation across many different types of literature. For instance, in the sphere of health misinformation, Chua and Banerjee⁷¹ categorized most misinformation as either a “dread” rumor or a “wish” rumor. “Dread” rumours are those that promote theories of worrisome outcomes, while “wish” rumours are those that promise positive outcomes. Similarly, Himelein-Wachowiak et al.⁷² categorized misinformation findings based on whether they were spread by a robot or by a human, allowing them to better understand the extent of the threat posed by robots to the proliferation of the infodemic. Finally, when analysing the content of social media posts containing misinformation in the midst of the Ebola outbreak, Sell et al.⁷³ categorized misinformation into four types: discord-inducing and non-discord-inducing, in order to evaluate the harm caused by misinformation, and political and apolitical, in order to assess the role of politics in the spread of misinformation.

Recent research examining the impact of health misinformation on social media during the COVID-19 pandemic consistently highlighted potential dangers that may be faced in similar future events. For example, health misinformation has been associated with a decrease in the perceived threat of health events, reduced compliance with measures designed to limit the spread of COVID-19, and lowered trust in treatments backed by scientific evidence.⁷⁴ These findings are consistent with sentiments echoed by many of the studies included in this review.

In response to the above, many researchers found it imperative that governments and social media companies make appropriate efforts to mitigate the spread of misinformation. The frequency of these calls to action reflects the seriousness of the threat to population health posed by misinformation in the era of COVID-19. Researchers of included studies not only called on governments and media organizations to respond to misinformation with educational material,^{31,40,46,48,51,62} but they also called for increased monitoring of social media and the proactive flagging and removing of misleading information.^{40,49,50,53,54,60} This is a common finding across literature on misinformation in the context of COVID-19. Several researchers75, 76, 77, 78 and organizations^79,80 have concluded that the COVID-19 “infodemic” should be combated by more deliberate efforts by large agencies. A sizable proportion of these strategies center around ‘inoculation theory’.⁸¹ Specifically, proponents of inoculation theory highlight the need for psychological inoculation for the general masses against misinformation, such as presenting videos to a general audience which educates them about common signs of false information. Beyond these strategies, the present review also revealed a need for a collaborative effort between governments and social media institutions. This appears to be the consensus among literature studying the social media COVID-19 infodemic.^82,83

Some social media users appeared to be able to correctly identify COVID-19 misinformation in the context of CAIM therapies. This finding is in contrast to the literature, which often suggested an overall persuasive effect of COVID-19 misinformation.⁸⁴ For example, research has indicated that fake news is 70% more likely to be retweeted on Twitter than true news.⁸⁵ Our review also revealed that those susceptible to COVID-19 misinformation appeared to be able to learn to refute it through education and guidelines. This concept is consistent with findings in the literature, in which several studies have reported that education about misinformation has a positive effect on users’ ability to refute misinformation,86, 87, 88, 89 especially if the quantity of corrective information provided is sufficient and provided early on in the information cycle.^90,91

An important barrier identified by many of the articles included in this review was the widespread generalizability concerns in the literature. This was due to a variety of reasons, including sample size concerns and the use of only one social media site when drawing conclusions.^39,49,53,54 This limitation has been found widely among the literature on social media misinformation in the context of COVID-19, as language restrictions, sample size concerns, and information exclusions have all been cited as reasons for generalizability issues across various studies.^74,92 In addition, many of the studies were limited by biases related to self-reporting of data.39, 40, 41^,44,56,63 Typically, in such cases, researchers had employed certain methodological controls such as reverse coding of similar questions to help control for participant biases⁹³; in several of the included articles, these measures were never undertaken.39, 40, 41^,44,56,63

