Analysis of pro-vaccine and anti-vaccine content published on YouTube, Facebook and internet

Paloma Piqueiras1 , Giorgio De Marchis 1 , Victoria Cuesta Díaz 2

INTRODUCTION

Several documents claim that vaccines are the most important public health achievement of the 20th century (e.g. World Health Organization, UNICEF, World Bank, 2009; Bean, 2011; Dube, Vivion and MacDonald, 2015, Deloitte, 2017). In terms of data, the World Health Organization (2009) has estimated that vaccines save around 3 million lives per year, equivalent to the population of Madrid or Rome. In terms of effectiveness, vaccination has managed to control a large part of existing diseases (Ehreth, 2003; World Health Organization, 2008) and has even led to the eradication of smallpox, a pathology that "once killed 30% of its victims" (Bean, 2011, p. 1875) or, in America, polio (Dube et al., 2015).

The high vaccination coverage rate in most countries indicates that vaccination remains a widely accepted public health measure. Thus, "more children are being reached with immunisation today than at any other time in history" (World Health Organization, et al., 2009, p. XX) and the benefits of vaccines "are increasingly extending to adolescents and adults, protecting them against life-threatening diseases" (World Health Organization, et al., 2009, p. XX). However, despite this evidence, there is still some mistrust in its application.

Traditionally, reducing the prevalence and incidence of vaccine-preventable diseases has depended on context, policy, science, public health and the media (Dube et al., 2015, p. 99). Practitioners have also played a major role in citizens' decisions about immunisation. In this regard, probably the most widely publicised example is that of Andrew Wakefield, a British doctor who in 1998 published in the prestigious journal The Lancet an investigation associating the MMR vaccine with autism. Although the study turned out to be a fraud, the level of hysteria caused by his claims led many parents to decide not to vaccinate their children for fear that they would develop autism. The result was a spike in measles cases.

Recently, the referral sources of information have changed and, therefore, other variables have appeared in citizens' decision algorithms about whether or not to vaccinate: "although health care providers are the primary source of medical advice, including advice about vaccination, the public is increasingly turning to the Internet" (Bean, 2011, p. 1875). In this regard, some research highlights the importance of the internet in opinion flows generated about vaccines (Dannetun, Tegnell, Hermansson, Giesecke, 2005; Cuesta and Gaspar, 2014; Cambra, Díaz, Herrero, 2016).

Several authors argue that the Internet has provided anti-vaccinationists with great opportunities for exposure and that it plays a decisive role in fuelling anti-vaccination sentiment (Davies, Chapman and Leask, 2002; Cambra, Díaz, Herrero, 2016). Following this logic, the emergence of social networks and the Internet has multiplied the possibility of finding erroneous content and, consequently, has led to public health decisions being made on the basis of inaccurate or misleading information.

Based on this belief that traditional models of information transmission and health message persuasion are changing, this article analyses the anti-vaccine and pro-vaccine content that is transmitted through blogs, Facebook and YouTube. It aims to identify the sources of information and the type of content shared among the user community.

OBJECTIVES

The overall objective of this research is to analyse the type of vaccination-related information that is distributed on blogs, YouTube, and Facebook groups.

The specific objectives are as follows:

• Identify sources of pro- and anti-vaccine information;

• Know the content they publish and the topics they debate on networks;

• Quantify the interactions that both anti-vaccine and pro-vaccine people are able to create.

• Find patterns of behaviour in the generators of anti-vaccine and pro-vaccine content according to the channel through which the content is distributed (blogs, Facebook, YouTube).

To achieve these objectives, the following research questions (RQ) and their associated hypotheses (H) are posed (Table 1):

Source: Own elaboration

METHODOLOGY

This article details the data collected on vaccines and anti-vaccines recorded in various blogs, YouTube channels, and Facebook groups. The data in this paper was collected between 23 December 2019 and 23 January 2020.

