The era of “big data” has offered our current digital economy different perspectives across a variety of areas, bringing together epidemiologists, computer scientists, economists, sociologists, bioinformaticians, and other scholars with large volumes of data and information, produced by and about people. Researchers are accessing routinely collected data and then aggregating and linking those data across sectors (eg, business, science, justice, health, and social care), each with its own advantages, limitations, and ethical issues [1]. Secondary use of routinely collected health care data holds great potential benefits in epidemiological studies primarily due to the large scale of preexisting data [2]. This use of routinely collected health care data to inform policy allows evidence to be collected without placing a burden on individuals. This is of particular importance when looking to tailor support for individuals who may face stigma or other barriers to taking part in research, such as individuals with a history of self-harm (SH). The COVID-19 pandemic exemplified the importance of using health care data to inform policy and practice.

Big data lies at the heart of efforts to aggregate information that will assist in identifying disparate needs across geographic regions, groups, as well as epidemiological trends [3]. The advantages of routinely collected data need to be balanced against ethical and legal challenges, such as patient anonymity, data security, fairness, and transparency [4]. Health care researchers identified the need to gain a better understanding of the public’s attitudes toward the use of their data by enhancing their participation in the research process and ensuring use of their data aligns with public interests [5]. While the public is generally supportive of the use of their health care data [4,6-8], certain fields, such as mental health data, arguably remain differentiated due to their perceived sensitive nature [9] and have unique ethical considerations [10]. Studies exploring public views on sharing mental health data found community attitudes to be largely positive [9,11-13]. The need to question the long-term efficacy of measures used, the validity of the types of data resources, and the risks of bias and privacy were highlighted [10,13-17].

In contrast to health data, social media and other internet-based platforms often lack clear and consistent procedures for the use of data generated online [18]. This rapidly growing technological landscape is changing our definition of knowledge and raising significant ethical questions, namely “Can data be used simply because it is openly accessible?” as well as critically evaluating the methods of collection and biases associated with it [19]. Guidelines have been developed outlining ethical factors, such as confidentiality, security, consent, potential harm, and levels of researcher control, when engaging with internet-mediated research [20]. While there is much debate surrounding the ethical principles that need to be applied in social media or internet-based research [18,21-23], these often do not involve the views of online users or make efforts to improve their level of awareness as to what is or can be done with their data [24].

Young people are particularly susceptible to both generating and sharing vast amounts of personal data, to the point of being characterized as a valuable currency in the digital economy [25]. Online presence within this age group as well as its forms changed over time from being more text based to increasingly changing into communication through imagery and videos [25], including sharing images of SH [26,27]. Contingent on appropriately securing and ethically handling these types of data footprints, they can arguably be used to identify patterns and risk indicators of those who struggle with mental health and would potentially benefit from early interventions that may prevent SH and suicide [28,29].

There is a wide misplaced acceptance that young people’s willingness to heavily engage with online platforms suggests that they do not care about their privacy [30]. However, some authors advocate little variance between adults and young people when it comes to their views on privacy in data sharing [31-33]. Researchers have used online platforms to examine the mental health and well-being of young people across various topics [34,35]; however, young people’s views on the use of their data for mental health research purposes, though somewhat explored [36], remain underreported. Moreover, the views of young people who have experienced SH have not been examined previously in relation to the use of their data for research. The often hidden and stigmatized nature of SH may give rise to specific concerns from young people that have implications for data management and ethical use.

Engagement with young people is essential to develop a better understanding of these issues and their views and was a key focus of this study. This survey set out to engage respondents with and without a history of SH in order to gain insights into their views and knowledge of how different kinds of data are anonymized and used for research and to determine whether there are any differences of opinions between those 2 groups. We also aimed to explore differences between those who had contacted health services for SH and those who had not.

This study was a web-based cross-sectional survey. The Checklist for Reporting Results of Internet E-Surveys was followed [37].

Ethics approval was granted by the Swansea University Medical School Ethics Committee (2018-0062). Respondents gave informed consent, were made aware that participation in the survey was entirely voluntary and no incentives were offered. Data were given anonymously with no personally identifiable information collected. Any potentially identifiable information given in free-text responses was removed before analysis (see details in Data Analysis section).

Respondents aged 16 to 24 years and residing in the United Kingdom were eligible. The survey was open to individuals both with and without a history of SH. SH is defined as intentional self-poisoning or self-injury, irrespective of the nature of motivation or degree of suicidal intent. This definition is necessarily very general because the nature of motivation or degree of suicidal intent is complex and may vary over time or between individuals [38]. A history of SH was based on self-report from respondents at the onset of the survey.

Recruitment occurred through a social media campaign conducted on Facebook (Meta platforms, Inc) and Instagram (Meta platforms, Inc) using visual material (eg, video clips and posters). Respondents were targeted based on age and location. Participants were also recruited via the Self-Harm Research UK research register. This is a register of individuals who have consented to be contacted for SH research [39].

Responses were collected between July 18, 2019, and January 31, 2020. The survey (including respondent information, consent, and debrief forms) was hosted on Survey Monkey (SurveyMonkey Inc) [40] due to its user-friendly interface and its capability to record anonymous responses. SurveyMonkey uses a unique user identifier based on cookies to ensure each respondent takes the survey only once.