4.3. Areas for future research

At present, information regarding demographic trends of those participating in CAIM discussion and misinformation is often limited to trends in political ideology and age.^39,42,44,57 Based on the findings of this review, it is clear that future research should elucidate several trends. These trends include social determinants of health, such as race and gender, that correlate with susceptibility to CAIM misinformation related to COVID-19. Additionally, research should explore the correlation between public healthcare investment and the spread of CAIM misinformation on social media, as well as the differing effectiveness of strategies to combat CAIM misinformation across demographics. The impact of demographic differences on misinformation diffusion is potentially substantial. For instance, patterns in misinformation spread vary considerably across different social media platforms,⁹⁴ and some have argued that these differences are due to cultural and demographic differences across platforms.⁹⁵ Policymakers and social media companies alike require a comprehensive understanding of demographic trends in misinformation spread as it pertains to CAIM in order to mitigate potential harms to social media users.

In addition, while this review found that CAIM misinformation about COVID-19 treatments often prompted individuals to perform unsafe behaviours, there is little data on whether social media users actually engaged in these behaviours after being exposed to said misinformation on social media. This gap in the literature has been highlighted by numerous researchers,96, 97, 98 and it is agreed that a better understanding of the cause-and-effect relationship will allow public health officials to garner a greater understanding of the threat posed by misinformation online related to COVID-19. For example, if evidence identified that a particular type or style of misinformation was more likely to result in behavioural changes, organizations could concentrate their efforts on mitigating its spread while ignoring less harmful misinformation archetypes.

4.4. Strengths and limitations

A key strength of this review is that the title and abstract screening was completed independently and in duplicate, while the full-text screening and data extraction process were completed in the same manner in triplicate. Furthermore, these aforementioned steps were pilot tested to ensure that the screening and data extraction were standardized. As we only included articles published in the English language, this serves as a limitation of our study as it may have resulted in missing, yet important findings, found in articles published in other languages. Similarly, although the definition of CAIM and the search strategy used to determine article eligibility are comprehensive, we may still have unintentionally excluded certain types of CAIM. Though the definition of social media and the search strategy used to determine article eligibility was also comprehensive, certain types of social media may still have been missed.

4.5. Conclusions

This scoping review consisted of a systematic search of the literature to identify the number of studies exploring COVID-19-related misinformation that also include CAIM therapies, and what these studies report about these interventions. From the 28 articles that met the eligibility and data extraction criteria, four major themes were identified, including the following: 1) misinformation prompts unsafe and harmful behaviours; 2) many eligible studies separated misinformation into different categories; 3) individuals are capable of identifying and refuting CAIM misinformation; 4) studies argue governments and social media companies have a responsibility to resolve the spread of COVID-19 misinformation. This review outlines the existing evidence base with respect to CAIM and COVID-19 misinformation on social media, which can provide a broad understanding of the subject matter to practitioners, researchers, and patients. It is our view that the findings from this review play a key role in comprehending the behavioral consequences of the dissemination of COVID-19 misinformation related to CAIM. These insights may be of value when it comes to the shaping public health strategies aimed at addressing and minimizing these concerns.

Conflicts of interest

JYN is an editorial board member of this journal but his membership had no bearing on the review process or decision. The authors declare that they have no other competing interests.

Funding

This study was unfunded.

Ethical statement

This study involved a systematic review of peer-reviewed literature only; it did not require ethics approval or consent to participate.

Data availability

Raw data associated with this manuscript is publicly available for download on Open Science Framework: https://doi.org/10.17605/OSF.IO/YTZ5E.

Authors' contributions

JYN: designed and conceptualized the study, collected and analysed data, drafted the manuscript, and gave final approval of the version to be published.

SL: collected and analysed data, made critical revisions to the manuscript, and gave final approval of the version to be published.

IM: collected and analysed data, made critical revisions to the manuscript, and gave final approval of the version to be published.

WP: collected and analysed data, made critical revisions to the manuscript, and gave final approval of the version to be published.

HC: analysed data, made critical revisions to the manuscript, and gave final approval of the version to be published.

DM: analysed data, made critical revisions to the manuscript, and gave final approval of the version to be published.