Introduction

Increasing digitalisation and use of social media is a two-edged sword.1 It creates opportunities to rapidly communicate and disseminate information to address social challenges, and is therefore an important tool for reaching individuals and communities.1 2 However, as emphasised by the United Nations,3 digital technologies and social media also have the potential of introducing misinformation to citizens.1 Misinformation, which refers to inadvertent misleading information to which the public may be exposed and share without intent to cause harm, can delay or prevent effective care,2 affect mental health,4 lead to misallocation health resources4 and/or create or exacerbate public-health crises.4 5 Disinformation or malinformation are other common terms, but refer to a purposive strategy to induce false belief, channel behaviour or damage trust and can be very difficult to prove and therefore we use the term misinformation.5

Misinformation can affect some members of society more than others (eg, those with lower digital, numerical and health literacy and/or cognitive skills are more vulnerable to misinformation),6 7 and therefore be more exposed to health threats, leading to greater social and health inequities.8 A systematic review conducted in 2021 found the groups most vulnerable to health-related misinformation include younger people, those with lower educational attainment, racial minorities and social media users.9 The disproportionate impact on women, trans and non-binary people has also been highlighted. For instance, a report published by Plan International Australia10 shows that those groups are bombarded with stereotypes and misleading facts about their bodies and their health. This was found to lead to feeling unsafe because of online health information and questioning whether to get COVID-19 vaccines.

Extensive COVID-19-related misinformation has spurred efforts to mitigate the spread of falsehoods and undermine public trust in evidence-based care. Such efforts were a focus in a report from the Broadband Commission for Sustainable Development11 and in the Global Commission on Evidence.3 Individuals can engage with misinformation through different sources (see table 1).12 In particular, while social media platforms are a key driver of misinformation,4 13 it is not well understood since data is not publicly available for analysis and because many popular platforms (eg, Instagram, YouTube, TikTok and Pinterest), use visual content instead of texts.12 For example, a study that analysed 800 vaccine-related Pinterest posts found that 74% were anti-vaccine in sentiment.14 15

Table 1

Internet sources of misinformation (table adapted from ‘Public health and online misinformation: Challenges and recommendations’)12

Although vaccines were the most common topic of misinformation before COVID-19,16 other common topics for misinformation include reproductive health, substance use or smoking, non-communicable diseases, pandemics, eating disorders and medical treatments.16 Governments have employed a variety of strategies designed to debunk misinformation, including monitoring and fact-checking, economic incentives and legislative policies.17–20 These strategies must be assessed and compared in terms of impacts and effects on health outcomes and behaviour change. For instance, one older evidence synthesis found that correcting misinformation has a moderate influence on belief in misinformation, rebuttals are more effective than forewarnings and appeals to coherence are more effective than fact-checking and appeals to credibility.21 However, the rapid evolution of platforms for information sharing, and growth and innovation of misinformation actors means that previously synthesised evidence may no longer be valid to understand current misinformation challenges. In particular, the rapid expansion of artificial intelligence (AI) has many implications for misinformation, which will need to be better understood. This includes the potential for AI to amplify or propagate misinformation, but also for it to be used as a tool to address misinformation (eg, through automated fact checking and credibility labelling).22 Given this, there is a need for new high-quality and routinely updated evidence syntheses from trusted sources that assess the comparative impact of different strategies.

In 2020, the International Telecommunication Union and UNESCO, sponsored the Broadband Commission for Sustainable Development. This commission developed a report about countering digital misinformation while respecting freedom of expression.11 The report provides a framework explaining the stages in the misinformation process (table 2). In addition, it provides a framework for 10 potentially effective responses to misinformation and the possible intersections with freedom-of-expression rights,11 which is outlined in table 3 along with a classification of five categories of governmental strategies to address COVID-19 misinformation identified in a non-systematic review conducted in 2021.23

Table 2

Five stages of the misinformation cycle (table adapted from ‘Broadband Commission research report on ‘Freedom of Expression Addressing Disinformation on the Internet’)11

Table 3

Potential responses to misinformation (table adapted from: ‘Broadband Commission research report on ‘Freedom of Expression Addressing Disinformation on the Internet’ and ‘Governmental actions to address COVID-19 misinformation’)11 23

Methods

This living evidence synthesis will be reported according to guidance from the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement.24

Registration

We used the PRISMA-Protocols checklist when writing our report,25 and our living evidence synthesis protocol was registered with the International Prospective Register of Systematic Reviews (PROSPERO).

Literature search

We will conduct our first round of searches in May 2023. This will include seven electronic databases (MEDLINE, EMBASE, CINAHL, PsycINFO, COVID-END inventory of best evidence syntheses, Epistemonikos and medRxiv), as well as sources for grey literature (Google Scholar, Open Science Framework and greynet.org). Search terms were developed with the collaboration of a library scientist on our team (TN) using medical subject headings and text words related to forms of misinformation and interventions. The search strategy is included in online supplemental appendix 1 and will be updated at 6 and 9 months after the original search (and potentially further depending on funding availability).

