Risks to Privacy With Use of Social Media: Understanding the Views of Social Media Users With Serious Mental Illness

Serious mental illness has a significant impact on individuals, families, and society because these disorders contribute to elevated risk of poverty, homelessness, substance use, early mortality, and suicide (1, 2). Many individuals living with serious mental illness, including schizophrenia spectrum disorders, bipolar disorder, or major depressive disorder, do not receive adequate treatment, and many available treatments are of poor quality or lack robust evidence (3). As digital technologies become more affordable and offer increasingly sophisticated features, there may be novel opportunities to bridge gaps in mental health services and deliver treatments of better quality and effectiveness to persons with serious mental illness (4).

Social media, in particular, has gained considerable attention as a potentially viable digital platform for reaching individuals living with serious mental illness (5). Recent studies have demonstrated that these individuals use popular platforms, such as Facebook, Instagram, and Twitter, at rates comparable to those for the general population (6, 7). Exploratory work has suggested that people living with serious mental illness use social media for connecting with others with similar mental health conditions, for seeking information, and for sharing personal experiences of living with mental illness (8–10). Therefore, growing recognition that online social networks are used daily by many individuals living with serious mental illness has sparked excitement surrounding the promise of leveraging these platforms for improving services. For example, pilot studies have demonstrated the acceptability of using Facebook to promote engagement and peer-to-peer support as part of lifestyle interventions delivered through community mental health settings (11, 12). In addition, a recent review found that peer-to-peer interactions, which are a common feature of social media, can potentially enhance engagement and retention in digital interventions among persons with psychosis (13). Judging from these early efforts, social media may yield new ways to enhance or augment various mental health services, and, as expressed by individuals with mental illness, this could include delivery of interventions to promote wellness and to support coping with symptoms of mental illness or navigating the health care system (14).

For future success, social media interventions will need to ensure that the benefits of these programs outweigh potential harms and that there are adequate protections to minimize risk. Commonly discussed risks of using social media for mental health interventions typically relate to participants’ privacy, informed consent, confidentiality, worsening of symptoms, and unintended consequences of disclosing personal information online (15, 16). To date, however, little is known about risks of using social media for mental health reasons from the perspectives of individuals living with serious mental illness. In this study, our objective was to determine whether social media users who self-disclose as having a serious mental illness express risks to their privacy when using social media for reasons related to their mental health. We examined whether respondents who express privacy risks differ on demographic and clinical characteristics from those who do not express privacy risks. Last, we asked individuals to explain the reasons why they may have reported experiencing risks to their privacy. Obtaining a better understanding of privacy risks from the perspectives of target users will be critical for informing the design of social media interventions that can safely support the needs of individuals living with serious mental illness.

Methods

Study Design

From May to December 2016, we recruited individuals on Twitter who self-identified as having a serious mental illness and asked them to complete a survey. Twitter is a popular microblogging platform with over 300 million active users as of 2019 (17, 18). Users post short messages called “tweets” containing up to 140 characters (in 2018, the limit increased to 280 characters). We searched Twitter for user accounts by using the following terms: “schizophrenia,” “schizoaffective,” “schizotypal,” “psychosis,” “bipolar disorder,” “major depression,” and “depression.” We could identify only Twitter users who self-identified in their profile or in a tweet or caption as having a serious mental illness and who had publicly accessible accounts. We used the Twitter platform to contact these individuals directly with a personalized tweet asking if they would be willing to answer a short survey. We then sent a follow-up tweet containing the link for the online survey to any Twitter user who responded to the initial tweet and indicated that he or she would be willing to answer the survey. Because this was an exploratory study, our target was to achieve 100 survey respondents who self-identified as having a serious mental illness. We did not compensate participants for completing the surveys. We received ethical approval for this study from the Committee for the Protection of Human Subjects at Dartmouth College.

Data Collection

We created the survey in English using Qualtrics online survey software. To minimize potential risks for survey respondents, we kept the survey anonymous by not collecting any personally identifying information. We collected demographic information, including age, gender, race-ethnicity, education, living situation, and employment status, including full-time employment, part-time employment, or currently in school; clinical information, including mental illness diagnosis, number of hospitalizations due to mental illness, and severity of symptoms of mental illness; and information on social media use, including frequency of use, types of social media platforms used, and reasons for using social media for anything related to mental health. Last, we asked participants about their experiences with risks of using social media for their mental health with the following question: “Are you ever concerned about risks to your privacy when you use social media for anything related to your mental illness?” If participants answered “yes,” they received an open-ended question asking them to describe their experiences of risks to their privacy (“If YES, please describe”).

