A better prediction model for patient surges from influenza? New Internet-based tool shows promise, say researchers

Is there a better way for ED managers to stay on top of flu outbreaks and the corresponding surges in patient volume? New data compiled by researchers at Johns Hopkins Medicine in Baltimore, MD, suggest this may well be the case. And the cost of having better information is attractive, too. The researchers found that Google Flu Trends, a free Internet-based tool that monitors Internet search traffic about influenza, can provide important, real-time information about the kind of traffic EDs are seeing from patients complaining of flu-like symptoms.1

For the study, researchers tracked Google Flu Trends data for the city of Baltimore, MD, and compared it with data on people seeking care for flu-like symptoms in the adult and pediatric EDs at Johns Hopkins Hospital between January 2009 and October 2010. What the researchers found was a strong correlation between hikes in Internet searches for flu information and surges in patients complaining of flu systems in the EDs.

"Google Flu Trends is available online, and you can have it designed for national level, state level, or even down to city-level data. And one of the things our research did was validate the city-level Google Flu Trends tool," explains Andrea Dugas, MD, the lead study investigator and an emergency medicine research fellow at Johns Hopkins University.

What this means, says Dugas, is that practicing ED physicians in Baltimore have another tool at their disposal to help them establish whether patients who are coming in with cough, fever, and other flu-like symptoms actually have the flu or another ailment. While such information can be helpful, investigators are hoping to put the data to even better use.

"Right now, the work we have done so far shows that [the Google Flu Trends] data correlate with what we are seeing in the ED, so it is representative of what is going on at that time, but what we would really like to have is advanced warning," she says. "We'd like to know what is coming next week or the week after, so right now we are working on validating this as a predictive model for what is to come in the future so that ... we can be prepared and ready for when that increased volume of patients with flu-like symptoms comes in."

Beyond having a better idea of timing, Dugas says that investigators are also hoping to develop ways to optimize their response to surges in patient volume. To work on that front, Hopkins hosted a conference in which they brought together ED and hospital administrators, as well as public health surveillance experts, to discuss how they can use the information they receive from surveillance systems such as Google Flu Trends in a meaningful way that will impact health care.

Pay attention to new tools, research

While ED managers understand that you can never fully predict what is going to happen or what your resource needs will be in a given day or week, they are nonetheless somewhat frustrated that they don't have a better handle on an issue that has a seasonal component to it, explains Richard Rothman, MD, PhD, the senior investigator on this study, and an emergency medicine physician at Johns Hopkins University School of Medicine.

"Every year, emergency departments around the country experience crowding, increased length-of-stay, and increased waiting times during the peak of influenza season, and up until now, there are no great systems out there that allow administrators and the people who figure out clinical staffing needs to predict that," he says. "The information that we gather and send back to health departments and then to the CDC [Centers for Disease Control and Prevention] has intrinsic delays in coming back to the administrators and clinicians on the ground ... so I think people are looking for new systems that may be quicker, easier to access, and more reliable."

Dugas concurs, noting that all EDs have plans in place that they can launch in the case of an emergency. The real question is when to go ahead and implement those plans. "Right now, there is not a great way to do that. The way we tend to do it is based upon what we are actually seeing in our own patient volumes, and then trying to make the best judgment based on that," she says. "The thought of how Google Flu Trends would tie into that is that since it is providing real-time data, we can see exactly what is going on in the country, in the state, or in our city right now, versus the CDC data that is collected, compiled, and gets back to us two weeks later, which doesn't really help us with making decisions at the time things are actually happening."

Eventually, Rothman believes that a combination of multiple systems will provide the kind of informational support that health care providers are looking for, and he notes that Google Flu Trends may contribute, but he is not recommending that ED managers base their planning on the tool just yet. In fact, he is not aware of any providers or organizations that are using the tool in a systematic way. "Individual providers and administrators are probably starting to look at the data, but most likely in the context of the other traditional methods that they rely on," says Rothman. "This is really in the early stages."

Rothman emphasizes that investigators still don't know to what extent people will use the tool, how generalizable the study results will be to other settings, and how reliable the findings are going to be in terms of guiding practices. But he acknowledges there is high interest in the tool, and he thinks ED managers should familiarize themselves with it.

"I would recommend that administrators get on the website, look at the tool, and pay attention to the next phase of research," says Rothman. "I don't think we are at the stage where major decisions should be based on using [the Internet tool], but that may change in the next six months, depending on what we see. I think people just need to keep an ear to the ground in terms of using new tools that are coming up for unexpected, major catastrophic events, such as a significant pandemic."


  1. Dugas A, Hsieh Y-H, Levin S, Rothman R, et al. Google Flu Trends: Correlation with emergency department influenza rates and crowding metrics. Clin Infect Dis. 2012;54:463-469.


  • Andrea Dugas, MD, Emergency Medicine Research Fellow, Johns Hopkins University School of Medicine, Baltimore, MD. E-mail: adugas1@jhmi.edu.
  • Richard Rothman, MD, PhD, Emergency Medicine Physician, Johns Hopkins University School of Medicine, Baltimore, MD. E-mail: rrothma1@jhmi.edu.