What can Twitter and Facebook tell us about obesity in the United States?

Tweeting about pizza? That’s data.
Tweeting about pizza? That’s data.
Image: Reuters/David Gray
We may earn a commission from links on this page.

Open your favorite social media app and scroll through recent updates from people in your network. Chances are, you will see a picture of a beautifully styled plate of food, restaurant recommendations or a post about what or where someone is currently eating. These postings could reveal prevalent food choices in a community—information that public health researchers are finding useful.

Unhealthy diets and a lack of physical activity can increase a person’s risk for health conditions such as, diabetes and obesity. Based on data from the Centers for Disease Control and Prevention (CDC) and the Institute for Health Metrics and Evaluation, more than one-third of adults in the United States are obese. One approach to reducing obesity rates is to lessen the risk factors for obesity. Social media provides a unique and real-time view of people’s attitudes and behaviors towards these risk factors, namely, unhealthy diets and physical inactivity.

On many social media platforms, users have the option of enabling geolocation. This means the exact geographical coordinates (i.e., latitude and longitude) where a person makes a post can be recorded. For example, if someone posts about eating a chicken salad at a particular restaurant in Seattle, we can tell what restaurant they are referring to based on their geographical coordinates. We can also aggregate social media postings to see how many people have posted about eating chicken salad in a particular geographical unit (e.g., a county).

In a study published in the American Journal of Public Health, researchers led by Quynh Nguyen, a social epidemiologist at the University of Maryland, took advantage of this geolocation feature on Twitter to study how expressions of happiness, food choices (healthy vs. unhealthy) and interests in physical activity correlated with obesity prevalence across 3,135 US counties. Over a 12-month period, the researchers tracked mentions of 1,430 popular foods, 376 physical activities—including recreational activities, gym activities and household chores—and 66 terms related to alcohol resulting in approximately 80 million postings.

The researchers found that counties with the highest number of tweets about physical activity had fewer premature deaths compared to counties with the least number of physical activity tweets. Also, physical inactivity was lower in counties with the highest happy, food, and physical activity tweets, suggesting people in these counties are more likely to exercise. Furthermore, obesity prevalence was 2.23% to 2.49% lower in counties with the highest percentages of tweets about healthy foods, physical activity and tweets expressing happiness.

There are some limitations in the data. One of which is a lack of information on the age, sex, race, and socioeconomic status of people who post about physical activity and food preferences on Twitter. This is important because some populations (e.g., poor individuals) might be unrepresented. Another limitation is that only about 1% to 2% of tweets have geolocation information.

However, this is not the first study to look at the association between attitudes and behaviors on social media and obesity rates. A similar study published in 2013 found that areas in the United States where people tend to like physical activities on Facebook had lower obesity rates. The agreement between these studies suggest that social media posts of food and exercise can help us better understand obesity prevalence in the US. Public health practitioners could combine these real-time data with information obtained from surveys, hospital visits and other sources to design community-based interventions to reduce obesity rates.