This is the ‘Accepted Manuscript’ version of: “Public engagement with clinical research on social media; which visual medium works best? A 5-year retrospective analysis” published in The Journal of Visual Communication in Medicine on 25/8/21, available online: https://doi.org/10.1080/17453054.2021.1950525

Authors: Ciléin Kearns, Allie Eathorne, Alex Semprini, Irene Braithwaite, Richard Beasley

Abstract

‘Public engagement’ describes a collaborative relationship between scientists and the public we serve. This has the potential to improve clinical research and is encouraged by key research funding bodies, however the objective evidence base for effective approaches remains limited. Social media algorithms determine what content users see and are known to weight post media differently. While visual content is understood to improve reach and engagement broadly, less is known about which kinds of visuals are most effective for engaging people with clinical research. We present a five year retrospective analysis of public engagement with Facebook posts made by an independent medical research institute, classified by their visual media content. Inclusion of visual post media was associated with positive effects on both reach and engagement. We present medium and strong evidence that this effect was most pronounced for comics compared to other visual media types. This data evidences objective value of using comics and other visual media for public engagement with clinical research. The metrics evaluated are easily accessible on many social media

platforms meaning this approach could readily be applied by other researchers to measure the impact of their public engagement efforts, and inform science communication strategies and resource allocation.

Keywords

Graphic Medicine, comics, medical illustration, public engagement, clinical research

Introduction

‘Public engagement’ describes a reciprocal relationship between scientists and the public they serve. This collaborative approach can improve research by informing study questions that address societal needs, practical and acceptable study design, can support recruitment, knowledge dissemination, and translation of results into practice.(1–6) Key research funding bodies such as the Wellcome Trust in the UK, the Health Research Council of New Zealand and the Royal Society Te Apārangi in New Zealand, recognise the value of and advocate for public engagement throughout the design, conduct, and sharing of knowledge gained from research.(7–9) While increasingly considered a valuable practice, research on public engagement is often limited by a lack of objective analyses of efforts, and widely accepted tools and metrics with which to do this.

The Medical Research Institute of New Zealand (MRINZ) is an independent research organisation, educational institute, and charitable trust. Over the past five years at the time of writing, it has been using social media to support public engagement with clinical research, study recruitment, and research dissemination efforts. Social media activity has been the responsibility of a small group of staff and interns for which social media is not a primary role or profession, and used to varying effect with and without paid advertisements and post promotion. The more recent half (30-month period) of this five year period has involved resourcing a medical illustrator to support these efforts by designing custom art to share with most of the original posts made by the institute on social media in this time period. This visual content designed has included medical and editorial illustration, comics, animation, data visualisation and diagrams (examples in Figure 1). Additionally, professional content from other creators and Facebook pages has been shared over the entire five-year period including photo, video, graphic design, publication figures, and illustrations.

Figure 1: Examples of visual content that accompanied posts and associated variation in reach and engagement. Top left to top right are examples of a comic, data visualisation, graphic design, and editorial/medical illustration. Bottom left to bottom right are examples of a comic, comic, data visualisation, and comic.

Algorithms are employed on social media platforms to determine which content a user sees, narrowing the near-limitless new content to a more manageable selection a user is likely to engage with based on past activity.(10) While visual content is understood to improve reach and engagement generally on social media (Figure 2);(11) it is not known if this also applies to engaging people with clinical research specifically, nor which kinds of visuals are most effective for this purpose. Such information could inform approaches by researchers and science communicators on how to best use social media to support clinical research aims, and resourcing decisions. In order to determine if the type of visuals used mattered when engaging the public on social media we conducted a five-year retrospective analysis of organic public engagement with the MRINZ’s Facebook posts.

Figure 2: Conceptual illustration of how posts that feature images can stand out from text in a vertical social media feed. Left: a feed of posts with no visuals used. Right: a feed that is predominantly text with one visual post that stands out. Text typically requires a longer and more careful inspection to decipher meaning than visuals. This may disadvantage interaction with text-only posts as their relevance is harder to interpret than visual posts when a user is scrolling rapidly through a social media feed.

