Social media platforms such as Yelp and Twitter have significantly altered the online landscape for restaurants. Now anyone with an Internet connection and an opinion can broadcast their thoughts to others interested in visiting. But what if public health officials could use Yelp and Twitter to track people mentioning foodborne illnesses online to detect outbreaks at restaurants? That’s exactly what researchers at Virginia Polytechnic Institute say is possible today, thanks to the number of people who take to social media to mention bouts of illness after eating out. In a new study published in Preventive Medicine, researchers studied restaurant reviews on Yelp of 5,824 food establishments dating from 2005-2012. They screened customer reviews for keywords related to foodborne illness such as “diarrhea” and “vomiting” and then analyzed every relevant review. Then they compared information from the reviews against food outbreak data from the U.S. Centers for Disease Control and Prevention (CDC). What they found was that foodborne illnesses reported by Yelp reviewers closely matched up with CDC statistics. Across five categories of food, the researchers found that the rates at which Yelp reviewers reported an illness and the rates of CDC’s reported illness information match up with a striking similarity:
- Meat and Poultry (implicated in 32 percent of Yelp illnesses, 33 percent of CDC illnesses)
- Vegetables (22 percent Yelp, 25 percent CDC)
- Dairy and Eggs (23 percent Yelp, 23 percent CDC)
- Seafood (16 percent Yelp, 12 percent CDC)
- Fruits and Nuts (7 percent Yelp, 7 percent CDC)
Based on the data, the researchers believe that social media reviews could complement traditional outbreak surveillance methods by providing rapid information on suspected foodborne illnesses, the implicated foods, and the restaurants involved. Reviews on sites such as Yelp or tweets about illnesses might assist public health authorities in detecting outbreaks earlier than would be possible using traditional methods, said Elaine Nsoesie, postdoctoral research fellow at Boston Children’s Hospital and Harvard University, and co-author of the study. When three to five customers of the same restaurant all go online to complain about getting sick, it might be a good indication that health officials should look into the situation. “We found that the reviews were very detailed in some cases,” Nsoesie told Food Safety News. “There were several cases where people talked about going to the hospital, or talked about a group of friends getting sick after eating out — very similar to what we see with foods implicated in foodborne outbreaks.” Still, the method isn’t perfect. Given that some foodborne infections can take two to three days for symptoms to develop, some people online may be incorrectly blaming illnesses on restaurants where they ate after they had already been exposed to the pathogen somewhere else. But, as a tool for early detection, it could prove quite effective, Nsoesie said. In fact, the Chicago Department of Health is already using an app called Foodborne Chicago to scan Twitter for keywords related to food poisoning. When it finds someone describing a foodborne illness incident on Twitter, it recommends that they file a report. Since March 2013, the app has catalogued more than 3,000 tweets, resulting in 193 reports.