The methodology has been developed in two steps: a) on the one hand, a comprehensive data recording procedure was stipulated and; b) on the other hand, the specificities of each search field were defined.

The general data collection scheme is the same for all channels analysed (source origin, content and interactions). However, the information recorded varies according to the research possibilities of each medium (YouTube, Facebook, and blogs). Consequently, it has been taken into account that on YouTube interactions can be counted in positive and negative likes, but that on blogs it is not possible to make this record. In the same way, the possibility offered by some blogs of sharing content directly on other social networks, such as Twitter and LinkedIn, has been taken into account, despite the fact that this option is not possible on the other channels studied.

The construction of the method is detailed below and, ultimately, the sample used is described.

Data recording procedure

The following specificities of each channel were taken into account when recording the data:

• YouTube

• Language: videos in English.

• Keywords: the title of the video contains the word "vaccine(s)" or "anti-vaccine(s)".

• Filters: using the application's search engine, we filtered by entering both the keywords described in the previous point and the dates indicated in the following point.

• Period: the date of publication of the video is equal to or after 1 January 2015.

• Blogs

• Language: content in English.

• Keywords: the word "vaccine(s)" or "anti-vaccine(s)" is contained in the text.

• Filters: The search process was done through the tags of the blogs themselves or by filtering through their search fields.

• Period: the date of publication of the post is equal to or after 1 January 2015.

• Restrictions: for a post to be taken into account, the number of posts that the author of the blog had made about the content to be analysed was counted. Thus, any blog with less than five texts referring to vaccines was discarded.

• Facebook

• Language: content in English.

• Keywords: the title of the group contains the word "vaccine(s)" or "anti-vaccine(s)".

• Filters: the option "groups" was selected in the search, discarding other research possibilities such as "publications", "pages" or "videos".

• Time period: the groups have a minimum activity of 10 publications per year since 1 January 2015.

• Restrictions: analysis of publicly accessible groups. Private groups were discarded due to the impossibility of viewing shared content.

Explanation of the fields recorded in each channel

Generally speaking, the analysis of the content of the chosen channels is structured in three main blocks:

1) The origin of the source to collect data on the authorship of the video.

2) The description of the video to detail the content and its quality.

3) The level of engagement generated to account for the impact the content has had on users.

Specifically, the record fields within each block are described below:

YouTube

• Origin of the source.

• Author's name: name of the person who is the author of the video.

• Name of the channel: name of the channel where the video is inserted.

• Country - continent: country of origin of the video or continent (in case the exact country is not known).

• Description of the content.

• Title: title of the video.

• Date: date on which the video was uploaded to YouTube.

• Duration: time in seconds.

• Summary: synopsis of the content of the video.

• Genre: mode of transmission of the information. There are five options:

• Monologue (the author of the video intersperses images with voice-over statements and his or her own presence);

• News (the video is about a news item from a news or TV programme);

• Report (the video is a report);

• Interview (a journalistic interview is broadcast);

• Debate (a debate on vaccines is shown).

• Primary source: the person who generates the video. There are five possible options:

• Private person (this is a blogger who includes the video on his or her channel);

• Journalist (the video is an excerpt from a news report or TV programme where the host of the talk or content is a journalist);

• Journalist collective (the source creating the video is a collective rather than a single journalist);

• Health professional (a video created and disseminated by a doctor).

• Secondary source: the secondary source only appears when the video reflects an interview or a report. It then records who is being interviewed or who is participating in the report. There are six possible options:

• Doctor (the person the journalist asks - the main source - in the interview or report is a doctor);

• Citizen (statements from citizens are included in the story we are watching in the video);

• Doctor and citizens (the people who appear in the story - where the main source is a journalist - are doctors who provide a scientific view and citizens who contribute their opinion);

• Official government statements (statements from a member of the government are inserted during the development of the story);

• Expert (the interview is conducted with an expert in the field but not a doctor).

