“In a collection of tweets there might be a story to tell about how people describe their health experiences, which might be hard to capture in any other way.”
Researchers and clinicians are just beginning to explore this abundance of community insight. They’ve been able to track and forecast the spread of Influenza and learn more about peoples’ mood variations, their perspectives on cervical and breast cancer screening as well as about the misunderstanding and misuse of antibiotics.
At UC, San Francisco, a group of researchers conducted a study of Twitter messages to better understand how people experience pain, in particular pain from toothaches, which accounts for about 40% of an estimated $80 billion in pain-related health care costs every year.
In the following interview, Barbara Gerbert, PhD, professor emeritus and lead investigator of the study, and Natalie Heaivilin, DDS, 2013 graduate from the UCSF School of Dentistry, share what they learned and how Twitter may help health professionals to prevent dental disease in the future.
Q: Why do you think people are using the Internet, and in particular social media, to share information about their health? And what does that mean for health professionals?
With the Internet people can investigate their own health online, from seeking information on how to achieve a better night’s sleep, to looking up images of lesions to determine if the one on their skin might be malignant, to checking for symptoms of a stroke. The Internet is always available and easy to access. An immediate health concern might be relieved without the hassles of making appointments, missing work or school due to pain or to seek health care, or dealing with insurance.
Additionally, Twitter and other social media provide an outlet for people to relate to others, including sharing details about their own health. That’s powerful. Of course not everyone shares such personal information, but there’s enough chatter for health care professionals to consider becoming part of the conversation. If what we read online influences what we do, then well-crafted public health messages could have a big impact.
And even prominent figures, who presumably have access to regular health care, will seek health information from Twitter. For example, broadcast journalist Anderson Cooper recently asked his 1.5 million followers on Twitter, “How do you know if you have bronchitis? Anyone know?” — Many of his followers recommended, some sarcastically, that he see a doctor; some provided personal stories about their own bouts with bronchitis, and others gave management advice based on their own experiences. Some health care professionals saw this as an opportunity to weigh in with science-based answers.
We don’t know why Cooper used Twitter to post his question and we don’t know if he acted on any advice tweeted his way, but the example shows that even a person with the means and accessibility to see a health care professional sometimes chooses to ask millions of strangers for immediate advice before seeking advice from a professional.
Q: What makes Twitter a useful research, and even clinical, tool?
A colleague of mine, Jennifer Gibbs, DDS, PhD, once suggested to me that there might be a story to tell about how people describe their health experiences on Twitter, and notably, that the story found on Twitter might be one that would be difficult to capture in any other way.
Most traditional health surveillance data are captured by standardized surveys and interviews with people, long after the fact. These surveys rely on participants’ memories and are cued by categories, and multiple-choice options, predetermined by researchers.
With tweets, it’s possible for us to learn about pain-related experiences in real-time. Without prompting, people are talking about the severity and location of their pain and how upset they are about curtailing their activities.
From a student’s perspective, Twitter research is interesting because it provides a window into thousands of personal health experiences that offer an invaluable learning opportunity. And as a future clinician, especially the negative impact expressed in tweets about dental pain has given me the impetus to learn all I can about how to comfort patients and alleviate dental pain.
Q: Were there any surprises or difficulties in conducting this study?
Initially, we thought that tweets would be easy to interpret because they are limited to 140 characters. We were also a bit concerned that we wouldn’t find anything interesting. As it turned out, we were wrong in both assumptions. In fact, tweets often contained juicy content despite the character limitation, and we spent countless hours discovering some recurring themes and debating the meaning behind them. As a result, we designed a rigorous coding system to categorize the tweets.
A few of our findings showed that 44% of the actions taken by Twitter users indicated they sought professional help, 17% missed work or school, and 14% involved actively asking the Twitter community for advice on how to deal with dental pain. These data can be useful given the high cost of health care associated with alleviating dental pain.
Q: Do you see any limitations to this type of research?
I think the most important limitation is that data from Twitter and other social media exclude people who do not use Twitter and who are likely to be the most vulnerable. However, there is a trend towards increased use of social media sites and the range of the age of users will likely expand over the next decade.
Q: What’s the potential for this type of research to improve health?
Never before has it been so easy for health care professionals to reach out to so many people and disseminate health information through a conversation.
The biggest question for us is how we can leverage Twitter and the Internet to learn what public health questions people have, and then how to disseminate science-based answers. This way we can counteract misinformation and misunderstandings about health conditions and treatment options, thereby augmenting current efforts used to prevent dental disease.
That’s right. Although tweets can be helpful to people, we also need to think about the potential harm from inaccurate information and how to address this risk. For example, a UC San Francisco study from 2012 found that only 14% of Twitter-delivered smoking cessation programs provided information consistent with the U.S. Tobacco Treatment Clinical Practice Guidelines. Additionally, researchers at Columbia University in 2010 found misuse and misunderstanding of antibiotics in 6% of antibiotic-related tweets, and a 2012 study conducted in Australia showed that 8% of concussion-related tweets contained health management information that was inaccurate.
There is clearly a need for people to easily access information about specific health concerns online. Health care professionals need to find ways to participate in social media to answer this need. Imagine if accurate health care messages were readily available for urgent conditions such as the symptoms of stroke, which then resulted in patients seeking prompt care. Imagine that the public could receive important information about the appropriate use of antibiotics, or that patients experiencing pain would know when to seek care and how to distinguish between the need for emergency assistance or routine care. The implications could be amazing.
This Q&A is part of a perspectives series from CTSI at UCSF.
- “5 Mins of u is better than dealing with cancer 4 a lifetime”: An exploratory qualitative analysis of cervical and breast cancer screening dialogue on Twitter (Lyles et al. 2013)
- Twitter=quitter? An analysis of Twitter quit smoking social networks works (Prochaska et al., 2012)
- ‘What’s happening?’ A content analysis of concussion-related traffic on Twitter (Sullivan et al., 2012)
- Epilepsy in the Twitter era: a need to re-tweet the way we think about seizures (McNeil et al., 2012)
- Public health surveillance of dental pain via Twitter (Heaivilin et al., 2011)
- The Use of Twitter to track levels of disease activity and public concern in the U.S. during the Influenza an H1N1 pandemic (Signorini et al., 2011)
- Diurnal and seasonal mood vary with work, sleep, and day length across diverse cultures (Golder & Macy, 2011)
- Using social media for research and public health surveillance (Eke PI., 2011)
- Dissemination of health information through social networks: Twitter and antibiotics (Scanfeld et al., 2010)