Data Analysis

We tabulated survey responses to generate summary statistics. We used descriptive statistics, including chi-square tests and Fisher’s exact tests (when the expected value of a cell was less than 5), to explore differences in demographic and clinical characteristics and in patterns of social media use between respondents who expressed risk and those who did not. We used Stata 14.0 for all statistical analyses. For the thematic analysis of open-ended responses, we applied a conventional content analysis approach involving open coding of the responses (19). We assigned codes to the responses as the data were studied and rereviewed the data to identify new codes (20). We grouped similar codes for the participant responses together into categories that were allowed to flow from the data (21). We avoided using preconceived codes or categories because this study was aimed at exploring privacy risks from the perspectives of social media users with serious mental illness, and there has been limited research on this topic to date. The lead author read through the open-ended responses several times and assigned codes based on key topics that were mentioned. Several codes were grouped together into broader categories, although many of these codes ended up being the name of the actual category. The second author reviewed the codes that were assigned to participant responses and the list of overarching categories. Both authors discussed the final categories to reach consensus. The thematic analysis of participants’ open-ended responses was used to complement the quantitative survey responses.

Results

Study Sample

We contacted the first 207 individuals we found from searching the Twitter platform who self-identified as having a serious mental illness. In total, 94 agreed to complete the online survey. Ninety respondents, who self-identified as having schizophrenia spectrum disorders (40%), bipolar disorder (37%), or major depressive disorder (23%), completed the survey (Table 1). Respondents were primarily from the United States (N=60; 66%), Canada (N=12, 13%), or the United Kingdom (N=12, 13%), although one respondent each came from Australia, Denmark, Malaysia, the Netherlands, Grenada, and Singapore. The sample reported being predominantly non-Hispanic white (84%). More than two-thirds (69%) were female, and more than one-half were age 40 years or younger (58%). About one-half of respondents had a college degree (52%), and nearly one-half were currently working (defined as full-time employment, part-time employment, or currently in school) (48%). More than one-half of respondents reported being hospitalized several times for their mental illness (61%), and most (81%) indicated that they experience moderate to very severe symptoms of mental illness that interfere with their daily activities, such as work, school, or recreation. Participants were frequent users of social media, with 93% reporting daily use (Table 2). Many participants reported using social media to connect with others who have a mental illness (72%), share personal experiences about living with mental illness (67%), and learn about strategies for coping with mental illness (50%).

Privacy Risks

Among the 90 respondents, 36% indicated “yes” to the question of whether they are ever concerned about risks to privacy when using social media for anything related to their mental illness. In Table 1, we compare demographic and clinical characteristics between participants who expressed concerns about risks to privacy and those who did not. A greater proportion of female respondents (42%), compared with male respondents (22%), expressed concerns about risks to privacy, although this difference was not statistically significant. A significantly greater proportion of respondents with a college degree (45%) were concerned about risks to privacy, compared with those without a college degree (22%) (χ²=8.65, df=2, p=.010). For employment status, a greater proportion of respondents who were currently working or in school (47%), compared with those who were not currently working (26%), expressed concerns about risks to privacy (χ²=4.31, df=1, p=.038). In Table 2, we compare respondents’ frequency and type of social media use by privacy risk reporting. (See Table 2 for a description of the types of social media.) Users of WhatsApp were significantly more likely than nonusers to express concerns about risks to privacy (58% versus 27%; χ²=7.41, df=1, p=.006).

Qualitative Findings

Among participants who expressed concerns about privacy, 28 (88%) completed the open-ended question that followed. After coding these 28 responses, we identified four overarching categories reflecting participants’ key concerns about risks to privacy with using social media for reasons related to mental health. Table 3 provides a summary of these categories and sample responses from participants. Below, we describe these categories in greater detail.

Threats to employment.

Threats to employment was a prominent topic to emerge from participants’ responses, with 11 of 28 (39%) respondents emphasizing their concerns that using social media for their mental health could negatively affect their current employment or seriously affect future prospects for obtaining a job or being promoted.

Fear of stigma and being judged.

Many participants described fears that others would find out about them and judge them, citing lack of tolerance or understanding of mental illness as concerns.

Impact on personal relationships.

Several participants mentioned concerns that either a spouse or someone that they know may find out about their use of social media for their mental health, such as disclosure of their mental health challenges online. Others indicated that they were concerned that using social media in this way could affect their dating life.

Facing hostility or being hurt.

Some participants expressed concerns about being harassed by spam (irrelevant or inappropriate messages), bots (automated responses or conversations online), or trolls (offensive or provocative online posts). One participant mentioned facing “haters or hackers” on social media, while others mentioned potentially being hurt if others find information about them online.

