Social support and positivity: Analyzing user-generated comments on the Instagram pages of two Brazilian cancer hospitals

Online social media platforms are an important tool for individuals who wish to exchange ideas, stories, feelings, information. Hospitals and public health organizations can use social media platforms, including Instagram, to engage, interact or communicate with patients, family members and anyone interested in health information or new discoveries. The purpose of this study is to understand the main characteristics of user-generated digital narratives on the Instagram pages of two Brazilian cancer hospitals. The study fills a gap in the literature on cancer communication by studying a language other than English in a Latin American context, and examining user-generated comments over a period of more than 6 years. The findings of the authors’ content analysi s show a preponderance of positive and social support messages, even though cancer is a disease laden with negative stigma. Religiosity, which can strengthen supportive social ties and create a stronger sense of empowerment and control, was also a hallmark of these narratives. In the content analysis of Brazilian cancer hospitals’ Instagram accounts, the predominant characteristic of user comments was that of positivity, even though cancer is a disease laden with negative stigma.


INTRODUCTION
Online social media platforms are an important tool for individuals who wish to exchange ideas, stories, feelings, and information. Social media usage is one of the most popular online activities. In 2020, an estimated 3.6 billion people were using social media worldwide, a number that is projected to increase to almost 4.41 billion in 2025 (Statista, 2020).
There is a growing use of social media and its extension to health promotion by services and individuals (Dadich and Khan, 2020;Hunt, 2015). Instagram is a mobile social media service for photo and video sharing (Forsey, 2020). On Instagram, users can share their opinions with others by liking or sharing posts or by writing comments underneath original posts (Chun and *Corresponding author. E-mail: cjurberg@bioqmed.ufrj There is no doubt that while people use social networks to exchange information on various themes, one of the most actively discussed topics is health (Gill et al., 2013). Studies observed that adults search for health information online and then speak with their primary caregiver about what they have read. Marrapondi (2017) states that: "There is an entirely unknown and unseen community seeking health information online beyond our primary users. Unfortunately, much of this information comes from unauthenticated, unreliable sources. This information is oftentimes erroneous or misleading, which can lead to wrong self-diagnosis and ineffectual or damaging treatment attempts by the layperson." Therefore, it is important to understand health information on social media and what kind of public engagement they produce, in order to reflect and propose effective communication against misinformation, wrong diagnosis or erroneous/late treatments. The present study examines users" comments on the Instagram accounts of two Brazilian hospitals dedicated to cancer treatment. The purpose was to understand the main characteristics of digital narratives posted by users, and their motivations for doing so. The authors decided to focus their study on Instagram because this is one of the most widely used social networks in the world, with 1,22 billion active users as of January 2021 (Statista, 2021a). Biancovilli et al. 45 Brazil has 69 million active users on Instagram, the third largest number of any country in the world (Salles, 2020). Nevertheless, the authors found a lack of studies dedicated to users who publish comments on cancer profiles on this platform. This study fills a gap in the literature on cancer communication by analyzing a language other than English in a Latin American context, and by examining user-generated comments over a period of more than 6 years in a growing social media. Studies that have documented cancer-related social media posts (Basch and Maclean, 2019a, b) highlighted the strong occurrence of content related to social support, discussions on treatment, and individual stories. However, some studies on the collective imagination in relation to cancer indicate that the population associates the disease with mostly negative feelings or events (Goldenberg and Arndt, 2008;INCA, 2007;Machado et al., 2017;Niederdeppe and Levy, 2007). Other studies also described the common association of cancer with war metaphors, such as a person"s "battle" or "fight" against cancer, which may suggest that cancer treatment is always painful, scary and difficult (Grant and Hundley, 2009;Hauser and Schwarz, 2020;Maganã and Matlock, 2018;Sontag, 1991). A Brazilian survey called "Conceptions of Brazilians on Cancer" (INCA, 2007) reveals that 70% of respondents associated cancer with death, pain and suffering. For this reason, the hypothesis of the present study is that the digital narratives of users who comment about cancer on Instagram profiles reflect society"s predominant perception about the disease. Our research questions were: Q1: What are the more frequent kinds of narrative amongst users who comment on Brazilian cancer hospital pages? Q2: Regardless of the institutional Instagram account, do users" narratives always follow the same approach?