• Stance: opinion on vaccines. There are four possible options:

• Pro-vaccine (in favour of vaccination);

• Anti-vaccine (against vaccination);

• Neutral (no opinion either for or against);

• Ambivalent (the two positions are confronted, thus establishing a debate between those interveners who are in favour and those who are against vaccination).

• Quality: this reflects whether the content is supported by scientific sources (quality content) or whether there are no clear references to the sources of the information (non-quality content).

• Level of interactions generated.

• Views: number of views of the video at the time of the registration date.

• Likes: number of likes of the video at the time of the registration date. Both positive and negative likes and dislikes are recorded.

• Subscriptions: number of subscriptions to the channel.

• Comments: positive, negative and neutral comments made to the video by users. For videos with less than 50 comments, the percentage of positive, negative and neutral comments was counted. For videos with more than 50 comments, a sample was selected (the first 20 comments) and the results were extrapolated to the total. To do this, the number of positive, negative and neutral comments in the sample was multiplied by 100 and divided by the total number of comments selected.

Percentage

Blogs

• Origin of the source.

• Author's name: name of the person who is the author of the post.

• Name of the blog: name of the blog where the post is inserted.

• Type of source: whether the source is official (an official body) or unofficial (an individual or group of professionals).

• Description of content.

• Title of the post.

• Summary: synopsis of the content.

• Type of content: whether it is based on information, on the author's opinion or on scientific data and statistics. Formulas may be combined.

• Position: Opinion regarding vaccines. There are three possible options:

• Pro-vaccine (in favour of vaccination);

• Anti-vaccine (against vaccination);

• Neutral (no opinion either pro or con in this post);

No results were recorded for the ambivalent stance, although this search criterion was taken into account.

• Date: the day the post was published.

• Quality: similar to the "type of content" column, but in this case, the aim is to distinguish whether or not the content shared is based on quality and notable sources.

• Level of engagement generated.

• Likes: number of likes of the post at the time of the registration date.

• Reads: number of reads of the post up to the date of registration.

• Subscriptions: number of subscriptions to the blog.

• Comments: comments made by users. There are 3 possibilities:

• Positive;

• Negative;

• Neutral.

• Shared content: This records how many times the content has been shared directly from the blog on other social networks such as Facebook, LinkedIn and Twitter.

Facebook

• Origin of the source.

• Source type: the name of the Facebook group and the type of profile (whether it is an official source or not) is recorded.

• Description of content.

• Posture: It is recorded whether the group has content:

• Pro-vaccine (pro-vaccination) or;

• Anti-vaccine (against vaccination).

• Content type: whether the content is scientific, informational or anecdotal. The options can be combined.

• Start of publications: start of the group's activity.

• Quality of the content: it is recorded whether it is:

• Appropriate (if it reproduces content about vaccines) or;

• Inappropriate (if the content shared is jokes, images, etc.)

• Level of engagement generated.

• Members: number of persons belonging to the group.

Sample

The sample of this research is composed of 48 YouTube videos, 207 posts from 5 different blogs with disparate trends and 7 public Facebook groups. The information is specified in Annex 1.

RESULTS AND DISCUSSION

This section describes the results of the research in terms of the research questions posed.

Research question 1 (RQ1): The information shared in the studied media is mostly issued by non-official sources. In this research, official sources are understood to be government bodies, local or regional administrations, hospitals and medical professionals.

Specifically, two YouTube videos belonging to medical specialists have been detected. In the remaining 46 videos, the authorship is divided between (Figure 1): bloggers (12 videos, 25%); the media (33 videos, 69%) and a group of journalists (1 video, 2%).

Figure 1: Authorship of the videos analysed.

Source: Own elaboration

Of the groups shown in Graph 1, it can be seen that videos from the media have a special characteristic: in 63.3% of the cases (21 videos) the media outlet uses official sources to provide quality content. Thus: a) in the 6 videos in which a news item is broadcast, official statements on the fact narrated are recorded; b) in 5 of the 10 videos in which interviews are recorded, the journalist asks experts; c) in 5 of the 10 reports analysed, statements by doctors are included; d) in the 6 videos showing debates on vaccination, there is only one in which doctors do not participate and; e) in the only video broadcast by a comedy programme, the journalist made a monologue without resorting to sources.

