Consumer reviews of restaurants and other facilities on Yelp could help track and prevent food poisoning outbreaks, according to a new study published in Preventive Medicine. Researchers at Virginia Polytechnic Institute looked at reviews on Yelp of 5,824 food establishments from 2005 to 2012, screening customer reviews for relevant keywords, then analyzed every review. This information was compared to data from the CDC for the same time period. The scientists found that illnesses reported on Yelp matched the CDC stats.
The Yelp reviews were of 13,262 businesses closest to 29 colleges in fifteen states. Those 5,824 businesses were categorized as Food or Restaurants. Researchers constructed a keyword list using common foodborne illness terms such as “diarrhea”, “vomiting”, and “puking”. They found that 4,088 reviews had at least one of those keywords.
Then they looked at personal reports and eyewitness accounts. Dates of illness, foods consumed, businesses reviewed, and number of sick people were calculated. They then identified bias by people who posted a large number of negative reviews compared to the median. They also focused on restaurants with more than two foodborne illness reports in the same year.
Foods were divided into five categories and illness complaints were matched to CDC outbreak reports. Meat and Poultry were implicated in 32% of Yelp illness reports, and in 33% of the CDC reports. Vegetables were suspected in 22% of the Yelp reviews, and 25% of the CDC reports. Dairy and Eggs were implicated in 23% of Yelp reviews, and 23% of the CDC reports. Seafood was suspected in 16% of Yelp reviews, and 12% of the CDC reports. Finally, Fruits and Nuts were suspected in 7% of Yelp, and 7% of CDC reports.
The report also found that California, Massachusetts, and New York had the most illness reports. Restaurant inspection reports for four of the restaurants implicated in more than one outbreak had at least one food violation in the last four years. They included contaminated equipment, improper holding temperature, and cleanliness of food and nonfood contact surfaces.
The report concludes by saying that these Yelp reports could be useful for monitoring foodborne illness outbreaks. Tracking reviews in near real-time could reveal clusters of illness. Since the foods implicated by reviewers matched the CDC reports, that information could also help public officials pinpoint foods responsible for the illnesses more quickly.
Still, the data source isn’t perfect. Incubation periods for different foodborne illnesses vary. Since it can take days or even weeks in some cases for the illness to manifest after infection, people often incorrectly blame their illness on the food most recently eaten. The data are also incomplete. More detailed analyses could refine these sources.