A total of 3700 individuals took part in the survey. On the basis of the inclusion criteria, 1765 (47.7%) respondents were included in the analysis. Respondents gave 10,710 responses to 7 open-ended questions. Supplementary analysis was conducted to examine differences in item completion rates across questions and demographics (Multimedia Appendix 2). Nonresponses made up less ≤6% for each closed-ended question. Participants with a gender other than man/boy, woman/girl, older age, a history of SH, and those having contact with services were less likely to skip individual items.

We stratified participants into 2 groups based on the presence (SH) or absence (NoSH) of a history of SH. Of the 1765 included in the analysis 1448 (82.04%; 95% CI 79.9-83.9) reported previous SH. The demographics of respondents are shown in Table 1. Individuals aged 20 to 24 years were grouped together due to a small sample size.

A total of 41.02% (594/1448; 95% CI 37.0-45.1) of those who reported SH did so with suicidal intent. Self-injury (1370/1448, 94.61%; 95% CI 93.2-95.7) was the most frequently used method.

This was followed by ingesting medication in excess of the prescribed dose (570/1448, 39.36%; 95% CI 35.4-43.6).

When asked whether help was sought following incidents of SH, 39.71% (575/1448; 95% CI 35.7-43.8) of the respondents reported not seeking help from any of the sources listed. Friends, family, and neighbors were the most common sources of help (555/1448, 38.33%; 95% CI 34.3-42.5) followed by psychiatric or mental health services.

Among the 1448 respondents in the SH group, 42.06% (609/1448; 95% CI 38.2-46.1) reported contacting health services following an incident of SH. Of the 609 respondents who contacted health services, psychiatric or mental health services (n=518, 85.1%; 95% CI 82.0-87.8) were the most frequently reported followed by GPs (n=346, 56.8%; 95% CI 52.8-60.8) and hospital treatment (n=240, 39.4%; 95% CI 35.5-43.4).

We examined young people’s opinions on the use of their routinely collected data for mental health research using a web-based survey. We evaluated any differences in the views of young people with or without a history of SH and those in contact with services for SH. Views on the use of mental health data for research were mostly positive with most emphasizing the importance of the benefit to the public and patients. Respondents were less likely to want to share their social media data, which they considered to be more personal compared to their health care data. Respondents stressed the importance of anonymity and the need for an appropriate ethical framework. Individuals who reported a history of SH and subsequently contacted health services were more willing to share mental and physical health data compared with those who had not contacted services. We did not observe any significant differences between the SH and NoSH groups.

The use of data for mental health research of public and patient benefit was viewed positively across all questions asked. Young people were less likely to share their social media and financial data than ethnicity, marital status, and mental and physical health data for research.

Individuals both with and without a history of SH reported broadly similar views regarding their data being used; however, there were clear indications across responses that the importance of data sharing and its use for mental health research for young people and their attitudes toward it was associated with their characteristics and experience. Responses indicated that those with a history of SH might be less likely to use NHS mental health services if they knew that their data would be shared with researchers (although numbers were small); however, this was reversed in those with a history of SH but contacted services. Individuals who reported SH and contacted health services were more willing to share mental and physical health data compared with those who had not. This is further supported in the work done by Kirkham et al [12] with NHS users and Watson et al [13] indicating that while there were concerns that data misuse might worsen stigma, attitudes were largely positive. Favorable experiences of mental health services were likely to result in a higher willingness to share mental health data.

In keeping with older generations [1,6-8], young people share a mostly positive attitude toward their health care data being linked and shared for research purposes, with many relating their positive opinions with the use of this data for public and patient benefit. Respondents stressed the importance of anonymity and the need for appropriate ethical frameworks in keeping with previous work emphasizing the importance of identity protection, data security, transparency, and safeguarding policies while ensuring the use of data is within the public’s best interest [4,5].

Young people were less willing to share their social media data and considered this more personal than health care data. While anonymity was consistently reported as important, respondents varied in their understanding of how their social media data could be used and anonymized, particularly images. This is supported by existing literature in relation to young people’s lack of awareness and understanding of privacy issues online [31,33]. This contradicts the concept that young people do not care about privacy simply because they are heavily and publicly engaged in the online world [45]. Views on the use of social media data were shared by an adult population in a qualitative study performed by Beninger [46], where users expressed diverse responses toward social media research, ranging from skepticism on how it could be used for research to acceptance and ambivalence that stemmed from the idea that social media is already public and as such may inevitably be used. While some views were similar across generations, adults were less reluctant to share their social media content, as long as conditions, such as privacy and consent were met [47].

Respondents expressed negative feelings about the use of their data for profit or advertising, and this was frequently referenced in relation to the use of their data by private companies.

Anonymity, consent, and research skills were key factors that influenced positive feelings toward sharing mental health data. Losing control over their data, linking with other datasets, and inability to withdraw from data sharing were concerns that influenced negative feelings and unwillingness to share data. Another common theme signified respondents’ willingness to share their mental health data only if other conditions were met, such as the ability to opt out of data sharing and if data are used to understand more about mental illness. Respondents appreciated the importance of mental health research to better understand mental health and inform services, especially when given more control over their data.