According to Tifentale and Manovich (2015), Instagram users "construct their identities and simultaneously express their belonging to a certain community. Thus, performing the self is at once a private act as well as a communal and public activity." As Kristensen and Prigge (2018) recognize, "the construction of a "data self" entails a new conception of lived experience, where the self is known through data, and the data simultaneously inform the self."

MATERIALS AND METHODS
This exploratory analysis aims to understand the digital narrative of Instagram users on the profiles of two Brazilian hospitals dedicated to cancer treatment: AC Camargo Cancer Center and the Hospital de Amor (in English, Love Hospital). These profiles were chosen over others because they are the most frequently updated among the Brazilian cancer hospitals" profiles that was found. In addition, both hospitals are internationally recognized for excellence in the treatment of cancer patients, diagnosis, teaching, and research on cancer (A.C. Camargo Cancer Center, 2020; Hospital de Amor, 2020). AC Camargo Cancer Center is located in São Paulo, the largest city of Brazil, with more 20 million inhabitants in the metropolitan area. This hospital was founded in 1953 and provides private and public health care, the latter financed by the Brazilian public health system (the Unified Health System, or SUS). It has the most important private cancer research center in the country (A.C. Camargo Cancer Center, 2020). Hospital de Amor was founded in 1962 and is located in the city of Barretos. It operates exclusively through the SUS, which means that all patients are offered free healthcare (Hospital de Amor, 2020). Both institutions accept donations to finance their activities.
In this study, the authors were especially interested in examining the narratives of users that addressed cancer. To this end, they monitored public engagement of the Instagram profiles of the two institutions. They started their investigation from the first post posted on each of the pages on Instagram and completed the collection of the corpus in March 2020. The AC Camargo Cancer Center profile was launched on October 28, 2013 and the Hospital de Amor profile was launched on June 7, 2014. The analysis of users" digital narratives on Instagram was based on the methodology proposed by Laurence Bardin (2011) for psychosociological investigations and for the study of mass communications. For Bardin (2011), pre-analysis means "a period of intuition, in which they seek to systematize the initial ideas that will lead them to a more precise scheme for sequential operations." At this stage, the objectives and questions of the research are defined, and the universe of the study is delineated, including the material that will be examined, the hypotheses, and the indicators that lead to the final interpretation (Bardin, 2011). This universe is called the corpus. In the present study, this comprises the comments of users on the Instagram profiles of the aforementioned Brazilian hospitals. As the comments are public and we suppressed their identification, preserving their privacy, there is no need for approval from ethics committees.
The comments posted under all the photos and videos on the two pages were manually captured and saved. The authors" preanalysis stage included the systematic reading of the corpus. Then, two raters separately codified the material after observing its representativeness, looking for homogeneity and relevance. This phase consisted of a transformation of the raw data into categories created to define the digital narratives. The two researchers took charge of viewing the corpus and discussing the categories used to classify each unit of content. A word was only considered as part of a category if it appeared at least 20 times. This number seemed ideal because they had to define a cut for this sample, which would make content analysis possible; the authors believe that this cut help them to better organize the categories and thus have a comprehensive overview of the conversations and narratives about cancer on the selected Instagram accounts. With the initial units, the quantitative and qualitative analyses (Minayo and Sanches, 1993) were automated with MAXQDA 2018 (Verbi Software, 2017); a software program designed for qualitative and mixed methods data. After that, the content was interpreted by identifying the presence, frequency of appearance, and meaning of the content units.
Subsequently, the authors looked for the common characteristics of the comments that would help them understand the nature of the users" engagement in terms of the previously established categories. Their aim was to understand the dynamics of public engagement, the types of narratives contained within the comments, and to seek out patterns. Online comments can be used to explore the public"s word-of-mouth communication, which refers to comments that publics make about an issue; this plays a significant role in understanding these publics" perception on a topic (Chun and Lee, 2017). With this investigation, the aimed was to collaborate with other professionals dedicated to cancer communication to explore, in a more rational and assertive way, the use of social media to transmit relevant health information.
Understanding what and how users communicate and what sentiments they express can be useful in the development and management of health communication strategies.