In the exploration of Facebook groups, no official group was found.

Among the blogs, the Real e Ilustre Colegio de Farmacéuticos de Sevilla (RICFS) was considered an official source, as well as the Pills Blog and the Live Med Iberia Blog. In the case of the latter two, although they do not represent any official body, their posts are written by members of the healthcare community (especially doctors and pharmacists).

Based on these data, Hypothesis 1 is validated:

H1: The sources of information transmitted on the internet are not official bodies.

Research Question 2 (RQ2): The content analysed responds to recurrent and topical themes, but the form of transmission and the focus of the information varies depending on the medium explored. In this respect, there are clear differences between a YouTube video, a blog or a Facebook post.

On YouTube, the publication of videos is directly linked to topical content. In this respect, it is clear that the media as well as bloggers and journalists publish their information when the vaccination is in the news. In contrast, in the two cases of videos published by doctors, the reason for broadcasting is not current affairs, but the transmission of their own opinions on the benefits of vaccination.

The content of the blogs is very disparate and it is difficult to find common patterns among them. However, in order to facilitate the presentation of the results, three categories have been established: 1) Recurrent information: an index of topics covered can be clearly established. This category includes Miguel Jara's blog, in which information on the Human Papilloma Virus (HPV) vaccine, criticism of the pharmaceutical industry and the supposed ineffectiveness of the measles vaccine occupy a large part of the posts; 2) Consulting: the authors' posts answer users' questions. This pattern of behaviour is detected in the RICFS, isalud and Live Med Iberia blogs; 3) Miscellaneous: The wide variety of topics makes it impossible to categorise them. The Pills blog is characterised by publishing pro-vaccine content with information supported by data and statistics, but its posts are not connected.

With regard to Facebook, 8 public groups with vaccine-related content were identified, but only 7 of them were considered in the final study (Table 2). This is due to the fact that one of the groups does not offer content in line with the study topic.

Source: Own elaboration

The content of these groups is very generic, but is associated with current affairs. In fact, media content is shared in all the groups. In this way, a user publishes a news item from a media outlet and the interested Facebook community comments on it and interacts with it.

The information provided validates the second hypothesis:

H2: The contents respond to current issues.

Research question 3 (PI3): This research question could not be answered, because the way of counting interactions or influence capacity is not comparable between YouTube, Facebook, and blogs.

It is possible to state that the interactions on YouTube videos are significant (approximately half of the videos analysed, 20 out of 48, have more than 100 comments) and that the ratings on Facebook group content are very low (there are hardly any posts with more than 5 "likes"). However, as the measurement methods are different, we cannot establish direct relationships or conclude which medium has more followers.

Nevertheless, as part of the results, several comparisons are provided according to the media analysed:

On YouTube, the level of interactions generated is remarkable. By authors: a) videos from the media have more than 606,000 views and generate 19,223 positive likes and 1,047 negative likes; b) the total number of views reached by the videos analysed which are broadcast by bloggers is 2,212,313. Likewise, the total number of positive likes is 170,383, while the negative ones barely exceed 4,000. Furthermore, there are 21,005 comments registered on these publications, of which 38% are in favour of vaccines, 41% maintain a neutral position and 21% declare themselves to be anti-vaccine (Graph 2); c) finally, the two videos made by medical professionals have a total of 133,887 views and have the peculiarity of obtaining more negative likes (134,211) than positive ones (991).

Figure 2: Trends in comments made on blogger videos.

Source: Own elaboration

In the case of blogs, it is difficult to make assessments. The data collected are very disparate because the measurement variables do not coincide in all publications. This means that for the Pills blog only the comments could be counted (29 in total) and for the Live Med Iberia blog only the reading rates of each entry could be recorded (an average of 1,060 per publication). On the other hand, the RICFS blog and the isalud blog do not specify any of the variables used to measure interactions. For this reason, only the data from Miguel Jara's blog are presented in detail below.

