Research question

The review aimed to understand what ethical guidance is available for researchers conducting online research. The underpinning research question was ‘what guidance exists on the ethical considerations that should be taken by health researchers conducting research with or about online health communities’ which guided the systematic search strategy. The overarching review question allowed the reviewers to search a wide range of available literature to capture what guidance is already available in the public domain and collate and review recommendations developed from other studies.

Aims and objectives

Specifically, this study aimed to:

• Map the available guidance on online ethics that is of relevance to health researchers
• Thematically describe the key ethical considerations covered by this guidance (e.g., consent, protection from harm, preventing identification, verbatim quotes)
• Highlight any areas of disagreement across the guidance
• Identify any gaps or ethical grey areas in current guidance.

Identifying relevant studies

The search strategy was developed using search terms based on the PCC framework [15,16]; population of interest (health researchers), the concept (ethical guidance and considerations) and context (research with online health communities). JH and JG led the development of the search strategy and search terms focused on 1) online 2) research 3) ethics and 4) policy and guidance. Searches were conducted across Scopus, Web of Science, Medline and Psycinfo to locate publications over a 10-year timeframe between 2013 and 2022. A ten year timeframe was chosen due to rapidly evolving trends across social media and technology, with this time frame best reflecting the current online platform landscape. Manual searching of reference lists was undertaken and grey literature searched using Google Scholar, Google and websites of key health related research bodies and regulators (e.g., NIHR, BPS, Royal College of Physicians).

Searches identified 3,294 records which were imported into Rayan online reference manager and duplicates removed (n = 1,177). Study selection utilised a two-step process to screen titles and abstracts (n = 2,117) and then full paper screening (n = 189). All screening was undertaken by JH, JG, EMC and RS and following inclusion and exclusion criteria set out in Table 1. Two reviewers each blind screened papers at both screening stages with conflicts resolved by group discussion and a third review where necessary. Research papers were selected for the review if they were conducting one of the forms of online research outlined in Table 1 in a health-related field and if their article made recommendations or guidance of their own for ethical practice as a result of their research experiences. Grey literature followed the same two-step process for screening. Twenty-seven articles and guidance were selected for data charting. All searches were conducted between February and June 2022, and were repeated in August 2023 however no new guidance was identified in these latter searches.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Inclusion/exclusion criteria for study selection.

https://doi.org/10.1371/journal.pone.0302924.t001

Risk of bias

Whilst quality assessment is not a requirement of scoping reviews, assessing the methodological and other qualities of studies can support in contextualising findings and enable interpretation [15]. Therefore, the selected articles were assessed for quality and bias using the Joanna Briggs Institute (JBI) Critical Appraisal Checklist for Text and Opinion [17]. The Text and Opinion checklist was considered most suitable as the scoping review had identified a range of different publications which made recommendations on ethical practice in online health research including organisational guidance documents, commentary and opinion pieces and methodological papers. As presented in Table 2, the checklist criteria focused on whether the ethical recommendations made were clearly identified and logical, had standing in their field of expertise, were relevant to health researchers and had congruence with existing research. The quality of the papers and guidance was independently assessed by one of the four reviewers (JH, JG, EMC, RS) using the predetermined questions. The quality assessment of the papers can be viewed in Table 2. The review papers largely met the criteria and after discussion, the decision was made to include all 27 papers in the final review.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. JBI critical appraisal checklist to assess trustworthiness, relevance and results of selected papers.

https://doi.org/10.1371/journal.pone.0302924.t002

Data collection, summarisation and presentation of results

A data extraction form was developed in a Microsoft Excel Spreadsheet charting all relevant data including authors, year of publication, key ethical topics covered and, for research papers, which if any ethical organisational guidance had been followed. The data form was piloted independently by two authors (JH/JG) on 10% (n = 3) of papers, and the remainder of included studies split between the two reviewers. Data was charted across four key areas 1) consent, 2) confidentiality and privacy, 3) protecting participants from harm and 4) protecting researchers from harm. The areas where initially informed by the key subheading across all included ethical guidance and refined through discussions involving all members of the team. All data was systematically extracted and charting was conducted by two researchers (JH and JG) and checked for consistency. The researchers then made comparisons across the selected papers to identify common themes and gaps across the existing guidance. Once data extraction was complete, JH and JG discussed the findings to identify the key points of ethical consideration within each theme which were then synthesised to create a narrative account of the guidance.