Most respondents strongly agreed that they should have the right to opt out of mental health data sharing and that it is important that mental health data are held by an organization they trust with charities, the NHS, and universities being the most trusted and private companies and industry the least. This aligns with views from the adult population showing strong support for the NHS and academic institutions for data sharing [16,48]. Only one-tenth strongly agreed with the statement that they would be less likely to access NHS mental health services if they knew that their data would be shared with researchers.

This is a large survey with >1700 respondents and represents the first of its kind with young people aged 16 to 24 years with a history of SH. It is also the first to compare the views of those who have contacted health services for SH and those who have not. The combination of fixed- and open-response questions provides a richness of data and a wider scale of insights that can be used to guide future research. Using a convenience sample allowed a rapid low-cost engagement with young people. However, convenience samples are inherently biased and not representative, and results must be viewed with a level of caution [49]. The sample of respondents primarily consisted of younger women and girls, and the creative used to recruit respondents held a message of “mental health data and research,” which is likely to be the reason for a large proportion of people with a history of SH. The small number of participants in the age group of 18 to 24 years as well as the small number of men and boys, meant that a meaningful breakdown by age and gender was not possible. Online survey respondents are often younger, well-educated, and have English as their first language [50,51].

Online surveys may be subject to fraudulent responses, a limited ability to target certain features in the sample needed, and may be limited by the absence of an interviewer to clarify questions [52]. Some of these limitations were mitigated by ensuring the visual material incorporated a clear message to target respondents and designing the survey questions to be simple and easy to comprehend. The use of social media to recruit respondents was highly effective in terms of recruitment, cost, and time when compared with traditional recruitment approaches. It also facilitated the participation of underserved groups, who may face stigma or other barriers to taking part in research [53,54]. People with mental health issues are more likely to fill out mental health–related questionnaires than the remainder of the population, with this difference amplified for online surveys. This does not present an issue for research looking to gather opinions of individuals with mental health issues but is problematic for prevalence research [50]. Individuals with a history of SH were purposefully sampled in this study, and, notably, there was high engagement from this group. This was likely related to the nature of the social media campaign, which used visual materials, such as videos and posters, and was conducted largely over Instagram.

Our findings, particularly patterns based on respondents’ characteristics and experiences, highlight the importance of involving young people from diverse backgrounds at all stages of research development. Those with a history of SH valued participation of those with lived experience in mental health data research more highly than those without a history of SH. They also had reduced levels of trust in the use of their data for research across all organizations compared to those without. There is work to be done to increase confidence in individuals with a history of SH. We need to develop a shared understanding of consent, the “five safes framework” [55], anonymization processes, and the public benefits of mental health data science. This could be supported by coproduced resources and platforms for discussion, for example, online or in educational settings. Carter et al [56] stressed the importance of public engagement on a social level, rather than simply focusing on the requirements of formal regulations, in order to facilitate understanding and trust from the public on the topic of sharing their health data. Respondents repeatedly stated that they would like to know about the research being done. One of the key factors influencing positive opinions toward the use of data for research was the desire for it to be used to help others. Raising awareness of the ways in which research can be used to improve services and lives would likely have a positive impact on attitudes toward data sharing. Any campaigns to raise awareness should be widely accessible to young people with social media likely to be a useful way to reach them. Accessible case studies, demonstrating the impact of research on peoples’ lives, lay research summaries, and real-world impacts on services should be widely disseminated beyond academic audiences. Research findings should be made accessible to individuals who have participated, in ways that are suitable for their age or background.

There appears to be a greater need to educate young people about how their social media data are used and their data availability for research on platforms as part of the terms of service. Standardized anonymization processes such as those suggested by the British Psychological Society [20] would represent a positive step toward clarifying how identities are protected when using these data for research. Procedures should include consideration of the republishing of photos and direct quotes in research outputs, as even with usernames or personal details removed, there is still the potential for these to be traced back to the individual [22,57]. Resources demonstrating how images can be labeled, processed, and analyzed without a researcher accessing them directly should be developed. Special consideration should be given to communicating with vulnerable groups including children and adolescents and people with mental health issues [22].

Future research could explore the acceptability of data use from a wider range of services such as schools, universities, or social care. In addition, recruitment to increase the number of boys or men and young adults would provide a more comprehensive overview of the opinions of this group.

This is the first survey to gather in-depth opinions on the use of big data for mental health research from young people, specifically including those with a history of SH. Young people are most definitely aware, and indeed they care about how their data are being used and for what purposes. They are largely positive about the use of health care data (mental or physical) for research and generally echo the opinions of older age groups raising issues around data security and the use of data for public interest. In addition, this study demonstrates that young people feel strongly about the use of their social media data for research, which was felt to be more personal. Those who had been in contact with health care services for SH were more positive about these data being used to improve services. There is a need for greater awareness of how different sources of data are used for research with a focus on anonymization, privacy, and the way in which data are used to help others.