No limit was put on the publication year in the keyword searches for relevant literature. No restriction was imposed on the age of the target populations. The selected articles had to be written in English. The articles needed to be research papers published in journals or presented at conferences. Other document types (eg, reviews, abstracts, editorials, workshop summaries, perspectives, opinions, diagnosis methods, and study protocols) were excluded [23]. Studies undertaken in any country were considered.

The target population was any group in the public worldwide who experienced social media use–induced health inequalities during the repeated resurgences of COVID-19.

The health inequalities discussed and summarized in this review could be any aspect related to any health issues, mental or physical. The health inequalities could be experienced anywhere worldwide, so long as they were induced by social media use and related to COVID-19. All studies satisfying these inclusion criteria were selected for the review.

Social media under discussion in this review referred to ways of sharing information, opinions, images, videos, etc, using the internet, especially social networking sites, including WeChat, WhatsApp, Facebook, Twitter, web-based health community forums, etc.

The comparator could be any form of health inequalities induced by social media. Publications with no comparison were also included because the aim of this review was not to determine the relative degrees of social media–induced health inequalities but to scrutinize the current status of social media–induced health inequalities experienced by people worldwide during the COVID-19 pandemic.

The outcomes of the selected studies we considered were as follows: participants’ physical and mental health inequalities induced by the use of social media and the associated contributing factors.

The designs of eligible studies were quantitative, qualitative, or mixed methods approaches adopted for investigating the outcomes mentioned above. Pilot studies and case studies were included because both types of studies could shed light on the study outcomes above.

Of the 9 finally selected papers, 8 (89%) studies [3,27,28,30-34] were published in 2021 and 1 (11%) [29] in 2020.

All the 9 selected articles were published in peer-reviewed journals. Each of the following journals contained 11% (1/9) of selected studies: BMC Public Health [3], Disability & Society [27], Research in Developmental Disabilities [28], Journal of Medical Internet Research [29], Journal of Racial and Ethnic Health Disparities [32], Annals of General Psychiatry [33], and Health Communication [34]. Of these, 2 papers [30,31] were retrieved from Social Media + Society.

Of the 9 studies, 3 (33%) studies were undertaken in China [3,27,29], 3 (33%) in the United States [30-32], 1 (11%) in Kuwait and Saudi Arabia [28], 1 (11%) in Ethiopia [33], and 1 (11%) in Germany [34].

We originally intended to identify specific locations of study participants to reveal the health inequalities potentially existing between different areas reported in the 9 selected studies. However, only 33% (3/9) of these studies mentioned the places where the participants were located: city, countryside, or town [3]; Hubei Province, China [27]; and historically Black colleges and universities [32].

A total of 67% (6/9) of studies were case studies [3,29-31,33,34] and the other 33% (3/9) were cohort studies [27,28,32]. The studies collected data from participants using web-based cross-sectional surveys and questionnaires [3,28,33], semistructured interviews [27,32,34], a web-based survey (extraction of data from web-based data sets) [29,31], and a mixed methods approach [27,30] (a WeChat ethnography research, participant observation, and semistructured interviews [27]; identifying COVID-19–related content through a keywords-based approach, introducing a measure of sustained amplification, and undertaking a qualitative hand-coding [30]).

In the 67% (6/9) of case studies [3,29-31,33,34], the sample sizes were 981, 1215, 1401, 212, 445, 929, and 22. In the 33% (3/9) of cohort studies [27,28,32], the sample sizes were 110, 190, and 21. It should be noted that in 22% (2/9) of these studies, the samples were 1215 tweets from 134 Weibo accounts [29] and 212,445 Twitter tweets from 137,746 unique users [31]. Most of the studies (6/9, 67%) [3,27,29-31,33] satisfied the standard of an appropriate number of participants (between 150 and 200) proposed by Dunbar [35]. The remaining studies (3/9, 33%) [28,32,34] used small sample sizes below this standard.