Quantitative data
The institutional profiles of AC Camargo Cancer Center and the Hospital de Amor had their own characteristics with regard to the frequency of posts and the different user approaches (Oliveira et al., 2019). AC Camargo Cancer Center has 91,900 followers, follows 80 other institutions (as of October 2021) and publishes 5.6 new posts a week, on average. Hospital de Amor has 182,000 followers, follows 357 other institutions (as of October 2021) and publishes 7.75 new posts a week, on average. On the AC Camargo Cancer Center page, we identified 806 users who made comments in the 747 posts of our sample; on the Hospital de Amor page, 2,469 users made comments in the 2,641 posts of their sample. It is important to mention that comments can be made by any Instagram users, regardless of whether they follow the pages or not. It was observed that 29 out of a total of 3,246 individuals commented on the Instagram pages of both institutions.
Based on 12 categories with 113 selected words (Table  1), 5,069 comments, from the total of 6,793 comments, were categorized because they contained one or more of the 113 words selected: 4,117 on the Hospital de Amor page and 952 on AC Camargo Cancer Center. Of these, 6,536 valid words (after excluding names, emoticons, and profiles) were used to build the narratives. As has been pointed out, the selected words had to appear at least 20 times in both profiles to be included in one of the categories.
It is important to note that the categories "War" and "Negative" had few words in each because other negative words did not achieve the minimum frequency. The authors included them because of their interest in comparing the present analysis with previous studies that have shown the frequent association of cancer with pain, suffering, fatality, and death (Goldenberg and Arndt, 2008;INCA, 2007;Machado et al., 2017, Niederdeppe andLevy, 2007), and the association of cancer with war metaphors (Grant and Hundley, 2009;Hauser and Schwarz, 2020;Maganã and Matlock, 2018;Sontag, 1991).
From this initial categorization, MAXQDA software was used to assist in both quantitative and qualitative analyses. The software performed 8,928 categorizations on 5,069 comments identified with the selected words, representing 75% of all comments. The remaining 25% were not categorized, because as they did not contain words relevant to the study.
Of all comments, it was possible to identify 45 strictly identical paragraphs that made them suspect they were bots. Bots are automated or semi-automated accounts designed to impersonate humans, with the aim of manipulating online conversations and opinions (Yang et al., 2019). In respect of Q1, the most frequent words within each category were "congratulations" within the category "Congratulations," with 1,643 entrances (14% of the 113 selected words); "God" was second in the ranking with 1,334 insertions (12% of the 113 selected words). It is worth pointing out that 80% of the comments categorized as "Acknowledgements" (Table 1) were expressed by users who identified themselves as women. This is because, in Portuguese, the expression 'thank you' is written differently according to the gender with which the person identifies; men say "obrigado' or "grato" and women say 'obrigada' or "grata".
Of the 8,928 categorizations, 9.2% (818) referred to the narratives of only two followers of the Hospital de Amor who routinely published comments in the analyzed corpus. In view of this, the results of the analyses considered both the comments of the "top 2," so identified by us, and their exclusion. Our intention was to assess whether the frequency of comments from the "top 2" interfered or not in our corpus (Table 2).
To answer Q2, Table 2 shows the ranking of the most frequent categories in each of the profiles on Instagram. It is important to note that the most prevalent category on the Hospital de Amor page was "Praise," and on the AC Camargo Cancer Center "Human beings," identified here as hospital professionals, companions, patients, volunteers, collaborators or so-called "ambassadors" of the institution, usually famous artists and musicians.
In the ranking of the categories, it is clear that the Hospital de Amor page was more influential in the general ranking of the categories, because it had more followers and more comments. The authors tried to find out if there would be a more frequent crossing between the categories chosen for the present study. In Table 3, it was seen that in the Hospital de Amor, users in more than 10% of their comments associate words (Table 1) of "Congratulations" (hug, congratulations, and so on) with religious words (blessed, amen, angel, blessings, God, faith, thanks, Jesus, light, and sir). On the AC Camargo Cancer Center page, users usually used praise (excellent, wonderful, and so on) associated with human beings (doctors, nurses, employees, and doctors). It is worth mentioning that the combination "Congratulations" and "Religiosity" without the "top 2" would have fallen two positions in the ranking of crossed categories (from the first position to third position).

