It’s a question every parent worries about in the winter: How bad is the flu season going to be?
That question, and the different ways to find answers to it, are opening up a lively discussion among health experts about the value of old-school scientific data and high-tech crowd-sourcing — and whether the new technology may someday help state and federal agencies analyze the spread of infectious diseases.
According to Cindy Burnett, an epidemiologist for the Utah Department of Health, “this flu season is a much more mild season than last year so far, but we haven’t peaked yet.”
That observation counters an analysis issued last week by Kinsa, a San Francisco-based health technology company, which declared “Utah is currently the sickest state in the country this week, with 5.7 percent of the population (over 182,400 people) experiencing flu-like symptoms.”
The difference comes down to methodology.
For this flu season, said Burnett, UDOH has changed how it tracks influenza-like illnesses. The new Moving Epidemic Method measures three data points: The incidence of flu-like symptoms, as reported by health care providers; the number of hospitalized cases; and reports from laboratories.
State officials don’t count individual cases of the flu, Burnett said, but looks for spikes in flu activity — like more people visiting doctors or clinics — that are “an indication that things are getting worse,” Burnett said.
The department collects data daily through the flu season, she said, and issues a weekly report from it. The reports are sent to federal health officials, who incorporate them into a national FluView report the following week.
“A week behind is not terrible for flu surveillance,” Burnett said.
Kinsa’s assessment of Utah is based on a single source of data: the undisclosed number of consumers who have purchased or received its thermometers and submit readings via a free smartphone app. The data source is narrower — but, the company argues, it allows faster analysis.
It aims “to know, in real time, how and where people are getting ill,” said Lauren Davis, the company’s vice president for marketing.
The company collects data through the app from users around the country, and aggregates it to create a nationwide map of where flu-like symptoms are most prevalent. The company makes money from selling the thermometers and by selling the aggregated data to vaccine distributors and other businesses, Davis said.
The San Francisco-based company has sold, or given away through schools, more than a million of its thermometers in the United States, Davis said. She wouldn’t divulge how many of those thermometers are in the hands of Utahns, but said distribution is representative across the country.
An individual’s health information, Davis said, is kept confidential and anonymous. Kinsa, she said, follows the standards of the Health Insurance Portability and Accountability Act, or HIPAA, designed to protect privacy of medical records.
Davis pointed to a study, by a pair of University of Iowa researchers, that found Kinsa’s national data on influenza-like illness to be as accurate as the federal government’s Centers for Disease Control’s figures — but in real time, rather than weeks after the fact, as CDC data is released.
Another study of crowd-sourced medical data, led by researchers at Boston Children’s Hospital, reported that “data from novel influenza surveillance systems can complement traditional healthcare-based systems” — if the crowd-sourcing draws from a large-enough sample of the population.
The Boston study evaluated the website Flu Near You, which is co-managed by Boston Children’s Hospital and relies on voluntary reporting of flu symptoms.
Internet-driven data for flu cases have had their ups and downs. Google launched a database, Google Flu Trends, in 2008, but it ceased publication in 2015. Google’s system relied on search results, which Davis said meant “there was just so much noise in that signal. People would type in ‘Bieber fever’ and Google would read that as ‘fever.’”
Officials at the Utah Department of Health have discussed such crowd-sourced data in connection with their own monitoring.
“Those types of surveillance are always on our radar,” Burnett said. Kinsa, she added, “is an interesting app. Time will tell if it’s something that’s really sustainable.”
Though Utah’s 2018-19 flu season has been milder than last year’s so far, it has produced two deaths among children — one in November and one earlier this month. UDOH does not release specific details of the cases, Burnett said, though in pediatric flu deaths, an underlying medical condition is often a factor.
In the past few years, Burnett said, the flu season in Utah peaked in late December or in January. (That is backed up by statistics kept by Columbia University.) This year’s season most closely matches the 2015-16 season, which didn’t peak until February.
“Only time will tell what will happen in the next few weeks,” Burnett said, adding that “it’s never too late to get a flu shot.”