Materials and Methods

Facebook Insights is a free online tool within the Facebook platform which provides a record of social media engagement on a Facebook page to the page’s owner(s).(12) Using Facebook Insights we collected engagement data for posts made by the MRINZ over a five year retrospective period from January 2016 to January 2021. Posts were excluded if their reach had been boosted through paid (non-organic) distribution, or if no data was available (e.g. if the post and associated statistics had been deleted). The ‘media type’ of each post was classified as ‘text’, ‘image’, or ‘video’ to allow comparison of performance by media type. Media subcategories were formed to break this down further into ‘text’, ‘graphic design’, ‘live video’, ‘animation’, ‘photo’, ‘data visualisation’, ‘editorial/medical illustration’, and ‘comics’ by a medical illustrator to allow for a more granular analysis of performance. Engagement is the sum of interactions by users with a post, such as likes, comments, reactions, shares, and link clicks. Engagement with media subcategories was our primary outcome of interest. Reach is an estimate of the number of unique people who saw a piece of content at least once; reach of each media subcategory was our key secondary outcome of interest. Other available data was reviewed including whether a post was original or shared content, whether it contained external links, link previews, contained professionally-created custom visuals, and for the subject matter of COVID-19.

Statistical analyses

Simple t-tests and simple linear models were used to estimate the difference between categorical variables of interest. Where necessary, Tukey tests were used for pairwise comparisons (to adjust for multiplicity). For the COVID-19 sub-analysis, linear models with COVID-19 as a categorical co-variate were used to estimate the difference between categorical variables of interest.

Results

A break-down of post inclusion is shown in Figure 3. 249 posts were included in the primary analysis of engagement, and 115 in the secondary analysis of reach (the count for which data was available).

Figure 3: CONSORT diagram for posts included in the retrospective analysis.

Figure 4: A - Facebook post reach by main media categories of text (N=9), image (N=94), and video (N=12). B - Facebook post engagement by main media categories of text (N=47), image (N=190), and video (N=12).

Analysis of reach and engagement

There was a trend for posts with images outperforming those with only text content (Figure 4) in terms of reach and engagement. In Tukey comparisons, differences in mean reach were not significant. In differences of mean engagement, only the 5.2-fold greater engagement with posts containing images versus text alone was significant (109 vs 21, difference 88.3, 95% CI 12.3 to 164.3, p=0.02, Figure 4B).

There were marked differences in the performance of text, images, and video became more pronounced when comparing media subcategories (Figure 5). In Tukey comparison of mean reach, significant findings included comics outperformance of animation (2.6 fold, difference 1992.3, 95% CI 110.3 to 3874.3, p=0.03), text (2.8 fold, difference 2088.1, 95% CI 206.1 to 3970.1, p=0.02), graphic design (2.7 fold, difference 2044.7, 95% CI 279.6 to 3809.8, p=0.01), and photos (1.9 fold, difference 1518.9, 95% CI 192.7 to 2845.2, p=0.01). Comics also outperformed other media categories in terms of engagement; this was significant compared to posts containing text only (15.1 fold, difference 297.8, 95% CI 161.7 to 433.9, p<0.0001), graphic design (13.8 fold, difference 295.8, 95% CI 159.2 to 432.4, p<0.0001), animation (4.0 fold, difference 238.6, 95% CI 30.7 to 446.6, p=0.01) and photographs (3.8 fold, difference 235.6, 95% CI 110.2 to 361.1, p<0.0001). The increased engagement seen with comics compared to live video and data visualisation was not statistically significant in Tukey comparisons.

Figure 5: A – Facebook post reach by visual media sub-category.

B - Facebook post engagement by visual media sub-category

Secondary analyses of other contributing factors

Illustrating posts with custom professional art

At the mid-point of the period of analysis, the MRINZ employed a medical illustrator who supported social media activity by designing custom visual content for many posts, and directed a more visually stimulating presence on Facebook. The mean engagement per post was 15.7 times greater (difference 170.0, 95% CI 124.6 to 215.5, p<0.0001, Figure 6) in the 30 month period following this change compared to the 30 month period before this additional support.

Figure 6: Average Facebook post engagement in the 30 month period prior to a medical illustrators involvement, compared to the 30 month period with their involvement.

Engagement with posts linking to clinical research publications, by post media (N=171)

Original content

Posts with original content by the MRINZ (N=202) had 4.9 times greater mean engagement than content that had been shared from other pages on Facebook (N=47, mean difference in engagement 85.1, 95% CI 22.1 to 148.2, p=0.008).

Figure 7: Link clicks on Facebook posts directing to formal research publications, by post media.

Including external links to clinical research publications

Figure 7 presents mean engagement on Facebook posts which featured direct links to medical research publications on journal websites (N=49). Although there appears to be a similar trend to that of engagement, these apparent differences were not significant in this small data subset.