Qualitative data
In the qualitative analysis, the authors sought to identify curious aspects of the digital narratives posted by users. Initially, they noticed a high frequency of two sentences on the Hospital de Amor page. These were "Congratulations, God bless you" and similar variations, such as "Congratulations, God bless you all," and "Congratulations, God bless you always." The use of the words "congratulations" and "God" was boosted through the comments of the "top 2" on the Hospital de Amor page, increasing the frequency of "Congratulations" and "Religiosity" categories. It is important to clarify that of the 174 comments containing "Congratulations, God bless you," 173 were made by the "top 2." Another occurrence that deserves to be highlighted was "Big fan" and the variation "I"m a fan," which were categorized under "Praise." In this case, the expression was used mostly by the "top 2" on the Hospital de Amor page. On the AC Camargo Cancer Center page, the most common term was "Congratulations!" with variations that were with or without the exclamation, and encompassing the word "all." The religious aspect hardly appeared here. The frequency of negative words such as "crying" and "pain" was low, with less than five mentions. Because they did not reach the minimum number of citations, they were not included in the categorization. Examples of negative sentences can be seen in the comments below: "My mother is a hospital patient and I am going to ask these questions of the doctors. I always look for information about this sad disease. Have a great day and thank you very much for the answer, A big hug" (Hospital de Amor).

"Very sad, a disease that badly mistreated my husband. He died of lung cancer, which is sad" (AC Camargo Cancer Center).
War-related expressions, such as "fighting," "war," and "warrior" was also almost non-existent. The word "combat" occurred only twice. "Warrior" was mentioned 11 times. The term "battle" appeared five times in the 6,793 comments analyzed. Only the words "save" and "fight" were recorded more than 20 times. The following selected comments include some of these terms: "Congratulations to these real heroes, who use their knowledge and dedication to save lives at all layers of society and offer the people the miracle that is asked of God, carried out by these angels sent to Earth to save those who God determines!" (AC Camargo Cancer Center) "Only together can we offer a better quality of life and fight the high mortality rate caused by this disease that can be cured, YES!" (Hospital de Amor) The categories "War" and "Negative" represented 0.65% and 0.35% of the total, respectively. But while there were few negative comments, the word "cure" and its variations, within the "Health" category, were also rarely mentioned. Two examples of a hopeful healing narrative can be seen in the comments below. The first associates religious love with human beings, while the second one links these terms to a congratulatory message to São Paulo State Oncocenter Foundation (FOSP) on its 40th anniversary: "May God bless this beautiful work! And that the people who seek you are healed, because they are filled with love. God be present in every moment!" (Hospital de Amor).
"Congratulations on 61 years of a tireless and persevering fight against cancer. I take the opportunity to inform you that FOSP turned 40 today, April 25, 2014. We, AC Camargo and FOSP, are happy to be allies in the healing and rehabilitation of patients affected by head and neck cancer." (AC Camargo Cancer Center).

DISCUSSION
By analyzing users" digital narratives in a qualitative and quantitative way, the authors observed that the four most frequent categories were the same for Hospital do Amor and AC Camargo Cancer Center pages: praise, congratulation, religiosity, and human beings. Thus, it can be said that the users" narratives were not substantially different between the two. These categories were present mostly in comments which expressed positive sentiments (as observed in the list of words for each category and in the most frequent combinations of categories). Some studies have pointed to a high semantic association of the term cancer with words such as pain, suffering, fatality, and death. Battle metaphors increase fatalistic beliefs about cancer prevention and discourage people to immediately see their doctor when imagining a cancer scare (Hauser and Schwarz, 2020;Niederdeppe and Levy, 2007). They did not find the same patterns in their sample. Indeed, there was a preponderance of positive messages, as has been observed in previous research. This may be due to the fact that, on Instagram and in other social media, publications with positive content tend to generate greater user engagement (Casaló et al., 2020;Picanço et al., 2018), both in health promotion and in other areas. For this reason, they believe that the prevalence of positive sentiments in our study is mainly the result of the type of content published by the two hospitals, although the analysis of those publications is not within the scope of our study. In a study with 60 Latino men and women in the United States diagnosed with cancer, it was observed the frequent development of coping strategies that involved positive reframing, which means trying to find positive emotions that help dealing with cancer diagnosis (Carrion et al., 2017), and social media also play an important role in this regard.
An analysis of public dialogues on cervical and breast cancer screening on Twitter, a microblogging and social networking service with 290 million active users worldwide (Statista, 2021b), shows that comments about mammogram are more positive than others, especially when they mention support for friends or report noncancerous screening results (Lyles et al., 2013). Social media is a space used to garner support and raise awareness about certain conditions, counter stigma, and reduce discrimination (Allen et al., 2020;Betton et al., 2015;Naslund et al., 2016).
Previous research concluded that reading others" comments on social media is as an important factor for online social support (Chun and Lee, 2017;Rozzell et al., 2014).