Only 56% (5/9) of studies reported the age of the participants, primarily focusing on those aged >16 years [3], ≤60 years [28], ≥18 years [32], 30 to 34 years [33], and 19 to 80 years [34]. With the information provided in the studies, it was impossible to calculate the average age of the participants, and these 56% (5/9) of studies clearly stated the sex of the participants: male and female. Of the 9 studies, 4 (44%) studies [3,29,31,34] had the public as participants; 2 (%) studies [27,28] chose people with disabilities as informants; the other 3 (33%) studies investigated elites [30], college students [26], and educated people with internet access [33]. Merely 22% (2/9) of studies [30,32] reported the races of the participants.

Various types of potential risks of bias were identified in the 9 studies. A total of 44% (4/9) of studies had a small sample size [27,28,32,34]. In all, 33% (3/9) of studies [3,29,33] reported an uneven size (Table 2). In addition, 11% (1/9) of studies [30] mentioned the bias of the lack of comparison between different types of social media and 11% (1/9) of studies [31] referred to weighing all edges equally as a potential bias. Moreover, 44% (4/9) of studies [3,28,32,33] reported that participants were either predominantly male or female, so there was sex bias in these studies. Besides, the study by Zeng and Li [29] pointed out another bias: only using descriptive statistics and content analysis and failing to investigate the psychology and behavior of the audience. These characteristics of the 9 selected studies are summarized in Table 2.

The types of social media, social media-use–induced health inequalities, associated factors, and proposed resolutions are presented in Table 3. Through this table, we intended to compare the main research results of the 9 selected studies.

A total of 33% (3/9) of studies [3,27,28] did not mention specific social media forms contributing to COVID-19–related health inequalities: internet [3], social media [28], and digital media [27]. The remaining 67% (6/9) of studies referred to concrete social media used: Sina Weibo [29]; Twitter [30,31]; Twitter, YouTube, and Google engine [32]; and Facebook, Twitter, Zoom, etc [33,34].

Overall, 4 broad types of social media use–induced health inequalities were revealed in the 9 studies: disparities in the access to COVID-19–related health information [3,27-29]; misinformation regarding COVID-19 and associated precautions [29-32,34]; Twitter accounts receiving disproportionate attention for COVID-19 content during the pandemic [30]; and those obtaining information from social media having significantly higher odds of experiencing psychological distress [33].

In the study by Wang et al [3], in the context of traditional media (eg, newspapers, radio, and television) failing to improve knowledge levels, people with different educational backgrounds acquired different amounts of COVID-19–related health knowledge through the use of the internet, leading to health inequalities during the COVID-19 pandemic. Such COVID-19–related health inequalities created the digital health knowledge gap between individuals with diverse education levels, which influenced them differently in terms of COVID-19–related health behaviors and medical decisions. In contrast, 22% (2/9) of studies [27,28] did not consider the study participants’ education but rather dealt with the COVID-19–related health inequalities between people with disabilities and people without disabilities. A study by Dai and Hu [27] described disabled people’s inadequate, accessible information about COVID-19 in comparison with nondisabled people owing to the gaps between policies and practices regarding the digital accessibility infrastructure. Specifically, the accessible information on the web designed for people with disabilities focused little on applicable information to meet their needs during the pandemic; official press conferences and television news on COVID-19 failed to use sign language interpreters to inform people with disabilities of the latest situations of the pandemic. There was no accessibility design on the newly developed website for disseminating information on COVID-19, making it inaccessible to people with hearing or visual disabilities. The study by Almusawi et al [28] was similar to the study by Dai and Hu [27]. Similar to the study by Dai and Hu [27], the study by Almusawi et al [28] was also concerned with people with disabilities, specifically participants with hearing loss. However, unlike the study by Dai and Hu [27], the study by Almusawi et al [28] focused on the reliance on different information sources: people with hearing loss mainly relied on social media while people with no hearing loss mainly relied on official government sources. Besides, unlike studies by Wang et al [3] and Dai and Hu [27], the study by Almusawi et al [28] touched upon the hearing participants’ low health literacy, which prevented them from accessing health information on the web.