Although the number of subscribers is not public, it has been calculated that Jara accumulated 575,858 likes on the publications analysed (an average of 4,500 per publication) and his content is massively shared by his readers (Graph 3) on Facebook (972,066 times), Twitter (14,501) and LinkedIn (632). In addition, the total number of comments on its posts is 1,100, a figure far higher than the number of comments made, for example, on the Pills Blog (29).

Figure 3: Miguel Jara's blog interactions counted in "likes" and content shared on other social networks.

Source: Own elaboration

Facebook interactions are the least representative of the entire sample. When compared with the other media analysed, we found that although the groups studied usually have more than a hundred followers, their activity and interactions are very low.

The frequency of content publication varies depending on the group studied, but, in general, content is shared weekly and monthly. On the other hand, the reactions to these shared contents are very scarce (an average publication has 3 "likes", practically no comments, and its impact index is, in almost all cases, less than 50 views).

Research question 4 (RQ4): The results of RQ4 on the ability of the media studied to generate pro- or anti-vaccine interactions are presented.

In the exploration of the data, it has been observed that both YouTube videos and blogs analysed tend to be pro-vaccination: 60.4% of the videos and 80% of the blogs are pro-immunisation. However, less than half of the Facebook groups (42.8%) are pro-immunisation.

A priori, it appeared that anti-vaccine content found Facebook to be a friendlier medium for sharing their information. However, this statement should be qualified by adding an important point regarding blogs: Miguel Jara's blog, the only blog detected with an anti-vaccination tendency, has twice the capacity to generate content and interactions than the other 4 blogs combined (Table 3). Thus, of the 207 entries analysed in the category, 128 belong to Jara.

Source: Own elaboration

The data presented in the analysis allows for the refutation of Hypothesis 4:

H4: Facebook and YouTube will attract more anti-vaccine content, while blogs will tend to share pro-vaccine information.

During the research, it was found that anti-vaccine content is shared more on Facebook and blogs. In contrast, pro-vaccination content is mostly found on YouTube.

CONCLUSIONS

Returning to the outline of the methodology, the conclusions detailed below are structured into: 1) origin of the source, 2) content of the information, and 3) interactions generated.

According to the origin of the source

• The anti-vaccine or pro-vaccine information found and analysed does not, in most cases, come from official sources.

No Facebook groups with governmental or administrative authorship have been detected. In the case of Facebook, only 2 of the 48 videos come from medical sources and, in the case of blogs, the only one that represents an institution of collegiate professionals is the Blog of the College of Pharmacists, in that it represents an institution of collegiate professionals.

• Information sources tend to be collective in the case of Blogs, but individual on Facebook and YouTube.

On Facebook and YouTube, it is most common to find individuals giving their opinion on vaccination. While it is true that most of them use a pseudonym to publish their content, each of the people who participate is easily identifiable insofar as their names are linked to a profile within the Social Network itself.

This is not the case for the Blogs: 4 of the 5 analysed are managed by a collective (RICFS) or by a compendium of professionals (El blog de Pills, Live Med Iberia, El blog de isalud). However, only in the cases of RICFS and Live Med Iberia do we know the name of the author of the information, while in the rest, the entry is signed by a collective (for example, "isalud team").

It should be noted that the case of Miguel Jara's blog is different. The author is perfectly identifiable and takes responsibility for the information he shares in his space.

According to content

• The content varies according to the medium analysed.

It has been identified that the topics shared on YouTube respond to the current situation. This means that the pattern of publications followed in this medium is linked to current events.

With regard to blogs, different categories have been registered to structure the content: a) Recurrent information; b) Consultancy, and c) Miscellaneous). However, no clear similarities were found in the information shared in each blog.

Finally, in the study of the Facebook groups, it can be concluded that information is posted according to current hot topics on vaccination.