Subject matter

In the 10-month period of posts in which COVID-19 was sometimes a subject; posts with COVID-19 subject matter had 5.4 times greater mean engagement than posts without COVID-19 as a subject matter (difference 302.3, 95% CI 144.1 to 460.4, p=0.0003).

Discussion

Primary outcomes

The main findings of this analysis were that there was moderate evidence of greater engagement with images compared to text, and in the media subcategories there was moderate evidence that comics had greater reach compared to animation and text, and strong evidence that comics had greater reach compared to graphic design and photographs. Further, there was strong evidence of greater engagement with comics compared to animations, editorials, graphic design, photos, and text media subcategories. Social media posts that included comics therefore enhanced both the mean reach (how many saw) and mean engagement (the proportion of those who interacted) with that content. There was no evidence of different reach for text, image, and video posts overall. Much of the public engagement literature published describes the intent of efforts and subjective contemplation of value rather than evaluation on objective metrics. Our review contributes objective data to the evidence base that helps to address this gap, and showed that visual media, and particularly comics, can contribute to public engagement with healthcare research discourse on social media.

This is in keeping with two previous studies by members of our team which demonstrated a role for comics and illustration in engaging the New Zealand public with clinical research on social media in study recruitment contexts.(12,13) One study demonstrated successful engagement on a local level of both a general adult audience, and a specific patient group with asthma. This facilitated rapid, cost-effective, and more equitable access to research participation than was typical in the MRINZ’s 20-year experience in clinical trials.(13) In a second study exploring two approaches to a national-scale online public health survey, we investigated the effect of comics and cartoon illustrations.(14) The illustrated version of the survey had greater engagement on social media and resulted in greater diversity in respondent age, gender, and ethnicity.(14) This new retrospective analysis adds further insight into the role of different types of visual media on engagement with clinical research online, and was strongly supportive of comics, a medium which may not be traditionally associated with clinical research.

What is different about comics as a visual medium?

Researchers may be surprised that comics outperformed other mediums more commonly employed by experts in visual science communication, such as editorial/medical illustration, graphic design, photography, and data visualisation. However, it is less surprising to those familiar with the field of Graphic Medicine - the intersection of comics and medicine – which has notably been embraced by science communicators to great effect throughout the COVID-19 pandemic.(15,16) In addition to their more usual channels online and in print; comics have been increasingly harnessed in more traditional ‘formal’ platforms such as government address and messaging in public health communication (16), journalism (17), and medical research publications (18–20). This adoption in a time of global emergency has been validating, inspiring, and challenges antiquated notions of a limited demographic appeal or subject matter for the medium.

Comics combine images and text in sequence to present narratives, offering a distinctly different medium of communication than these constituent parts in isolation. They are processed differently to other visual media (such as video) as the reader plays an active role in the experience. Readers control the pace of content delivery by their reading speed; must imagine what happens in the gutter (the blank space between panels) in order to make sense of the narrative; and determine the ‘voice(s)’ of characters, sound effects, and narration. Interpreting meaning from visuals may involve lateral thinking with the use of visual metaphors and conflicting text and imagery, two examples of techniques commonly employed in comics to emphasise a point. Reading comics therefore demands ‘active’ attention unlike many other media forms, and offers a slightly personalised experience for each reader.

Clinical research topics are often niche areas of science which are unfamiliar to most people, and so may be considered less approachable for a general audience in their typical presentation for a scientific audience. As a widely-consumed entertainment medium, comics may be viewed as inherently more approachable, and facilitate greater access to science-related subjects. The structure of comics can support science communication in many ways, such as pacing information delivery over a series of panels allowing ideas to be built up in a series of manageable steps. Comics are being increasingly embraced to make medical subjects more widely accessible, for example to share pandemic science, mental health, and health journalism.(17,21–23) Their application may extend beyond the informative to clinical practice too; Delp and Jones conducted a randomised controlled trial in the emergency department and found that patients who received cartoon instructions for wound care had greater recall, understanding of, and compliance with these at 3 days post discharge, compared to those who did not receive cartoon instructions for wound care.(24) This suggests a clinical role for the medium beyond engaging people with information. Adherence with care is a major rate-limiting factor of healthcare interventions in all specialities, and critical in clinical research to determine if interventions are the cause of differences between groups studied. Having established evidence that comics can support public engagement with health research; we wonder if they may also be able to support adherence to care in both research settings and clinical practice.