Consent

All guidance discussed the complexities of obtaining informed participant consent in online research with BSA [9] and AOIR [11] noting particular difficulties for big data studies due to the sheer number of participants involved. According to the BSA, whilst informed consent is not legally required to obtain data from public online spaces, it cannot be overlooked from an ethical perspective [9,10], particularly when topics are sensitive in nature [10]. The BPS [8] assert that if data is not considered in the public domain and there is no scientific justification for undisclosed use, consent should always be gained. The AOIR [11] state that consent should consider environmental factors and expectations of the online communities and users under investigation. A number of practical difficulties were noted in obtaining consent including users having left online spaces and the potential disruption to online communities [10]. The BPS [8] outline the need to verify participant characteristics including vulnerability to coercion and ensure consent processes are not overly complicated and lengthy. BSA [9] advocate for a participatory approach of ongoing communication and recommend researchers consult forum moderators for permission (although they acknowledge moderators may not speak for all users) [10].

The majority of papers (n = 13) concurred that consent was more ambiguous in public online spaces [8,11] and was an ongoing process requiring continual reflection and communication with participants [4,18,19,23,25,27,30,32,33,35,36,38–40]. Eight papers repeated the BPS assertion that not seeking consent for public domain online data could be justified when the scientific and social value outweighed the associated risk [18,19,23,30,33,36,38,40]. Examples included data on non-sensitive health topics taken from large publicly accessible forums (for example Mumsnet) and blog or vlog data produced for a public audience [25,27,33]. Two studies [21,22] took a more extreme view that all online spaces are private, and consent should always be sought. However, in these instances both were discussing recruiting participants from online spaces to take part in conventional research.

In line with AoIR, seven papers gave attention to the expectations of online communities in relation to research [19,25,30,33,36,39,40]. For example, Burles et al [25] suggested that disrupting online illness support communities could cause members to change or reduce the support they provide to others. Three papers agreed with the BSA that consent should be sought from community moderators [40] using the examples of a private fathering forum [33], an online youth counselling platform [36] and closed Facebook groups [33,39]. However, seven papers acknowledged challenges of verifying age characteristics online [26,28,29,32,34–36] and the ability to give or capture informed consent online [36], and recommended consent processes relative to topic sensitivity [35] developed in consultation with clinical and subject experts [36]. Five papers acknowledged the challenges of ensuring participant information was read by participants [27–29,32,36] with solutions including FAQs [28], online quizzes [28,29] and multi-stage consent forms [32]. The scale of online datasets was also viewed as a barrier to seeking individual level consent [8,9,11]. Solutions proposed included seeking consent for quotes used in publications [11,27], as well as the posting of debriefing messages on the online communities or a dedicated study site [27,36,40]. Finally, three papers discussed the challenges of withdrawing consent in online research, for example when data included in a study is deleted by the user [4,36] or in prospective studies when the sensitive data has not yet been produced, with Clark et al [27] giving the example of using Twitter posts to predict post-natal depression.

Confidentiality/Privacy/Anonymity

As the guidance on consent indicates, all three guidance documents acknowledge that privacy is challenging to define in online spaces [8,9,11]. The BPS generally defines a public space as one where participants could reasonably be expected to be observed by strangers [8]. However, their internet mediated research guidance acknowledges this distinction is not so clear cut for online research because online interaction can simultaneously take place publicly (for example a public online forum) and privately (from the home) [8]. Where this ambiguity exists the BPS [8] and BSA [9] propose several practical steps to ensure participants are sufficiently protected from harm which are discussed in the section below. As the only international guidance included, the AoIR [11] gives a wider contextual outlook at differing perspectives between disciplines and across countries. They highlight the difference between deontological approaches to privacy (protecting the rights of autonomous individuals) in Europe versus the more utilitarian approaches (achieving a greater good for society) taken in the UK and US and advocate for “ethical pluralism” which acknowledges these legitimate differences but develops shared norms and practices. The AoIR highlights several contextual factors which must be considered alongside the practical steps outlined by the BPS [8] and BSA [9]. This includes legal frameworks (e.g., GDPR), platform terms and conditions, management of data, adequate anonymisation of large data sets and reliance on companies giving access to their API which can favour certain countries, universities and researchers.

Twelve papers reflected on the blurred lines between public and private spaces online and users varying expectations of privacy in relation to research [4,21,23–27,29–33,36]. For example DeCamp [29] made recommendations which aligned well with the guidance, stating: 1) researchers should understand site Terms and Conditions regarding privacy and research participation, 2) if a site requires registration and moderator approval, there is some expectation of privacy, and 3) sites with express purpose of sharing individual health stories might have some expectation of privacy and researchers should engage with moderators and users before beginning the research. Two papers recommend the publication of a privacy statement via study or research group specific websites [18,32,35]. The issues highlighted by the AoIR in relation to different legal and cultural understandings of privacy internationally were rarely discussed, with only three papers providing any reflection on this [26–28].

