heartbeat

Data analytics charts new path to heart attack survival

As more federal, state and local agencies embrace data analytics, they’re finding it more efficient and less expensive to host their data sets in the cloud. Indeed, they’re even offloading the process of scrubbing and standardizing data to analytics vendors that offer guidance in how to apply state-of-the-art data mining approaches to critical governmental problems.

“Vendors are building large data centers for organizations that want to get their business analytics in the cloud,” said Shawn McCarthy, an IDC Government Insights research director. “I’m anticipating that more agencies will want to do this because they don’t need to be installing, maintaining, upgrading or patching their software.’’

One such organization was Wake County Emergency Medical Services in Raleigh, N.C., which has reported life-saving returns on its investment in cloud-based data analytics.

Wake County EMS tracks cardiac arrest outcomes. Recently, it changed its recommendations for how long to continue CPR on cardiac arrest patients based on a data analysis project that it outsourced to SAS Advanced Analytics Lab for State and Local Government.

Beginning in 2005, when hypothermia treatment became commonplace for cardiac arrest patients, Wake County emergency response crews found they needed to administer CPR to patients for much longer time periods – up to an hour, instead of the 25 minutes that was the industry best practice.

In order to come up with new CPR guidelines for its crews, Wake County EMS hired SAS to analyze a database of all of its responses to cardiac arrest patients from 2005 through 2012, including the amount of time spent on CPR and the patients’ outcomes. 

“SAS was able to do all the regressions, the statistical analysis and to control for all the variables. They gave us a plot of our data that basically showed we have a logical survival rate to continue CPR for an hour,’’ said Dr. Brent Myers, Medical Director of Wake County EMS. “We were able to use the data to change our approach to cardiac arrest patients. We are presenting our findings to the national association for EMS physicians in January in Tucson. We think our data will also influence EMS care elsewhere once people see our results.’’

Wake County, which includes North Carolina’s capital Raleigh, has a population of 950,000. Wake County EMS responds to approximately 88,000 calls per year, including 500 cardiac arrests that require CPR. The agency operates 40-plus ambulances manned by paramedics.

Wake County EMS has maintained a database about its cardiac arrest patients and whether EMS crews restored a pulse doing CPR for 20 years. But in 2005, Wake County EMS signed cooperative agreements with area hospitals to receive data on how these patients ultimately fared from their medical treatment. This is the database that Wake County EMS hired SAS to analyze.

“We thought our data was clean and ready to go until SAS looked at it,’’ Myers said. “It took about six months for them to clean it up and analyze it.’’

Myers said the initial project with SAS cost around $100,000. Now, Wake County EMS is in discussions with SAS to host all of the agency’s emergency call-related data on a permanent basis.

As a result of the initial project, Wake County EMS has changed its recommendations for cardiac patients. If a patient has a flatline on a heart monitor, emergency crews will stop CPR after 25 minutes. However, if they are able to get the heart re-started through CPR, they will continue CPR for as long as an hour without worrying that the patient will end up in a vegetative state.

“When we looked back at our data from 2005 to 2012, it showed that if we had stopped administering CPR at 25 minutes, which was the old guideline, we would have terminated 100 people that ultimately left the hospital,’’ Myers said. “Now we have the data to make the change to administering CPR for an hour. In a very data-driven way, we feel confident that we are practicing evidence-based medicine as we go forward.’’

Wake County EMS hopes to conduct additional data analytics projects with SAS that could affect how its paramedics provide medical care. For example, the agency would like to study the medical outcomes of patients that it treats for asthma-related problems. It also would like to track patients who call 911 with nonemergency related issues.

“Between 25 and 30 percent of the patients who call us are not emergency patients. Instead, they need regular medical care. What we would like to do with those patients is discover why they are not getting the regular medical care that is needed and whether they qualify for Medicaid and don’t know it,’’ Myers explained. “We’re hoping to reduce the number of calls and costs associated with noncritical care, so these patients can go to a clinic instead of using emergency care.’’

Myers said that the type of data analytics that SAS is providing Wake County EMS represents a shift in emergency care.

“It’s a significant change to have integrated data across the continuum of care,’’ Myers said. “At EMS, we had all of our data, but how does that help the patient if we don’t know ultimately how the patient did once they left the hospital? What’s different now is the idea of integrated data that’s not siloed at the hospital or the clinic. That’s where we need the power of analytics to help us interpret all of that data. What’s new is that we’re not just talking about how EMS did, but we’re talking about how the patient did.’’

Reader Comments

Wed, Feb 12, 2014 Chris

I read an article in the WSJ (96 Minutes Without a Heartbeat 17 May 2011) that talked about using a capnograph to measure CO2 as a way to determine viability compared to a time based determination. This article seems to support that longer applications of CPR are justified based on data instead of a device.

Thu, Jan 16, 2014 Chris

Below is the link to an WSJ article about capnograph use in determining when determining the lenght of time that CPR is useful. http://online.wsj.com/news/articles/SB10001424052748704281504576327592395352666

Please post your comments here. Comments are moderated, so they may not appear immediately after submitting. We will not post comments that we consider abusive or off-topic.

Please type the letters/numbers you see above