The effect of involving science communication professionals

The rate that new social media content is added to platforms each day has demanded that social media platforms filter a more limited selection of content for each viewer. The content and order of presentation that is seen is determined by social media algorithms. Facebook has made several changes in recent years to improve the experience for users, such as attempting to filter out low-quality clickbait, political extremism, and misinformation about COVID-19.(10) In 2018, a study of Facebook posts over the previous 18 months showed a 50% drop in engagement, thought due in part to a ‘meaningful interactions’ update to the algorithm which Facebook reported was intended to promote content from friends and family over that of businesses, and content that sparked meaningful interactions such as conversation through comments on a post.(25) In a time period marked for many Facebook pages by diminishing organic reach and engagement efforts, we note that our engagement was significantly higher in the latter 30 month period of study. This coincided with a significant change in the MRINZ’s approach to social media by involving a medical illustrator in custom media creation for posts (Figure 4). Our analysis showed strong evidence of higher engagement on original post content compared to that of shared posts from other pages. This emphasises the importance of developing or acquiring science communication capability within a research team or organisation. Further evidence to this point included strong evidence of higher engagement on posts containing custom visuals created by a professional (which included posts not created by the MRINZ’s medical illustrator) compared to those with visuals that were not created by a professional.

Wrangling changing algorithms

Our finding that images were the most engaging media type contrasts with the intent of updates to the Facebook algorithm to favour video content which were introduced in 2016, 2017, and 2019.(10) Our analysis also contrasts with findings reported by Facebook that images with large amounts of text perform worse, and active measures made in recent years to reduce the distribution of images that contain more than 20% text, notably reducing the distribution of Facebook and Instagram ads violating this rule.(26) Combining text and images is an integral component (although not mandatory) of the medium of comics, and certainly a key feature of all of the comics in the posts analysed in this study. The authors hypothesise that if there was any penalty for this in the period of analysis, it might have been compensated for by other aspects of the Facebook algorithm. For example, if emotional resonance with humour and narratives led to more engagement in the form of ‘reactions’ (such as ‘love’ or ‘care’) which were weighted more than ‘likes’ in a 2017 algorithm update. Exploring ‘why’ certain comics resonate, and with whom, would be an interesting area for further exploration.

Is there a role for comics in formal research dissemination?

Our data suggests that amongst posts containing direct links to publications, those using visuals had increased engagement, although this observation was not significant in our small sub-sample of posts. Given the moderate and strong evidence from our analysis that comics increased reach and engagement of posts overall and compared to most of the media subcategories; it would be of interest to establish with a larger sample whether they can also support other aspects of clinical research such as the dissemination and impact of publications, recruitment to studies, and reaching underserved demographics.

Limitations

The Facebook Insights metrics used represent posts made over five years by a Facebook Page of a single clinical research institute in New Zealand, across a range of research programmes in different clinical specialities. Most of the subcategories of visuals analysed in the latter 30 month period of the study were created by a single medical illustrator and so likely are influenced by their particular skills in certain categories of image creation. We expect that other creators with greater expertise in different media subcategories might see engagement skewed towards their expertise. As such, the data presented should be considered more as evidence of what can be achieved with certain media, and may only represent NZ public preferences. The approach of a single institute may not translate readily to others, and the appeal of different subjects and visual art styles may be widely varying in a public audience. Medium subcategories were subjectively assigned by a single professional medical illustrator with experience creating in each of these approaches; others may have categorised differently.

As a retrospective review of real-world Facebook post performance there was no randomisation or controls for each post made that might enable direct comparisons of different approaches to sharing material. Followers of the MRINZ’s Facebook page are more likely to see new posts from the MRINZ in their feed and therefore there is an element of self-selection of an unknown portion of the audience viewing content from the page. We also note that not all content posted to a Facebook page will be seen by all of those who follow it, and that we are unable to determine the proportion of engagement from followers versus those who do not have an existing connection to the page. These uncertainties may limit generalisability to a public audience with no prior connection to a Page.