Misinformation was the most prevalent topic in the 9 selected studies. Of these, 56% (5/9) of studies [29-32,34] dealt with COVID-19 misinformation on various social media platforms and related precautions. The study by Zeng and Li [29] discussed disparities between East and Central China and West China in health information and misinformation released on a popular social medium in China named Sina Weibo. The contributors were government public health authorities’ inadequate willingness and ability to use government Weibo accounts, their inactive operation of social media, and the Weibo accounts’ failure to respond to public inquiries and huge amounts of COVID-19–related misinformation promptly. In contrast, 4 studies [30-32,34] investigated COVID-19–related misinformation on Twitter, Facebook, Zoom, YouTube, Google, Facetime, etc, which led to negative outcomes of COVID-19 prevention and treatment, psychological problems, and misleading orientation. The underlying factors included purposeful or purposeless amplification of COVID-19 information, people’s preference for social media over government organs, and people’s orientation-seeking.

Table 3 shows that although the proposed resolutions to COVID-19–related health inequalities are mostly specific to each of the 9 selected studies, what they have in common is intervention on the part of different players including government public health authorities [3,27-30,33], university authorities [32], and scientific communities [31,34]. The intervention measures are concerned with the establishment of relevant platforms [3,27], the development of related programs [3,32,33], the improvement of communication strategies [3,27-30], and the investigation of information-seeking, information application, and information orientation practices [31,34].

With advances in new media and IT, it is of great value to study the COVID-19–related knowledge gap and digital technology–induced unequal health information distribution [3], which were caused by the “Digital Divide” [36], that is, the gaps in the access to and use of the internet among different social groups leading to knowledge gaps [36]. Access gaps may not necessarily breed COVID-19 knowledge gaps because people use the internet media as the most frequent and dependent way to acquire COVID-19–related information [3]. In this case, what induced COVID-19–related health inequalities was use gaps: disparities in intensity, behavior, content, literacy, and other aspects when using the internet media [37,38]. To address the health inequalities caused by the access and use gaps of social media, the following resolutions were proposed: (1) improving the pertinence in the ways and contents of social media–based communication; (2) building social media platforms for authoritative scientific COVID-19 knowledge communication; and (3) developing social media literacy and science literacy of the public [3].

Compared with the people with no disabilities, people with disabilities faced more barriers when accessing health information during the COVID-19 pandemic for two main reasons: (1) the government’s commitment to information accessibility was not always fulfilled, leading to the neglect of the needs for information in people with disabilities; and (2) the newly established web sites for COVID-19 information dissemination lacked accessibility design, thereby being inaccessible to people with disabilities, especially the people with hearing or visual disabilities [27,28]. To eliminate these health inequalities, Dai and Hu [27] proposed a self-initiated and volunteer-driven Disability Support Network, which fully considered various dimensions of information in terms of formats, content, and situations for people with disabilities, and government authorities and public service sectors taking more proactive steps to provide inclusive communications and information in emergencies for people with disabilities on social media. Besides, it is necessary to bridge the gap in health literacy for people with hearing loss using social media and web-based unstandardized dialectic writing and adopting different ways of disseminating information linking to web-based videos in sign language, to ensure equal access to health care and universal compliance with health directives at the population level [28].

The use of social media by public health authorities (eg, Center for Disease Control and Prevention) helped popularize daily health information through Weibo accounts, especially during COVID-19 [29]. However, the high dropout rates of Weibo accounts in some areas and the unequal distribution of Weibo accounts between the eastern region and the middle and western regions caused health inequalities among people in terms of access to helpful information on epidemic prevention and control [29]. The passives state and low willingness and ability of social media operations of Chinese public health authorities and their failure to respond to public inquiries and large amounts of misinformation on social media during the epidemic promptly made misinformation on social media even more rampant, causing even greater health inequalities. Governments in the central and western regions need to learn from the similar experiences of governments in the eastern region, maintain their social media activity by updating daily information frequently, and use social media to release public health information to the public promptly [29].

