What can you learn from your data?
Using the information you have
Just how many data points are collected for every patient every day in your hospital? It numbers in the thousands, and much of the information is never used. Imagine what you could do if you harnessed all the data you have at your fingertips.
That’s a discussion the Institute of Medicine’s (IOM’s) Clinical Effectiveness Research Innovation Collaborative (CERIC) hoped to spur when it released a report in mid-April called “Making the Case for Continuous Learning from Routinely Collected Data” (available online at http://www.iom.edu/Global/Perspectives/2013/~/media/Files/Perspectives-Files/2013/Discussion-Papers/VSRT-MakingtheCase.pdf).
While it’s not directly aimed at hospital quality and safety professionals, there are still lessons in it for them, says one of the writers, Michael Murray, PharmD, MPH, executive director at the Regenstrief Center for Healthcare Effectiveness Research at Purdue University in Indianapolis.
Secondary uses of data
“Our impetus was to get the public thinking about the secondary uses of their data — that other people can benefit from it,” he says. But he thinks that people who deal with data every day also fail to understand the potential at their fingertips.
In early May, for example, he was talking to a patient safety group. “They had all this data on smart pump alerts from a variety of facilities,” says Murray. “They had formed a collaborative to determine how each facility can best use alert data and are working to be able to view it through a digital hub. They can see where alerts are coming from, what kind of drugs the patients are on.”
The group was riffing on issues like how to couple that data with outcomes data from the hospitals. “You could perhaps determine what happened after the alert. Think of what a rich data source that might be — you could end up preventing harm at other hospitals that use similar technology.”
Doing something like this is usually reserved for the largest institutions that have data analysis units on site, Murray says. But groups of hospitals that might not have that capability internally might be able to work with a university that can help mine large sets of data. “There is a lot of cleaning, organizing, interpretation, and analysis you have to do to make sense of big data,” he says. “Many organizations just do not have the individuals or systems to accomplish that without help.”
This has been done in other industries for a long time. The report makes note of the use of data in the financial industries, which can take large data sets and integrate them easily. Healthcare has lagged, though.
With the creation of the Office of the National Coordinator at the Department of Health and Human Services, however, there has been a push to catch healthcare up.
An example cited in the report is the health information exchange created in Indiana, in part by the Regenstrief Institute (http://www.ihie.org).
More disparate networks, like the HMO Research Network, a group of about 20 healthcare organizations, collaborate using administrative and clinical data for research purposes. The information is not centrally stored, thus protecting the integrity of proprietary information while still aiding the group as a whole.
A thousand points of data
So how do you figure out what you have and what you can do with it? Start by getting a group together from every single department that collects data of any kind, says another one of the report authors, Eric B. Larson, MD, MPH, MACP, vice president for research and executive director and senior investigator at the Group Health Research Institute in Seattle. Have each department take an inventory of what it collects and what it does with each piece of data. It could be that you have no idea of the vast quantity of information you have access to.
At the same time, Murray says you should be thinking about how that data might help answer specific questions you have or solve a particular problem. “If you have a broad question, you might want to partner with another institution that can help you analyze a lot of data. There is funding available for this kind of project.”
A broad project might be combining information on a particular drug. Most new drugs are tested on just a few thousand people. Being able to combine data from many different sources might help determine if there are problems with a drug that you do not see in 3,000 test subjects, but you do in millions of doses.
For narrower concerns that relate more to your own particular set of circumstances, it’s possible you can collect and analyze the data without the help of specialists with fancy computers, Murray says.
Larson says he sees quality professionals benefitting from looking for ways to draw off routinely collected information to address quality concerns.
Maybe you can use natural language processing programs to look for notations by physicians relating to a particular kind of complication. This is the kind of thing you should be thinking of, he says.
“It’s already happening in billing. They are much more efficient because of electronic health records. So why can’t you use them the same way to detect safety problems or to monitor for problem areas — a ward where there are more falls, a shift that has a cluster of nosocomial infections.”
There is no limit to what you can do, Larson says. “Put your effort where you have the greatest patient risk. You should be looking at your data and finding ways to exploit it. What do you collect ever day, and what is the potential for its use? Think big picture.”
For more information on this topic, contact:
• Michael Murray, Pharm.D., MPH, Executive Director, Regenstrief Center for Healthcare Effectiveness Research, Indianapolis, IN. Telephone: (317) 423-5504. Email: firstname.lastname@example.org.
• Eric B. Larson, MD, MPH, MACP, Vice President for Research, Group Health Executive Director & Senior Investigator, Group Health Research Institute, Seattle, WA. Telephone: (206) 287-2988. Email: email@example.com.