• Published content tends to be either pro-vaccination or anti-vaccination, depending on the medium through which the information is conveyed.

There is a certain tendency to support vaccination on YouTube and, in a certain sense, there is also a palpable rejection of anti-vaccination content. However, in the videos shared by the media it is common to see how space is given to people who declare themselves to be anti-vaccine, giving them a prominence comparable to that of pro-vaccine people.

Anti-vaccine content is more common in blogs (128 posts are anti-vaccine content compared to 79 pro-vaccine posts) and on Facebook (4 of the 7 groups share anti-vaccine content).

• The way in which information is transmitted varies according to the channel studied.

While on YouTube the focus is on debates and interviews, on Facebook and blogs it is common for a subject to convey information or opinion in a one-way manner.

According to the interactions generated

• Interactions on Facebook are much lower than those generated on YouTube and blogs.

The level of comments and "likes" achieved on YouTube is much higher than on other channels, perhaps because users are able to interact with the content more easily.

However, it is also possible to generate interactions through a blog. The paradigm is Miguel Jara: he accumulates a large number of comments and has the capacity to have his content reproduced by users on other social networks such as LinkedIn or Twitter. However, this is not the case with the other 4 blogs analysed, whose level of interactions is negligible.

On Facebook, on the other hand, interactions are very limited.

• It cannot be determined that the interactions on the blogs are real.

While on Facebook and YouTube the authors of comments can be identified, this is not the case with blogs. In this medium, the names of the people who share content are not public and access to them through comments is not possible. This makes it difficult to identify DOLs (Digital Opinion Liders) and casts doubt on whether the author has as many followers as the interactions suggest.

• Interactions generated towards anti-vaccine content are more representative in blogs.

This conclusion can be verified by analysing the five blogs proposed in this research. Of these, only one is anti-vaccine and its level of followers, comments and interactions is significantly higher than the other four. It is also true that the content generated by Miguel Jara is much higher than that produced by the other blogs.

In the case of YouTube, the opposite trend is detected: the number of comments and "likes" increases if the content shared is pro-vaccine. Thus, the only two videos registered as anti-vaccination accumulate a total of 11 comments, 1197 views, 37 positive "likes" and 3 negative "likes". These are derisory numbers considering that between the two channels they have more than 340,000 subscribers.

In the analysis of the Facebook groups, no conclusions can be drawn about the interactions carried out, since, although the number of members is higher in the anti-vaccine groups (1,000 versus 352), the level of interaction is low in all of them.

REFERENCES

List of blogs

Roig, V. (2020). El blog de pills [Blog]. Recuperado de https://www.elblogdepills.com/

Isalud [Blog]. Recuperado de https://www.isalud.com/blog/

Live Med Iberia [Blog]. Recuperado de https://www.livemed.in/blog

Jara, M. (2020). Miguel Jara [Blog]. Recuperado de http://www.migueljara.com/

Real e ilustre Colegio de farmacéuticos de Sevilla [Blog]. Recuperado de http://www.farmaceuticosdesevilla.es/blog/

List of Facebook groups

¡La importancia de las Vacunas! [Nombre de usuario] Recuperado de https://www.facebook.com/groups/541904929740761/

Afectados por el mercurio [Nombre de usuario] Recuperado de https://www.facebook.com/groups/604058640089804/

Anti Antivacunas [Nombre de usuario] Recuperado de https://www.facebook.com/groups/307166246860080/about/

Científicos provacunas [Nombre de usuario] Recuperado de https://www.facebook.com/groups/1232552076771162/about/

El peligro del mercurio en las vacunas [Nombre de usuario] Recuperado de https://www.facebook.com/groups/nomasmercurio/

No vacunas forzadas - No SB277 [Nombre de usuario] Recuperado de https://www.facebook.com/groups/novacunasforzadas/

STOP VACUNAS [Nombre de usuario] Recuperado de https://www.facebook.com/groups/STOPVACUNAS/

References