Accessing social media information for pharmacovigilance: what are the ethical implications?

Recent history has seen advances in the technology available to the general public for communication purposes, and a significant increase in the use of the internet, in particular social media networks. To date, there are on average 330 million monthly active users of Twitter, and 2.07 billion monthly users of Facebook. ¹ These figures will likely increase as more and more people gain access to the internet and utilize such applications.

Use of these networks has become an integral part of modern life, so it seems sensible to explore their use for information-gathering purposes. Patient discussion forums and health and support networks may potentially also be utilized.

Social media platforms have proved to be useful in obtaining information, for example, in order to identify trends in public health, such as for influenza.^2–4 This medium provides a platform for users to discuss views and opinions as related to their health, including the use of medicinal products and their side effects. The discussions can be local or global, so there is the possibility of obtaining a broad range of information related to people’s experiences.

Guideline on good pharmacovigilance practices (GVP) Module VI ⁵ does contain guidance for market authorization holders (MAHs) to screen internet/digital media under their responsibility, and to collect valid cases identified from non-company-sponsored digital media they become aware of; however, there is no additional legal requirement for screening for, or guidance on how to manage, data from social media.

Regulators and MAHs have a duty to monitor the safety of medicines, and information from social media is important as it arises from direct user experience – it is patient-based, unsolicited and up to date. ⁶ These platforms may be a unique and possibly robust source of information about health and the use of medicines and side effects, both in the identification of adverse drug reactions (ADRs) and signals. These data can potentially add to the information received from traditional post-marketing reporting methods, where under-reporting is an understood and accepted limitation.

However, there are many challenges around accessing, identifying, and analysing information from social media, and obtaining it in a ‘useable’ format. Solutions to these challenges are being explored and tested, and involve analysing and converting user posts to a useable format to extract drug and reaction information.^7–10

One of the most interesting challenges is the ethical implications associated with accessing, utilizing and acting upon information from social media platforms.

For the majority of social media accounts, users can set their own privacy levels, either as public (visible to all) or private (visible to connections only). Does an account set as public mean there is an implied consent for use of data? ¹¹ One could argue that the data exist mainly in the public domain, and so therefore, yes, there is implied consent. However, public data can be accessed by anyone and the use of the data would occur without discussion or explanation as to how they are being used, and what for. Therefore, while there may be an argument for consent, it is not informed consent. This could be problematic from a regulatory perspective in relation to consent and data protection laws. (How about the inclusion of an ‘opt-in’ tick-box popping up when a new account is created, consenting to use of the data by third parties? Again, the regulatory and data protection impact would need to be considered.)

Also consider that personal health-related information obtained from social media may be useful to other industries – for example, ADRs from social media may be viewed as non-disclosure by insurance companies, resulting in litigation perhaps against the patient;^7,12 further, such personal data could be used for illegal purposes such as identity theft. MAHs would therefore need to ensure that the data gathered from social media remain private, confidential and secure – these factors are important in ensuring that public trust is maintained.

There is also the question of follow up. Is it possible to follow up with a user to confirm or request more details about an event? Social media is an informal arena, global and open to all. As such, users are a mix of people with different levels of education, the language used may be challenging, the text used may be culturally varied, it may contain spelling mistakes, etc. Extraction of information may not be straightforward – clarification and verification may be required.

If an MAH becomes aware of a serious ADR or an emerging safety issue, should they follow up? Ethically and morally the answer is of course yes, but given the lack of legislation around consent, and data privacy, it may not be possible. However, in the case of serious safety concerns, by not acting the MAH may leave themselves open to criticism at the very least, and possible legal action due to negligence.^7,12,13 If numerous posts were extracted which contained information on users feeling suicidal after taking a particular medicine, for example – and this information was new for this medicine – a responsible MAH should act. But can a responsible MAH act? Not taking action in such an instance seems to go against the very spirit of pharmacovigilance.

To further complicate matters, many people may be using multiple platforms in parallel, with different user names – so the same information may be posted on more than one site, introducing double counting and making it difficult to identify duplicate posts. Within a discipline in which incidence levels are used in the statistical methodologies applied for signal detection and in determining how to manage a safety signal, this is problematic.

Social media is open to all, and provides another platform for those who perhaps wish to discredit the use of medicines and the pharmaceutical industry in general. If information cannot be verified, false cases could be identified and collected by the MAH. A number of false cases may result in a false signal. If follow up is not possible, then the signal has to be treated as real. Analysis of this false signal is a waste of the MAH’s time, and inaccurate conclusions regarding the safety of a product may be drawn and have far-reaching implications, depending on the nature of the concern and the indicated population, potentially impacting patient safety.

It should also be noted that not everyone has access to the internet in the same way. A subset of users will exist who are of low computer literacy, have lack of, or restrictions on, their internet access, and maybe are even censored. These users may reside in a region of high usage of the drug in question, and they may experience ADRs. Additional screening measures may be implemented; however, for the reasons mentioned, information from these patients may be limited.

Also, limitations on the types of information posted may be present. In the UK the greatest number of users are aged 25–34, with the fewest users being in the 45–54 age group; similar observation have been made for the United States and the Netherlands. ¹⁴ So social media may not be as useful a source of information on the use of drugs in the elderly population.

Similarly, there may exist bias in terms of what is reported – and more importantly, what is not reported. We should expect minimal posts concerning ADRs considered embarrassing, such as sexual transmitted diseases or gynaecological problems, or observed in relation to taking a medication prescribed for such conditions. We may see little information regarding teenage problems (acne, development, etc.) even though this age group has high social media usage. Teenagers would most likely prefer to discuss such matters in an anonymized environment, and not in an arena in which their contacts can observe them.

So, while social media is a valuable source of information for use of medicines, it should not be considered as representative of all users, but rather as an additional source of information.

In the future, social media and the internet in general may be one of the most important sources of safety information. Increasingly, the pharmaceutical industry is looking to develop and implement digital-based solutions for patients and health care professionals (HCPs). The WEB-RADR project, launched in 2014, seeks to ‘utilise the powers of social media and new technologies for pharmacovigilance purposes’. ¹⁵ Consisting of industry, regulatory and academic leaders, the project is hoping to develop a mobile app that allows users to report ADRs directly to the national competent authorities, and develop data-mining tools for screening of social media.

The ethical issues mentioned above require consideration and clarification. However, while people may indeed consider their social media data as personal and private, and therefore not to be used by the pharmaceutical industry – only 34% of people queried were comfortable with information from their posts being utilized, with 90% agreeing prior consent was needed – 83% of those surveyed would be more inclined to report ADRs via a social media-type platform, given the correct measures. ¹⁶ But how to obtain quality reporting from social media users? Education, perhaps, but then how to implement? This could possibly be taken up as part of the WEB-RADR project.

A collaborative effort is needed between pharmaceutical companies, HCPs, regulators and the public in order to get full use of social media as a tool for pharmacovigilance.