Participant harm

The BSA [9] and BPS [8] both argue that online research should maximise benefits and minimise harm through values of protection, respect, dignity, and privacy [9] and gaining informed consent and anonymisation are vital to this. The BPS highlight social responsibility to respect the social structures of existing online groups and the consequences of undertaking research upon group cohesion and trust. Several practical steps are recommended to address this. These include: familiarisation with the online space of study to ascertain if participants perceive it to be public [9], discussions with group moderators on the best way to research their online groups [8,9], anonymisation of data through removal of personal information [8,9] including online pseudonyms [8], not including potentially embarrassing or sensitive information [9], paraphrasing or combining quotes [8,9], age verification and not naming online communities [8]. The guidance also note that these actions should be proportional to the risk of harm, for example sensitive topics and more vulnerable groups including children and young people, women, certain emotional states such as grieving and/or trauma, illnesses; and minorities such LGBTQ+ communities [8,11]. Research on sensitive topics where risks are high and the ability to control is low should be avoided and consideration must be given to researchers’ responsibility in reporting online users discussing negative well-being and criminal activity [8].

Sixteen of the studies recommended the anonymisation of data [4,19,23–28,30–33,35–37,40], achieved through removing online pseudonyms, sensitive health information, geolocation data and the names of sites from which data was collected. Benton et al [23] advise caution when linking data across multiple platforms as individual online personas could “out” sensitive health information. Ten studies recommended paraphrasing verbatim quotes [4,18,23–26,30,33,35], and this varied from changing minor words to prevent retrieval form a search engine, through to the fabrication of synthetic quotes based on participants’ words [23,25]. Eighteen papers felt potential harm was dependent on the sensitivity and anonymity of each online group and that vulnerability should be sensitively negotiated as the research progresses [4,18,19,21–24,26–32,34–36]. Three papers suggest this is addressed through familiarisation with site netiquette and potential impacts on cohesion and trust [29,31,33].

Researcher harm

Two of the three guidance discussed protecting researchers from harm [9,11]. This was considered particularly important for research relating to politically sensitive topics or where researchers hold identities (e.g., ethnicity, minority identity, sexual identity, political activism, etc.) that could trigger strong ideological reactions [11], placing them in publicly visible, vulnerable positions and at risk of abuse [9]. Both guidelines called for the support from individual institutions, stating that researchers should receive ethical approval prior to commencing research and continue to discuss challenges with the committee throughout [9]. Furthermore, institutions should develop policy detailing support procedures for researchers experiencing online threats or harassment related to their work [11]. The BPS [8] also highlight the challenges of ensuring integrity and scientific value when researchers have less control over who can participate, environmental conditions and responses during the research and variations in the research procedures caused by different hardware and software.

Seven papers considered the potential researcher harms from online research. Two papers noted that researchers should be cautious of putting themselves at risk of online abuse particularly if they were from groups vulnerable to harm [18,24]. Hunter et al [4], in line with AoIR [11], discuss the emotional impacts of exposure to sensitive or politically extreme information online, recommending clearly defined researcher roles and regular debriefs when exposed to content. Five papers reflect on the impacts upon research integrity raised by the BPS [8,26,30,33,36] with three papers suggesting that seeking approval from institutional research ethics committees is the best way to ensure methodologically and ethically sound research [30,33,36].

Limitations

This scoping review only included studies which were of relevance to health-related researchers. Given the lack of guidance from health organisations, this may have excluded some broader ethics documents which are being used by health researchers to inform their studies. We limited our scoping review to guidance and papers published in the past 10 years to reflect the fast-changing nature of online communities and online research methodologies. It is possible this may have excluded some earlier guidance in health-related fields, although if these documents have not been updated then their relevance to current health related researchers may be limited. Our review also identified some emerging issues which were beyond the scope of this paper including the ethics of digital images [11] and the ethical impacts of big data in health, particularly prospective studies which predict sensitive health issues such as mental health or chronic conditions [27,51] which warrant further research. Similarly, we only included papers which presented some form of recommendation or guidance to other researchers on ethical online research practices. Further research could consider online health studies more generally to understand which key ethical principles are being applied, although the early stage of our review suggests that reporting of ethical practices in online research studies is inconsistent.