Two researchers at Microsoft made headlines
last month when they announced that by looking at large samples of search engine queries they believed they were able to identify people who were likely to have pancreatic cancer and not know it.
Earlier this year, Novartis Pharma researchers reported
on a pilot project in which they used profiles from Facebook, Twitter, blogs and other online forums to track multiple sclerosis patients and see what led them to switch drugs.
In another sign of the times, their study was published in the Journal of Medical Internet Research,
a relatively new scholarly journal.
Clearly, social media research is new territory for software companies, science and medicine.
But with it come new ethical and legal concerns, says Eric Swirsky, JD, MA, of the University of Illinois at Chicago (Photo, left
). Writing in the American Journal of Bioethics,
Swirsky and two colleagues described the issues facing IRBs.
· Privacy: Are posts on Twitter or Facebook completely public?
· Consent: Does agreeing to the terms of service on a search engine make the queries used public information or should researchers get consent?
· Voluntariness: Is the voluntary aspect of study participation removed when the information gathered is from social media or search engines?
· Anonymity: When information is de-identified, who does the de-identification? How is anonymity guaranteed?
· Research intent: Will subjects be targeted for marketing? Will their data be sold to pharmaceutical or insurance companies?
· Scientific Validity: How can researchers be certain that the information they gather through such channels is authentic? What about demographics and representative sampling?
Such questions usually go to Institutional Review Boards (IRBs), bodies that set policies for medical research and monitor projects to protect the rights of subjects. But most IRBs simply have not addressed the questions on using data from social media, Swirsky says.
Privacy is a major concern. One school of thought holds that anything on social media is fair game, Swirsky explains, just as you assume whatever you do in a park is public.
“The assumption that is made by researchers is that people know the risk, so they don’t have a right to privacy.” In fact, he added, some experts think that the very idea of privacy in the United States is antiquated and needs to change. These experts argue that improving public health requires valuing society collectively more than valuing the rights of individuals. Viewed from that perspective, there is absolutely nothing wrong with using search terms or information from public profiles to conduct clinical research.
Swirsky’s more nuanced view is that posting personal information on social media is not quite the same as being in a public space. Social media is different because, he says, even though they probably shouldn’t, people have an expectation of at least some privacy.
Swirsky uses food as an example. If you take a sandwich to the park, people can see that you are eating a sandwich, but they would have to ask you to learn what kind it is or how it tastes. It would be unusual for groups of people to come up and ask about the sandwich and it would seem intrusive. But online, people might well take a photo of their sandwich and describe it in detail.
“In the public space, one understands the lack of privacy and we keep to ourselves, more or less," Swirsky says, "Online, users have more of an expectation of privacy and thus share more.”
Just as you must give consent each time you have a procedure of any kind done at a hospital, Swirsky says, so should you be required to give permission for your social media data to be used in research.
Another concern is the reliability of information in social media profiles. Researchers have no way of knowing whether what people say on Facebook or Twitter is true. There is no guarantee the opinions expressed on social media are the actual opinions of the users, “due to the phenomenon of trolling,” he says, one in which internet users try to deliberately provoke people by posting inflammatory or extraneous messages.
In a related concern, Swirsky questions whether the information gathered from Facebook would be representative, since social media users do not reflect the actual demographics on an area.
Using information gathered by search engines or social media platforms also means the people who provided it are not voluntary participants in a study, a key ethical requirement for traditional studies, Swirsky said. By using information gathered from search engines or social media platforms, the voluntariness aspect is removed.
Not only do people who post things on social media not know if their information is being used for research, they do not know the intent of the research, another ethical concern.
The Novartis study might be a case in point. Its authors concluded that “social intelligence can be used to improve outcomes research, and that it may be able to add information that is not generally available in claims databases.”
Did the subjects approve having their online postings used to further a drug company’s research? It's not clear if anyone asked them
Social media platforms, particularly Facebook, generate profits through creating, and selling, a product. The product is the personal information of the users. Companies purchase that information in order to market to specific users. It is not that far-fetched to imagine pharmaceutical companies or insurance companies paying for information gleaned from web searches and status updates.
The issues are not going away anytime soon. Younger researchers who are now entering the field now are people who grew up using social media, and such platforms are intuitive for them.
Social media permeates modern society. Even those without personal profile pages on specific platforms know about those platforms and have a clear idea of how they work. For many users, going on social media is akin to hanging out with their friends.
Whether or not serious research has found its place there is still open to debate.
Research using social media can be conducted legally and ethically, however, Swirsky points out, “It requires the correct analysis each time.”