Patient and public involvement

We engaged citizen partners in the development of the original grant proposal to the Canadian Institutes of Health Research, in team meetings and will continue to do so throughout the entire project, including presentation of results, interpretation and preparation of user-friendly materials for dissemination. One of our citizen partners (MS), has championed efforts for citizen engagement in evidence syntheses through COVID-END, Global Commission on Evidence and the Cochrane Consumer Network, and will lead the citizen team that includes three additional citizen partners (MB in Manitoba, CL in Quebec and JP in Nova Scotia).

Study period for the systematic review

The study commenced in May 2023 and completion is anticipated in May 2024.

Eligibility criteria

We will include original articles without language restrictions that evaluate one or more of the potential responses to health-related misinformation listed in table 3 (eg, monitoring and fact-checking, counter-misinformation campaigns, credibility labelling, educational, curational, narrative, technical and algorithmic, economic, investigative, legislative and other policy) for addressing health misinformation. We will include experimental, quasi-experimental, observational studies for any populations, settings and diseases (ie, we will not limit to only COVID-19 misinformation). Outcomes considered will include change in attitudes/behaviour, health benefits, harms and costs. We will exclude evidence syntheses but will review their references to identify additional studies to include.

Eligibility assessment and data extraction

Retrieved studies from all electronic databases will be imported into Covidence. Two reviewers will independently screen all search results for inclusion. We will assess the titles and abstracts of the references to classify them as ‘potentially relevant’ or ‘exclude’, with disagreement resolved by consensus. We will retrieve the full text of all potentially relevant articles, and two researchers will independently review them to make a final inclusion assessment. Any disagreement at this stage will be resolved by consensus.

One reviewer will extract findings from each included study using a standardised form, which will be checked by a second reviewer for accuracy and consistency with the protocol. We will use a pilot exercise before extracting data where two reviewers will independently extract data from a sample of five studies to identify any areas of the form that are unclear and may require revision. The extraction form includes bibliographic information (eg, title, authors, year of publication); study design information; disease focus (COVID-19 or other focus); location of study; population addressed (eg, age of participants, users of different social media or other information platforms, gender and sex of participants and any equity consideration); type of misinformation addressed based on those outlined in table 1 (including whether it is inadvertent, purposeful or not specified); the stages of the misinformation cycle outlined in table 2; type of responses to misinformation using the framework presented in table 3; outcome(s) measured (eg, change in attitudes/behaviour); and effects identified (eg, benefits, harms, costs). We have deliberately scoped outcome measures broadly to include all potential health outcomes described in the literature.

Risk of bias in individual studies

We will appraise included studies for risk of bias and certainty of evidence. For risk of bias, we will use the Cochrane risk of bias tool for any experimental studies. For observational study designs, we will use a version of ROBINS-I (Risk Of Bias In Non-randomised Studies - of Interventions) that was enhanced for assessment of cohort studies in a series of living evidence syntheses evaluating COVID-19 public health and social measures.26 27 We will also use the GRADE (Grading of Recommendations, Assessment, Development, and Evaluations) approach for assessing the certainty of evidence for the outcomes identified. One reviewer will independently conduct the assessments, which will be checked for accuracy by another reviewer. Any discrepancies between reviewers will be solved through consensus.

Data synthesis

Data analysis will involve quantitative analysis if meta-analytical pooling is possible, which will be based on heterogeneity (assessed with I2) being ≤50%. We anticipate that individual studies will vary in effects observed and will estimate the pooled mean effect size using the random effects model. We will use two indicators of publication bias (Rosenthal’s Fail-Safe N and Duval and Tweedie’s trim and fill procedure).28 For interpretation and imprecision assessment, we will use a minimally contextualised approach that considers whether the 95% CI includes the null effect, or when the point estimate is close to the null effect. We will present subgroup estimates of effect and certainty of the evidence if we observe significant differences between subgroups. For the first version of the living evidence synthesis our a priori approach will be to analyse COVID-19 and non-COVID-19 oriented strategies with further subgroup analyses determined in collaboration with our team, which includes partners from government agencies and citizens. If meta-analytical pooling is not possible, we will synthesise quantitative data according to outcomes and interventions addressed and present a narrative summary of findings using tables and charts, and disaggregating by sex and/or gender, irrespective of whether differences were found.

Updating

We anticipate producing the first version of the living evidence synthesis by the end of August 2023 with updates produced in November 2023 and March 2024. The search strategy will be adapted to include new knowledge about strategies, outcomes and sources of information to ensure we capture all the relevant scientific evidence regarding misinformation. Adopting a living approach for this evidence synthesis given that the context of misinformation is constantly shifting (eg, with increased use of AI) and may change applicability of findings over time and evidence is being produced rapidly in this area which has the potential to change conclusions about approaches over time.22

Ethics and dissemination

There are no individuals or protected health information involved and no safety issues identified.

The systematic review does not involve individuals or protected health information and there are no safety

issues identified. Each version of the living evidence synthesis will be published through the Global Commission on Evidence and COVID-END websites to ensure the most recent update is always publicly available. In addition, we publish a full version in a peer-reviewed journal that will also point to where the future updates will be housed. Moreover, we will publish plain-language materials (a summary and videos) that adopt user-friendly and engaging formats and that will be produced with our citizen partners.