When the authors analyzed the combinations of categories present in the same comment, they saw the strong presence of the category "praise." Amongst the ten most common category combinations, it was present in six. The most common combination (with the "top 2") was "congratulations" + "religion" (n = 390). They believe that the expression of praise, congratulation, and feelings related to religiosity is a way for users to sympathize with doctors and hospital staff, and patients who are undergoing cancer treatment. Some of the authors of these comments seemed to be patients, family members, or friends of patients, as the content of the texts indicates. Even so, some users (n = 29) commented on both pages, which led us to conclude that cancer is a topic that arouses the interest of people regardless of any personal link with the hospitals. Among the motivations that lead a person to publish content on social networks are the desire to maintain connections and friendships, the desire to share their impressions of something on the internet and the longing to help others (Ghaisani et al., 2017). Social and emotional influences play an important role in media sharing behavior, and positive emotions are associated with higher incidences of media sharing than negative ones (Goh et al., 2009).
The strong occurrence of comments with religious content that is, mentioning words like "God", "blessed", "lord" and "faith" is another feature of their sample that caught their attention. The second most frequent word in the authors" corpus was "God", and in the ranking of categories on the two Instagram pages, "religiosity" is on the top 4 in all cases, with or without the top 2 users. This is in line with previous studies that indicate the strong presence of speeches of faith in moments such as the diagnosis of cancer, by patients and friends or family members (Soothill et al., 2002). An investigation conducted in Croatia with cancer patients showed that depression was less prevalent in the high-religiosity group of breast cancer patients, and religiosity is a more easily accessible coping resource in situations of limited predictability (Aukst-Margetiç et al., 2005). The same relationship between religiosity and a better quality of life among cancer patients was found in a study with Iranian Muslims (Zargani et al., 2018) and in Saudi Arabia (Al Eid et al., 2020).
Religious comments in online interactions can be a way for users to develop supportive social ties and a stronger sense of control or empowerment (Chun and Lee, 2017).
It is important to highlight that of all 8,928 categorizations, 9.2% (818) referred to the narratives of only two followers of the Hospital de Amor, who routinely published comments during the analyzed period and are probably bots. Bots on social media are used, among other reasons, to provide more likes, more shares and more clicks to social media sites (Flew and Iosifidis, 2019). It is relevant to remember that "misinformation could be spread across social media through various techniques, including automated bots and human agents who support some kind of propaganda" (Yang and Horning, 2020).
The authors noted a considerable female presence in the digital narratives of their corpus, since 80% of the comments categorized as "Acknowledgements" were written by users who identified themselves as women. They do not have access to user demographics on the two Instagram pages, but men and women write words such as "thank you" or "thankful" differently in Portuguese language, as there is a gender inflection in this case. This fact seems to indicate that most digital narratives on the two Instagram pages are created by women. The same trend was already observed in other investigations on gender and engagement in social media (Brandtzaeg, 2017;Ross et al., 2015). A study that analyzed posts of individuals with mental illness on Twitter revealed significant differences between the content shared by female and male users; males express higher negativity and lower desire for social support compared to women (De Choudhury et al., 2017). A systematic literature that examined gender differences in healthcare utilization of lung cancer patients revealed that women tend to get diagnosed at an earlier stage and have a higher probability of using inpatient cancer-care services (Rezwanul et al., 2020).

Conclusion
In the content analysis of Brazilian cancer hospitals" Instagram accounts, the predominant characteristic of user comments was that of positivity, even though cancer is a disease laden with negative stigma. It was believed that this is a positive aspect of these speeches, since it can reduce the readers' fear about the diagnosis of the disease; this can even indirectly lead to an increase in the number of preventive tests. The authors saw a strong presence of social support messages, which seemed to be aimed to motivate patients and medical staff. Users" positive discourses can also be influenced by the content of the posts where the comments were published, but we have not been able to verify this in this study.
Religiosity was also a hallmark of these narratives, and the same relationship between the diagnosis of cancer and the manifestation of faith, and positivity is observed in other studies, in different cultural backgrounds. In addition, women are also the ones who author social support messages the most in social media. Further study may explore the sociodemographic characteristics and the gender of the users who produce these types of narratives about cancer, to better understand their aims and motivations. It would also be interesting to investigate specific ways of engaging different audiences in health promotion issues, a subject that is becoming increasingly important.