Mental health information online : what we have learned from social media metrics in BuzzFeed ’ s Mental Health Week

Introdução: O número de sites com foco em conteúdo de saúde mental vem crescendo rapidamente. Considerando a necessidade crescente de acesso a informações precisas sobre tratamento em saúde mental, é importante entender a promoção dessas informações on-line. Objetivo: Analisar as interações da Semana de Saúde Mental do BuzzFeed (BuzzFeed’s Mental Health Week – BFMHW) em seu próprio site e em plataformas de mídia social relacionadas (Facebook, Twitter e YouTube) usando métricas de entrega de informações em tópicos de saúde mental. Métodos: Extraímos métricas de mídias sociais das 20 postagens com o maior número de interações no site da BFMHW e de 41 vídeos disponíveis na playlist da BFMHW criada pelo perfil BuzzFeed Video no YouTube. Analisamos o formato e o conteúdo usados nos métodos de publicação do BuzzFeed, bem como as seguintes métricas de mídias sociais: exposição (presença on-line, visualizações e tempo on-line), influência (curtidas) e engajamento (comentários, compartilhamentos, respostas e interações do BuzzFeed). Resultados: A análise das variáveis revelou que o envolvimento do público está associado ao número de mídias em que o conteúdo é publicado: visualizações no YouTube e compartilhamentos no Facebook (0,71, p <0,001), interações totais no Facebook (0,66, p <0,001) e número de interações totais no BuzzFeed (0,56, p <0,001). Conclusões: Nossos resultados sugerem que o YouTube pode ser um importante canal de informações, incluindo atividades e envolvimento em outras mídias, como o Facebook. As informações podem alcançar o público de forma mais eficaz se forem exibidas em mais de uma mídia e incluírem experiências pessoais, algum humor no conteúdo e informações detalhadas sobre o tratamento. Descritores: Saúde mental on-line, métricas de mídia social, engajamento, BuzzFeed, Semana de Saúde Mental, psiquiatria. Abstract


Introduction
Technology has had a great impact on our daily lives, particularly on information distribution and communication processes. 1 The current mode of information distribution has redesigned the way how people search for health information, and it has become an important aspect of treatment and care in mental health. Access to information may aid treatment, as many health problems and mental health disorders benefit from psychoeducational strategies. 2,3 Granting broad access to mental health content online has made it possible for people to access information and support 24/7.
While there is a vast amount of accurate mental health information available online and a growing number of websites focused on mental health content, 4 stigma is still common and is aggravated by false, inaccurate data that spreads over social networks. 5 Therefore, the pattern of online information distribution may either influence treatment positively, or it may delay adequate treatment and adherence by fostering stigma and suggesting ineffective or inaccurate treatment options.
By taking advantage of the most effective online strategies used to disseminate adequate knowledge, we may be able to reduce stigma and enhance social inclusion of people with mental illness. 6 Furthermore, patients with difficulties such as social anxiety, depression and introversion seem to provide an insight into their mental health status on social media. 7 Thus, social media could be a useful strategy to promote engagement in treatment interventions and research, 8,9 especially when a large number of studies currently face difficulties recruiting participants. 10 Access to accurate information could be a starting point for patients to wish to seek and receive treatment, 11 therefore it may be useful to study the case of a website that has received thousands of visitors talking about mental health, using content marketing and social media strategies. We chose to study BuzzFeed's Mental Health Week (BFMHW) because it was a mental health-focused event (week) promoted by a website that has a global audience and is a pioneer in the field of social news and entertainment. BuzzFeed publishes posts across many social media platforms and covers a wide range of topics using diverse, creative posts, which often "go viral." One of its core components is the association between informational content and marketing strategies, with a secondary focus on entertainment. 12 In December 2015, BuzzFeed promoted the Mental Health Week. 13 Several posts were published on their website, with consistent approaches discussing mental health problems and disorders.

Metrics of format and content
In addition to social media metrics, we considered it equally important to assess extradescriptive metrics by analyzing the format and content used in BuzzFeed's publications. All descriptive data were extracted as follows: a single author read or watched each post or video and arbitrarily classified their content or format considering four major groups for further examination:

Data analysis
The variables were analyzed using the Shapiro-Wilk normality test. All non-parametric data were expressed as medians (interquartile range) and parametric data as means (standard deviation  However, differently from frequency, the number of views  (Figure 3).

Pattern of narrative
Of the website posts, 75% used graphs/lists as the pattern of narrative and only 25% used self-reports/ experience as main stories. Among the videos, an array of different narrative elements were found, with the majority featuring self-reports/experience (58%). Of these, 75% were real-life self-reports and the remaining 25% were interpreted by actors from the BuzzFeed team. Fictional stories accounted for 24% of the videos; of these, 60% were short interpretations and 40% were web series. We also found that 12% of the videos used audio/images; only 5% were interviews (graphs A1 and A2 in Figure 4).

Content
In the analysis of content, experience and mixed content were the most common categories, accounting for 85% of the website posts. Among the YouTube videos, the experience category was identified in 81% and mixed content in 78% (graphs B1 and B2 in Figure   4). BuzzFeed's website posts also had information as content in 40%, humor and inspiration in 35% and pop cultural elements in 25%. Tips and expression were the least common types of content found, accounting for only 15% of the videos and 5% of the posts, respectively.
We observed that information, humor and tips showed similar trends in both posts and videos, as opposed to expression and inspiration, which demonstrated contrary trends.  (graphs C1 and C2 in Figure 4).

Publication format
In this analysis, we found that 75% of the website posts used lists as their main element, in contrast with YouTube videos, where lists accounted for a much lower percentage (17%). Videos also tended to show the "what" format, with 44%, compared to 20% of the posts. The "who" approach ranked second in videos (39%) and last in website posts (5%). Long text format was found in only 5% of the posts and long audios in 21% of the videos. Finally, the "why" format was present in only 2% of the videos (graphs D1 and D2 in Figure 4). According to these data, 73% of the YouTube videos used a persona/character.

Social media metrics
Exposure Some exposure variables were presented above      Table 2). The material extracted from the YouTube playlist showed a higher median of likes and favorites (17,786) than the items shared on Facebook (8,463) and Twitter (226) ( Table 3).  (Table 3).

Shares and retweets.
When we looked at posts shared on social media from the sample taken from the BFMHW website, the median number of shares and retweets was much higher on YouTube (19,178) than on Facebook (285) and Twitter (112) ( Table 2).
However, when we looked at the sample of videos from the BFMHW playlist on YouTube, the median of shares and retweets was higher on Facebook (4,510) than on YouTube (1,203). Finally, Twitter showed the lowest number of interactions (92) ( Table 3).

Comments, replies and BuzzFeed interactions.
Regarding comments, replies and BuzzFeed interactions in the sample of posts on the BFMHW website, a median of 9,687 actions was found on YouTube, vs.
40.5 on Facebook and only 2 on Twitter (  (Table 3). Of 44 posts and videos, the majority addressed treatment, followed by anxiety disorder and suicide. In this sense, it is important to highlight that evaluating how intrinsic content quality influences metrics was not within the scope of this study -rather, our aim was to analyze the pattern of information distribution online.

Pearson and Spearman correlations
We can say that, in this specific case of appropriate content, exposure was correlated with engagement; whether the same would occur with posts with lowquality content remains unknown. Nevertheless, among social media metrics, engagement rates could be used as a measure of the effective delivery of key messages, as it takes into account the person's actions as a measure of actual contact with the online content. 21 In this sense, we suggest that mental health initiatives online should consider video format and publication in more than one media channel.