In the 10-month sub-analysis of posts at the end of the study period, there was strong evidence that posts with COVID-19 as a subject matter had greater engagement compared to those in which COVID-19 was not the subject matter. This illustrates how contemporary global events can have a large influence on social media post engagement and may conflate findings in small samples of posts. There are likely other variables that influenced the reach and engagement of our posts which are not accounted for in the analysis, such as who shared posts and where (e.g. to a personal page of a person with a small following versus to a large and active public group); the written content of posts such the use of hashtags and tagging other accounts; and the time of posting relative to audience online activity. The variables which matter likely change over time with audience interests and evolution of the Facebook algorithm. An awareness of such factors is important for those involved in science communication on social media. We could not determine the audience characteristics who were engaging with each post, but have included Facebook’s own estimates of the Page’s overall audience characteristics in the supplementary material. This suggests a greater proportion of female than male audience mainly in the 18-44 years age range, but we note this is an estimate, and represents an unknown proportion of the overall audience engaging, limiting utility. Similarly, the number of Facebook page ‘Likes’, and the more recently initiated ‘Followers’ over time was not considered in this analysis given the uncertainty for how these measurements have interacted with the reach and engagement of a post over time as the algorithm has changed. There is data to suggest dramatic falloff in the relationship between these measures of audience size, reach, and engagement, during the period of study, which are thought to be related to algorithm changes to encourage paying to advertise one’s posts to reach one’s own audience.(25) Additionally, data on these metrics was only available for a portion of the studied period, in which time the MRINZ’s Page following changed minimally. We note that individual post performance metrics were taken at different temporal distance from the posts with the more recent having had less time to accumulate engagement. A minimum period of ‘one month from posting date’ was used for all posts included. In our experience content engagement drops off exponentially within a few days of a post, and so any effect of this is expected to be low and predominantly affect the most recent posts. In our case these were the better-performing despite this theoretical disadvantage (Figure 6).

Lastly, our analysis does not consider how well the intended messages were communicated or the perspectives of those engaging. Although this was practical for a large retrospective review, these may be important considerations when attempting to measure the impact of public engagement. Future research could specifically be designed to explore such areas.

Conclusion

The inclusion of visual media in Facebook posts by a clinical research institute increased their reach to and engagement of a public audience online. We present medium and strong evidence that this effect was most pronounced for comics compared to other visual media types. There was strong evidence of increased public engagement with health research on social media with involvement of a medical illustrator creating tailored visual content to support these messages. These findings add evidence to Graphic Medicine, Medical Illustration, and public engagement literature with a 5-year data set exploring several visual mediums. As visuals may easily be shared in a similar way across many social media platforms, these findings may have wider implications for public engagement online. Our analysis indicates that public engagement with creative social media could potentially be harnessed to promote publication readership, study recruitment, and other aspects of clinical research, but these require further data to investigate. The metrics evaluated here are freely and easily accessible on Facebook, meaning this approach could readily be applied by other researchers to measure the impact of their public engagement efforts, informing strategies and resource allocation.

Acknowledgements

The authors wish to thank the New Zealand public for their enthusiastic engagement with the MRINZ in our clinical research and discourse online via social media. This helps to make our clinical research possible, address questions that matter, and translate findings into practice.

Declaration of conflicts of interest

RB reports grants from Health Research Council of New Zealand. The authors declare no competing interests.

Funding

The Medical Research Institute of New Zealand receives Independent Research Organisation funding from the Health Research Council of New Zealand (IRO grant [18/002]).

References

1. Alamri A, Rogers P, Kearns C, Doke T, Al-Habib A, Servadei F, et al. Social media for dissemination and public engagement in neurosurgery-the example of Brainbook. Acta Neurochir (Wien). 2019 Jan;161(1):5–9.

2. Banner D, Bains M, Carroll S, Kandola DK, Rolfe DE, Wong C, et al. Patient and Public Engagement in Integrated Knowledge Translation Research: Are we there yet? Res Involv Engagem [Internet]. 2019 Dec [cited 2019 Dec 2];5(1). Available from: https://researchinvolvement.biomedcentral.com/articles/10.1186/s40900-019-0139-1

3. Elsabbagh M, Yusuf A, Prasanna S, Shikako-Thomas K, Ruff CA, Fehlings MG. Community engagement and knowledge translation: Progress and challenge in autism research. Autism. 2014 Oct;18(7):771–81.

4. Holzer JK, Ellis L, Merritt MW. Why We Need Community Engagement in Medical Research: J Investig Med. 2014 Aug;62(6):851–5.

5. UyBico SJ, Pavel S, Gross CP. Recruiting Vulnerable Populations into Research: A Systematic Review of Recruitment Interventions. J Gen Intern Med. 2007 May 10;22(6):852–63.

6. Morris ZS, Wooding S, Grant J. The answer is 17 years, what is the question: understanding time lags in translational research. J R Soc Med. 2011 Dec;104(12):510–20.