The role of social media in breeding misinformation attracted the attention and aroused the concern of Wang et al [3]. Social media make it very easy for misinformation and fake news about COVID-19 to spread to the public [31]. A good case is a misinformation on Hydroxychloroquine on Twitter. Interestingly, Donald Trump and his supporters turned out to be the most influential actors in advocating hydroxychloroquine as an effective treatment for coronavirus on Twitter [31]. People’s trust in social media sources over government organs (eg, Center for Disease Control and Prevention and World Health Organization) [32] and their preferred information-seeking and orientation-seeking practices via social media [34] made misinformation on social media platforms even more unconstrained. Misinformation regarding the pandemic frequently appears on social media platforms, serving as a source of health risk [39,40].

Therefore, the media outlet should be more responsible for monitoring health message dissemination [41]. An effective way of countering misinformation lies in the gatekeeping of incorrect information, which helped to fight against the spreading of misinformation during COVID-19 owing to its ability to mediate and fact-check the accuracy of the contents [42]. Another way is to use the disproportionate voice of the crowdsourced elites on the web because people crowdsourced a small set of accounts on social media when clear information about COVID-19 prevention and protection was missing [30]. To mitigate misinformation in college students, university authorities need to consider carrying out programs to aid them in navigating social media for information-seeking and considering the high probability of misinformation [32].

“As networked societies become more accustomed to relying on information from varying sources on social media outlets and other cyberspaces (even for critical medical knowledge), the implications of how they interpret and apply that information in physical spaces is a significant consideration” [31]. Therefore, investigating people’s information- and orientation-seeking practices in and through social media–based communication about COVID-19–related social media coverage [34] can help find practical approaches to minimize misinformation on social media, which most possibly caused health inequalities during the pandemic. What is needed to combat COVID-19–related misinformation on social media is (1) a keen sense of responsibility and the capability to think critically before sharing any information regarding the SARS-CoV-2 virus on social media platforms [31] and (2) a rational judgment of which social media sites are trustworthy and the ability to read and interpret health information on social media critically [32].

Gallagher et al [30] studied the Twitter accounts receiving disproportionate attention during the COVID-19 crisis and the variation across demographics, finding that the public crowdsourced journalists, media outlets, and politicians more than epidemiologists, public health officials, and medical professionals. COVID-19–related health inequalities may arise owing to (1) crowdsourced elites varying across demographics in terms of race, geography, and political alignment; (2) different subpopulations preferentially amplifying elites who are demographically similar to them; and (3) different subpopulations crowdsourcing different elite accounts (eg, journalists, elected officials, and medical professionals) in different proportions [30]. Paradoxically, by working with COVID-19 elites, epidemiologists, public health officials, and medical professionals to popularize scientifically informed health guidelines and debunk misinformation, it is most likely to leverage the crowdsourcing potential of social media to achieve more health equality [30].

In the context of a severe public health emergency, the public depends heavily on media coverage to stay informed [34]. COVID-19 has bred a massive “infodemic” [43] where various social media bombarded people with misinformation and myths about the COVID-19 pandemic, which intensified their groundless anxiety and stress about COVID-19 [33]. As social media exposure was associated with anxiety [44], it is necessary to develop an intervention plan to intervene in people’s psychological distress, especially targeting those who predominantly received COVID-19–related health information on social media platforms [33]. People, especially those experiencing greater psychological distress, need to exercise extreme caution when deriving information on COVID-19 from social media and better use information delivered by the World Health Organization’s “infodemics” team [45]. Moreover, they are encouraged to communicate with others about social media coverage of COVID-19 health information to understand better and evaluate pandemic-related information [34].