Discussion

We found that about one-third of individuals who self-identified as having a serious mental illness online reported concerns about risks to their privacy with using social media for their mental health. It is important to note that participants who were concerned about risks to their privacy were more likely to have a college degree. This is consistent with literature showing that people who have a higher level of education also have greater health literacy and digital literacy (22). Specifically, digital literacy is reflective of greater user knowledge and is associated with increased awareness and understanding of privacy-related online behaviors (23). Although our sample consisted of frequent users of social media with overall high levels of education, we cannot assume that they are familiar with available options to protect their privacy online. Even though we did not ask about participants’ familiarity with or use of privacy settings on their social media accounts, research from the general population suggests that nearly one-half of social media users report some difficulty managing their privacy settings (24). The high education level of our sample is consistent with research from the general population showing that a greater proportion of social media users have college degrees, compared with high school education or less education (25). This also shows that we were unable to reach people with serious mental illness with lower education levels, and, therefore we may have a limited understanding of their perceptions of risk.

Second, participants who were concerned about risks to their privacy were also more likely to be currently employed, which aligns with the finding related to education, because educational attainment is correlated with employment in persons with serious mental illness (26). Furthermore, the impact of privacy risks on employment status emerged as a prominent theme from participants’ open-ended responses. Many participants expressed concerns about risks with using social media as it relates to their current job tenure, prospects for promotion, and efforts to seek future employment opportunities. This finding is noteworthy, given the challenges people with serious mental illness face in obtaining and keeping steady employment (27), as well as the benefits of gainful employment, including improved functioning, reduced utilization of outpatient mental health services, increased self-esteem, and successful recovery (28–31). Therefore, the development and delivery of interventions using social media will need to balance potential benefits with the risks that disclosing information online could affect employment prospects. This finding could also inform the WorkingWell smartphone application for supporting individuals with serious mental illness in the workplace (32), by illuminating the importance of including additional instruction with safety tips about how to use social media while avoiding sharing information online that could negatively affect employment.

Participants in this study were highly active on social media, and they disclosed their mental illness diagnosis publicly on Twitter. This could help explain why about two-thirds of participants reported not having any concerns about privacy when using social media for their mental health. It is also plausible that many participants have embraced online peer-to-peer support, because they reported connecting with others who have a mental illness, sharing personal experiences about living with mental illness, and learning about strategies for coping with mental illness as reasons for using social media. Prior research has explored potential benefits of interacting online with others living with similar health conditions, including feeling empowered, learning new skills, and feeling less alone (5, 8, 33, 34). Although a review of online social networking among people with psychosis found little evidence of risks (35), prior studies have described risks of self-disclosing stigmatizing health conditions online for various patient groups (36).

Among participants who reported concerns about risks to their privacy, we identified four broad categories of risks. As described earlier, risks pertaining to employment are especially relevant for people with serious mental illness. Risks related to fear of stigma and being judged, facing hostility, or being hurt are consistent with prior literature indicating that social media use could have unintended consequences of subjecting individuals to targeted stigma or cyber-bullying (16). It is noteworthy that participants mentioned concerns about how their social media use might affect their personal relationships, because this potentially parallels existing research showing that people with serious mental illness are acutely aware of how their diagnosis affects their relationships with others in offline contexts (37). Many individuals with serious mental illness experience difficulty forming and maintaining social relationships (38), and they can go to great lengths to manage how they interact with others in public because of feelings of shame and fear that others might find out about their diagnosis (39). A similar scenario could occur on social media platforms, where individuals with serious mental illness recognize the need to carefully manage their online interactions, given concerns that others may find out about their diagnosis, thereby having a detrimental impact on personal relationships.

Several limitations warrant consideration. First, we recognize that our findings should be interpreted cautiously given the exploratory nature of this study and the small sample of social media users who had disclosed personal information about their mental illness online. Therefore, participants’ responses cannot be generalized broadly to people living with serious mental illness or to those who do not openly disclose their mental illness diagnosis online. Second, there were limits to generalizability because a large proportion of our sample had a college degree and were currently employed, and there was limited racial and ethnic diversity. Recent research shows that lower income groups, as well as members of racial and ethnic minority groups, use social media at rates comparable to that of the overall population (40). This highlights the need to expand on our findings to determine how social media could support the delivery of interventions to hard-to-reach and underserved individuals with serious mental illness who are often underrepresented in traditional mental health service delivery settings. Last, we relied on self-reported diagnoses and were unable to confirm participants’ responses by using objective clinical data. Because a diagnosis of serious mental illness is associated with stigma, and because we did not compensate participants, it seems unlikely that participants would have been dishonest in their self-disclosure of having a diagnosis of serious mental illness.

Conclusions

Our findings demonstrate why it will be essential to involve social media users with serious mental illness throughout the development and delivery of social media interventions, given that their major concerns about risks relate directly to several key aspects of their daily lives. Interventions should draw from the perspectives of these individuals to ensure that potential risks related to stigma, harm, or self-disclosure are clearly presented and that content or material is included to educate participants about how to protect themselves. Future research will be necessary to examine additional risks not considered here, such as the potential for worsening mental health symptoms (41), delays in seeking professional help (42), and exposure to misinformation (16). For now, our study contributes novel findings that can immediately guide the ethical and safe use of social media as part of interventions, programs, or services for persons with serious mental illness.