7. Wellcome Trust. Public engagement support for researchers [Internet]. Available from: https://wellcome.ac.uk/what-we-do/our-work/public-engagement-support-researchers

8. Royal Society of New Zealand. Public engagement guidelines for researchers, scholars and scientists [Internet]. 2016. Available from: https://www.royalsociety.org.nz/what-we-do/research-practice/public-engagement-guidelines/public-engagement-guidelines-for-researchers-scholars-and-scientists/

9. Health Research Council of New Zealand, Ministry of Health, Ministry of Business, Innovation & Employment. The New Zealand Health Research Prioritisation Framework: Maximising the benefits of health research for New Zealanders [Internet]. Health Research Council of New Zealand, Ministry of Health, Ministry of Business, Innovation & Employment; 2019 [cited 2021 Feb 18]. Available from: https://www.hrc.govt.nz/resources/new-zealand-health-research-prioritisation-framework

10. Cooper P. How the Facebook Algorithm Works in 2021 and How to Make it Work for You [Internet]. Hootsuite Blog. 2021 [cited 2021 Feb 18]. Available from: https://blog.hootsuite.com/facebook-algorithm/

11. Videos vs. Images: Which Drives More Engagement in Facebook Ads? [Internet]. Databox. 2021 [cited 2021 Mar 15]. Available from: https://databox.com/videos-vs-images-in-facebook-ads

12. Facebook. Facebook IQ: Audience Insights [Internet]. Facebook. Available from: https://en-gb.facebook.com/business/insights/tools/audience-insights

13. Kearns C, Baggott C, Harwood M, Reid A, Fingleton J, Levack W, et al. Engaging Māori with qualitative healthcare research using an animated comic. Health Promot Int. 2020 Dec 10;daaa111.

14. Kearns C, Kearns N, Braithwaite I, Shortt N, Eathorne A, Semprini A, et al. Using comics and curiosity to drive pandemic research on a national scale. J Vis Commun Med. 2020 Nov 18;1–11.

15. Czerwiec MK, Williams I, Squier SM, Green MJ, Myers KR, Smith ST. Graphic medicine manifesto. University Park, Pennsylvania: The Pennsylvania State University Press; 2015. 195 p. (Graphic medicine).

16. Kearns C, Kearns N. The role of comics in public health communication during the COVID-19 pandemic. J Vis Commun Med. 2020 Jul 9;1–11.

17. Taylor W, Shwed A. COVID-19 Myths, Debunked. The NIB [Internet]. 2020 Mar 22 [cited 2020 Apr 18]; Available from: https://thenib.com/covid-19-myths-debunked/

18. Green MJ, Wall S. Graphic Medicine—The Best of 2020. JAMA [Internet]. 2020 Nov 30 [cited 2020 Dec 2]; Available from: https://jamanetwork.com/journals/jama/fullarticle/2773653

19. Callender B, Obuobi S, Czerwiec MK, Williams I. COVID-19, comics, and the visual culture of contagion. The Lancet. 2020 Oct;396(10257):1061–3.

20. Kearns C, Fisher D, Chong YS. The infective nurture of pandemic comics. The Lancet. 2020 Dec;S0140673620325502.

21. The COVID-19 Chronicles [Internet]. National University Singapore: Yong Loo Lin School of Medicine. 2020 [cited 2020 Oct 13]. Available from: https://medicine.nus.edu.sg/the-covid-19-chronicles/

22. Noe MN, Levin LL. Mapping the Use of Comics in Health Education: A Scoping Review of the Graphic Medicine Literature. Graph Med [Internet]. 2020 [cited 2020 Jul 25]; Available from: https://www.graphicmedicine.org/mapping-comics-health-education/

23. Bean C, Sturm J, Allie I. Let’s talk about it: a graphic guide to mental health [Internet]. 1st ed. Vermont: The Center for Cartoon Studies; 2020 [cited 2021 Mar 15]. 24 p. Available from: https://www.cartoonstudies.org/cartooningprojects/mentalhealth/

24. Delp C, Jones J. Communicating Information to Patients: The Use of Cartoon Illustrations to Improve Comprehension of Instructions. Acad Emerg Med. 1996 Mar;3(3):264–70.

25. Erskine R. Facebook Engagement Sharply Drops 50% Over Last 18 Months [Internet]. Forbes. 2018 [cited 2021 Feb 18]. Available from: https://www.forbes.com/sites/ryanerskine/2018/08/13/study-facebook-engagement-sharply-drops-50-over-last-18-months/

26. Facebook. About Text in Ad Images [Internet]. Business Help Center. [cited 2021 Feb 18]. Available from: https://www.facebook.com/business/help/980593475366490