By Chelsea Rice
One of the last warm Saturdays in September, my boyfriend and I planned to celebrate his birthday at a Cambridge restaurant that friends had praised as their favorite brunch spot. The food tasted great: We shared a plate of oysters to start, and he enjoyed eggs Benedict for his main course while I opted for a breakfast take on the classic BLT sandwich, mainly because it was served on a croissant, a buttery weakness of mine.
But upon arrival back home…our brunch backfired.
I ran out to pick up the birthday cake. When I returned, I found the birthday boy almost paralyzed by stomach pain, feverish and violently ill. While he spent the rest of his birthday in a migratory pattern between the bathroom and bedroom, I waited to see if I would get sick and searched online to learn how to report the food poisoning.
Turns out that here in the Boston environs, residents call a special number at the Boston Public Health Commission. It connects with a public health nurse who asks questions about symptoms, what you ate over the past few days, and where you ate it. On that memorable day, I shouted this option into the bathroom at my sick partner, but he was so nauseous he could barely talk to me. Reliving what he had eaten in the past 48 hours was the last thing he wanted to do.
As a health journalist, I know it’s important to report food poisoning — one in six Americans gets it every year. But as a consumer, filing a public health report can be an intimidating and impersonal process for a very personal — and vulnerable — experience.
We told all our friends and canceled the rest of the festivities, vowing never to return to the scene of the crime, but I still wondered: Were we selfish? Could we have helped others with our story? Have other diners had a similar experience at that restaurant?
According to the restaurant’s worst reviews on Yelp, the list is long. (I’m not naming it to give it the benefit of the doubt — maybe it was having an off day? — but the picture is grim.) The worst reviews include bugs in the plates, under-cooked proteins and foreign objects that broke a tooth. There are even a few that reported diners getting sick after eating there. And those are just the brave few who posted. I was alarmed that this restaurant still had a line and a reservations list with complaints like that. There’s no way a health department could ignore claims like these, I thought, if they were written up in an official report.
Little did I know, there’s a new bridge between social media and public health that is finally crossing that divide.
In 2011, a research group out of Boston Children’s Hospital published a study using extracted, keyword-related Yelp reviews, showing that the ingredients people described in their reviews about food-borne illness matched up with relevant ingredients that the CDC reported were involved in food-borne outbreaks for that time.
Now, that team is taking their work to help cities across the country address and more accurately monitor food-borne illness with HealthMap Foodborne Illness, part of a larger social media disease-tracking initiative based at Boston Children’s.
“When someone talks about diarrhea on Twitter they are really looking for people to care, and that’s really what it’s all about.”
Dr. Elaine Nsoesie and Dr. John Brownstein, who co-founded the project, are working with New York City, Chicago, St. Louis and other major cities to customize their foodborne disease tracking tool for each city’s needs.
“It’s hard to make people come to you,” said Brownstein. “People aren’t engaged necessarily in public health.” But if you can tap in to their online voices, he said, “you can actually get a huge amount of information that would not come from another vehicle.”
In Chicago, the city’s public health department monitors Twitter in a social media tracking initiative that HealthMap customized for them called FoodborneChicago. The tool filters tweets that are geocoded to a specified area through a system that recognizes key words related to food poisoning. Think “sick,” “food,” “vomiting,” “diarrhea,” “poisoning” and various combinations like “restaurant made me sick” or “vomiting after that lunch.”
A public health official can monitor the filtered tweets live, and sort them into “relevant” (ambulance icon), “questionable” (question mark) and “irrelevant” (trash can), as in the image above. The system learns to better recognize food poisoning-related posts over time. The tool accurately classifies tweets 85 percent of the time, according to Jared Hawkins, part of the HealthMap development team and informatics junior faculty at Boston Children’s.
“It’s easy to complain on your social media and get sympathy, so if you complain on Yelp or Twitter you get a connection, but a report to a public health department goes into a black hole where there isn’t any sympathy or interaction,” said Brownstein. “When someone talks about diarrhea on Twitter they are really looking for people to care, and that’s really what it’s all about.”
The system takes advantage of that natural hunger for comfort to gather data. Once a tweet is validated as relevant, the tool takes the public health official to a separate screen where they can see the sick person’s publicly available profile information. At this point, HealthMap FoodBorne Illness prompts the health official to reply to the post with a dynamic range of responses from empathetic to authoritative to actionable.
If the official chooses “Actionable,” the system sends the sick person a link to a mobile-friendly online form to fill out the specific information about their experience, including when and where the foodborne illness occurred.
“When you have food poisoning, you don’t know who to call, if anyone,” said Hawkins, who also has a Ph.D. in immunology. “And while you can stumble onto the online form if your city has one, it’s usually many pages deep in the public health website,”
“People are sick or mad, but they aren’t mad enough to fill out a 10-page PDF.”
“People are sick or mad,” he added, “but they aren’t mad enough to fill out a 10-page PDF. But they will tweet about it. So it reaches a much wider audience. And if a tweet can be a real report, that is huge.”
So how might this work in Boston? According to HealthMap, 10 tweets a day on average are relevant to foodborne illness within a 25-mile radius of Boston. In October so far, there have been 57 tweets. The high was in April, with 119.
Currently, the Boston Public Health Commission and Boston Inspectional Services do not make social media a regular part of their monitoring process to identify and report foodborne illness, according to spokeswomen at both departments.
But according to Brownstein, there is definitely potential there. Here’s how our city compares with some other major cities across the country on the number of tweets related to foodborne illness:
So is it “TMI” to post to social media about food poisoning? Not when your city’s health department is watching. They might be the only ones who really care — and as social-media surveillance mounts, they may actually do something about it.
Chelsea Rice is a digital health journalist living in Boston. She’s previously written for Boston.com, The Boston Globe and HealthLeadersMedia.com. Follow her on Twitter @ChelseaRice.
Sample page from the HealthMap tool: