Cardiac Arrest Registry to Enhance Survival (CARES) Summary Doc 5.5.09.pdfDisaster Medicine. The CDC and the AHA are working together to reduce the death rate from heart disease and

Cardiac Arrest Registry to Enhance Survival (CARES)

Bryan McNally, MD, MPH (Principal Investigator) Arthur Kellermann, MD, MPH (Co-investigator)

Allison Crouch, MPH (Program Coordinator) Amanda Perez, BA (Data Analyst)

May 1, 2009

INTRODUCTION

Out-of-hospital cardiac arrest (OHCA) is the leading cause of death among adults in the United States and Western countries. It is estimated that approximately 300,000 - 400,000 deaths occur every year. Most of these deaths are due to a fatal heart rhythm disturbance called ventricular fibrillation. Nationally, few communities actively monitor and report their survival rates from OHCA. The range of survival in these communities for ventricular fibrillation is anywhere - from 2% to 35%, a striking difference, since the approach to the care of these patients is uniform and there is no evidence that patients in one part of the country are different biologically from another. The CARES (Cardiac Arrest Registry to Enhance Survival) Program is a collaborative effort of the Centers for Disease Control and Prevention (CDC), the American Heart Association (AHA) and the Emory University Department of Emergency Medicine, Section of Prehospital and Disaster Medicine. The CDC and the AHA are working together to reduce the death rate from heart disease and stroke by 25% from the years 2000-2010. One of the CDC’s initiatives is to develop a model national registry to accurately measure our progress in the treatment of OHCA. Using the Utstein style of statistics for OHCA, CARES is capable of identifying and tracking all cases of cardiac arrest in a defined geographic area. The ultimate goals of CARES is to help local EMS administrators and medical directors identify who is affected, when and where cardiac arrest events occur, which elements of the system are functioning properly and which elements are not, and how changes can be made to improve cardiac arrest outcomes. CARES utilizes an internet database system that reduces time involved in registering events, tracking patient outcomes with hospitals, and response intervals associated with First Responders and EMS providers. Multiple reporting features can be generated and monitored continuously through secure online access by CARES participants and allow for longitudinal, internal benchmarking. Presently, the odds of surviving an episode of out of hospital cardiac arrest in the United States vary by a factor of 10 to 20, depending on the community in which it occurs. Disparities in outcome this extreme are unacceptable and are what the CARES project will be able to identify and allow communities to improve upon. As more communities participate in CARES, confidential, external benchmarking can occur between similar systems across the United States. To date, there are over 20 communities participating in the program nationally and CARES will continue to add more sites as the program continues to grow. Please refer to the CARES website (https://mycares.net) for more information on the program. Do not hesitate to contact the CARES staff ([email protected]) for additional questions and updates.

CARES Current Sites

Cardiac Arrest Registry to Enhance Survival (CARES) OPERATIONAL OVERVIEW

Overview The goal of the CARES program is to establish a model of unifying all essential data elements from three, independent sources, which currently record fractured data of a single, cardiac arrest event. The CARES system is building this model by establishing a relationship with emergency medical services (EMS) agencies, hospitals, and computer-aided dispatch (CAD) systems. Through these sources, access to specific data elements and participation allow for understanding of data flow and the ability to develop an efficient and automatic data collection and outcome reporting system. The collection and reporting system is provided by a restricted-access, secure, internet database developed by Sansio/Scanhealth, Inc. and managed locally by the CARES program staff. All participants can view their individual statistics and de-identified, community-aggregate statistics. EMS agencies EMS providers initiate a CARES event based on criteria set forth initially by the project coordinator and the data dictionary. A contact person at each EMS agency (CARES liaison) ensures adherence to the criteria and provides routine communication with the project coordinator for issues, concerns, and questions. The data can be submitted in three ways: completion of the CARES form, direct entry into the website database, or exporting electronic data from field software programs. Hospitals Hospitals receive notification to provide outcome data through an email, which is automatically generated by the database. These four basic questions include emergency department outcome, hospital outcome, disposition location, and neurological status at time of discharge. Through establishing a relationship with the hospital, orientation to the project and website was conducted, HIPAA concerns were addressed, and adequate identification of patients within their system was obtained. CAD system CAD data is collected through automatic export/import and/or direct entry into the website database. Once contact and agreement is made with each CAD system administrator, three time elements (Call Received, Dispatch, and Arrival) for the EMS and First Responder are sent and/or entered after matching the event based on date, approximate time, and location of the event. In various combinations, multiple CAD systems provide response times for EMS and First Responders. All event and First Responder information is identified by EMS during the initial CARES report. Review The CARES database is monitored daily by the project staff, which continues to address key issues. These include CARES form and dictionary revisions for EMS and hospital users, updates in website data entry and report formats, identifying and meeting with additional communities for program participation, summary analysis of pitfalls and solutions for project ‘tool kit’, and advising agencies with system improvement in areas related to the CARES program.

CARESSurveillanceNetwork

Direct data entry from EMS by desktop, mobile ePCR, or scannable form

CARES staff has access to agency, regional, and national data

Hospital contact enters outcome data at their convenience monthly

CAD data entry when available

1 - Street Address (Where Arrest Occurred)

1 - City 1 - State 1 - Zip Code

Cardiac Arrest Registry

14 - Location Type

Farm

Mine / Quarry

Industrial Place

Recreation/Sport

Street/HwyPublic Building

Educational Inst. Airport

Home/ Residence

Nursing HomeResidence/InstitutionPhysician Office/Clinic

Hospital

Jail

Other

SH3001 (1 of 1), Rev 3, 04/06 Copyright 2008 Sansio (Page 1)

Arrest Information

First Responding Agency

Presumed Cardiac EtiologyTraumaRespiratoryDrowning

17 - Presumed Cardiac Arrest Etiology

ElectrocutionOther

Yes

No

16 - Arrest After Arrival of 911 Responder

Witnessed Arrest

Unwitnessed Arrest

15 - Arrest Witnessed

3 - Last Name2 - First Name

5 - Date of Birth 6 - GenderMaleFemale

YesNo

18 - Resuscitation Attempted by 911 Responder

Ventricular FibrillationVentricular TachycardiaAsystoleIdioventricular/PEA

23 - First Arrest Rhythm of Patient

Unknown Shockable RhythmUnknown Unshockable Rhythm

Dead in Field

Pronounced Dead in ED

Ongoing Resuscitation in ED

27 - End of the Event

Resuscitation Information

First Cardiac Arrest Rhythm of Patient and ROSC Information

Resuscitation not initiated at scene due toobvious signs of death, DNR, resuscitationconsidered futile, or resuscitation is notrequired

Resuscitation terminated at scene due tomedical control order, protocol/policyrequirements completed

Transported to Hospital with or withoutROSC

26 - Out of Hospital DispositionYesNo

24 - ROSC

YesNo

25 - Sustained ROSC

YesNoAED Present but not Used

21 - Was an AED Used During Resuscitation

Lay Person

Lay Person Medical ProviderFirst Responder

22 - Who First Applied Monitor/Defibrillator, AED

AED Malfunctioned

# 0f AED Shocks

# 0f Manual Shocks

Not ApplicableLay PersonLay Person Family MemberLay Person Medical ProviderFirst ResponderResponding EMS Personnel

Hospital Destination

Not Applicable

Lay Person Family Member

Part A : Dem ographic Inform ation

Part B : Run Inform ation9 - Call #8 - Date of Arrest

/ /

4 - AgeDaysMonthsYears

11 - Fire/First Responder 12 - Destination Hospital

AsianBlack/African-American

American-Indian/AlaskaNative Hawaiian/Pacific IslanderWhite

Unknown28 - Race/Ethnicity

Responding EMS Personnel

Hispanic/Latino

20 - Who Initiated CPR

YesNo

30 - Was hypotherm ia care provided in the field

Hypotherm ia Information

31 - When w as hypotherm ia care initiatedDuring ResuscitationAfter Resuscitation

-

Utstein Survival ReportAgency Group: National | Date of Arrest: From 10/01/2005 Through 03/31/2009

-

-

May 08, 2009 ©2000-2009 Sansio. Sansio - myCARES™ 1 of 3

-


-

-


-


-

-


DemographicsSample Data

911 to EMS ArrivalSample Data

911 to EMS DispatchSample Data

NumberofUtsteinSurvivorsbySite:

SiteA‐10 SiteE‐8 SiteI‐3 SiteM‐16SiteB‐2 SiteF‐6 SiteJ‐5 SiteN‐4SiteC‐8 SiteG‐7 SiteK‐4YourSite‐3SiteD‐13SiteH‐4SiteL‐4

NumberofEventsofCardiacENologybySite:

SiteA‐703 SiteE‐139 SiteI‐97 SiteM‐836SiteB‐38 SiteF‐29 SiteJ‐221 SiteN‐77SiteC‐185 SiteG‐178 SiteK‐188YourSite‐67SiteD‐164SiteH‐143SiteL‐138

By Ruth SoRelle, MPH

If you had to have a heart attack,Atlanta may be the best place to do

it. Since June 2006, when the AmericanHeart Association began its RestartAtlanta’s Hearts, nearly 34,000 FultonCounty residents have been trained incardiopulmonary resuscitation.

Add to those another 1,292 peoplewho are trained in the Heartsaver/Auto-mated Electronic Defibrillator programand the 52 AEDs that have been placedat 29 sites over those 18 months, andyou may have a new rival to Seattle’slegendary cardiac survival rate.

Propelling this unprecedented activi-ty in Georgia is the need to improve sur-vival from out-of-hospital cardiac arrestbuttressed by the Cardiac Arrest Reg-istry to Enhance Survival (CARES), acombined effort of the U.S. Centers forDisease Control and Prevention, theEmory University Department of Emer-gency Medicine’s Section of Prehospitaland Disaster Medicine, and the Ameri-can Heart Association. Already, severallarge U.S. cities contribute informationto CARES. Included in the registry orsoon to be included along with Atlanta

are Boston; Columbus and Cincinnati,OH; Louisville, KY; Raleigh-Durham, NC;Nashville, TN; Kansas City, MO; Austinand Houston, TX; Tucson, AZ; and SanFrancisco.

The five-year project, funded with $1.5million, is designed to put some order inthe mishmash of reporting cardiac arrestdata to help individual communities iden-tify and correct their own deficiencies. It

is already having an effect in Atlanta, saidMike Willingham, CCEMT-P, the EMScommunity relations director of theAmerican Heart Association.

Restart Atlanta’s Hearts “would notbe possible without the CARES registrycooperative between Emory Universityand the Centers for Disease Control andPrevention,” he said. “We use the basic

EMERGENCYMEDICINE NEWS

Volume XXX, Number 2February 2008

By Anne Scheck

When two scientists filed a patentapplication for “contrast enhance-

ment” in nuclear magnetic resonanceimaging more than 20 years ago, they citedthe “reduced burden of toxic contrastmaterial” as an unprecedented advantagefor the new and improved compounds.One of them was gadolinium.

Now, however, the very agent thatinspired so much optimism back then isthe subject of a boxed warning by theU.S. Food and Drug Administration.Gadolinium-based contrast agents havebeen linked with a rare but tragic sideeffect — nephrogenic systemic fibrosis

— when used in MRIs for patients withacute or chronic severe renal insufficien-cy, renal insufficiency due to hepato-renal syndrome, and in liver transplantpatients in the perioperative period.

However, “gado,” as radiologists callit, is only one imaging agent to makeheadlines recently. In the past fewmonths, findings linked a possibleheightened cancer risk to radiation dos-ing for CT, results that were published inNovember in the New England Journal

of Medicine (2007;357:2277), and reportsthat iodinated contrast agents used inCT cause more allergic reactions thanthose utilized for MRI. Other accounts

www.EM-News.com

Continued on page 8

FDA Orders Boxed Warning on Gadolinium

CARES Program Working toImprove Cardiac Arrest Survival

CME in every issue!

InFocus See p.by James R. Roberts, MDThis Month:

➤ The Medical Effects of TASERs

Learning to Live with the LLSA See p.by Daniel K. Mullin, MDThis Month:

➤ Children and Brain Injuries; Women andAMI; Trauma and Pregnancy

The only independent emergencymedicine publication with CME!

E M E R G E N C Y M E D I C I N E ’ S O N L Y I N D E P E N D E N T N E W S M A G A Z I N E

11

23

PER

IOD

ICA

LS

LETTERS . . . . . . . . . . . . . . . . . .3

VIEWPOINT . . . . . . . . . . . . . . .4

QUICK CONSULT . . . . . . . . . . .8

INFOCUS . . . . . . . . . . . . . . . .11

ID ROUNDS . . . . . . . . . . . . . .16

TOXICOLOGY ROUNDS . . . . . .20

LEGAL NOTES . . . . . . . . . . . . .20

LEARNING TO LIVE . . . . . . . .23WITH THE LLSA

CAREER SOURCE . . . . . . . . . .26

CLASSIFIED . . . . . . . . . . . . . .28

Continued on page 10

Our

Year of Publication

IN THIS ISSUE

Vin

cent

Gia

rran

o

10 EMN ■ February 2008

dataset from CARES to determine ournumber of cardiac arrests, bystanderCPR rates, bystander AED rates, andoverall survival rates. We would not beable to fund or coordinate this effortwithout these measurable data. It is crit-ical also that CARES measures ‘patientsurvival to discharge,’ allowing us torealize ‘true saves’ from sudden cardiacarrest.”

Bryan McNally, MD,MPH, an assistant professorof emergency medicine atEmory, is the heart ofCARES, said Arthur Keller-mann, MD, MPH, the chairof emergency medicine atEmory and a co-leader ofthe project. When Dr.McNally began the projectin 2004, statistics describingsurvival rates from cardiacarrest nationwide were ahodgepodge. While many

communities over the past 20 years havedeveloped a variety of methods forreducing the time between an individ-ual’s collapse and the start of CPR andeventual defibrillation, few communitiesmonitor how well these strategies work.Those who do found that survival ratesvary widely by factor of 10 or more. With-out a way to evaluate the four criticallinks that make up the chain of survival— immediate notification of EMS, rapidprovision of CPR, early defibrillation,and swift provision of definitive care —improvement is almost impossible.

“Basically a national report wouldallow for several things,” said Dr. McNal-ly. “It would report Utstein survival [aninternational definition of survival forcardiac arrest] in the participating com-munities. Each community has access toits own data, and can compare them to ageneral report as a benchmark.”

Silos of DataCARES has three stores of data, Dr.McNally said. “The idea is to link thethree silos:” computer-assisted dispatchdata, EMS data, and hospital and finaldisposition data. Software developed bySansio makes the information availablethrough a secure web site, and stream-lines bringing the different datasetstogether.

Computer-assisted dispatch systemsautomatically export or directly enterdata into the CARES web site database,he said. The time the call is received andthe dispatch and arrival times for EMSor first responders are logged.

Convincing EMS officials that datacould be collected to comply with thefederal Health Insurance Portability andAccountability Act of 1996 took time, Dr.McNally said. EMS can enter data inthree different ways, one of which is byscanning it optically after the run andthen having it digitally scanned. It auto-matically fills out the web form after thescan.

The web-based system allows EMSpersonnel with a password to enter thedata from any web site, he said. EMSsystems with laptops work with CARESpersonnel to help integrate the CARESdata into the electronic medical record.Then the information can be down-loaded to the registry daily.

C ARESContinued from page 1

Continued on page 15

N e w s

Questions Askedabout MI Patients

■ Did the patient arrive in theemergency department?

■ Where was the patient treatedin the hospital?

■ What was the patient’s finaldisposition?

■ What is the patient’s neurolog-ical status?

February 2008 ■ EMN 15

Each hospital involved in the projecthas a point person who answers fourquestions about each cardiac arrestpatient:

■ Did the patient arrive in the emer-gency department?

■ Where was the patient treated inthe hospital?

■ What was the patient’s final dispo-sition?

■ What is the patient’s neurologicalstatus?

“The beauty is that a lot of definitionhas been worked out beforehand,” Dr.McNally said. “We are collecting data inthe same way on the same platform dur-ing the same time period. We want todevelop a model for a national cardiacdeath registry so we can compare appleswith apples.”

“We would not be able tofund or coordinate thiseffort without thesemeasurable data.”

Mr. Mike Willingham

The registry looks at other responsesas well, he said. “Are people doingbystander CPR? Was an AED used? With70 percent of cardiac arrests occurringat home, you would expect an AED to beavailable then.” Those data are alreadyhaving an effect, he said. The AHARestart Atlanta’s Hearts initiative is help-ing train people to use AEDs and also topinpoint the spots in the communitywhere they would have the most effect.The program also educates the public onthe importance of dialing 9-1-1 when aperson is in cardiac arrest and startingCPR immediately, he said.

Atlanta has used information on thetime between calling 9-1-1 and arrival ofEMS to change its system for dispatch-ing ambulances, he said. Previously, thecentral dispatcher with the policedepartment in Atlanta called FultonCounty, which then dispatched theambulance out of Grady Memorial Hos-pital. Now, the police call Grady EMSdirectly, he said. It saves minutes, and ina year, they will be able to determine theeffect on survival time.

Data on more than 4,500 patientshave been entered into the system so far,and Dr. McNally plans to publish a report

on the audited results as soon as possi-ble. More importantly, however, individ-ual systems can look at their own data,compare it with the national numbers,and figure out ways to strengthen theirown systems.

As Dr. McNally contemplates theend of funding for the project, he hopesto find ways to continue improving thequality of care for out-of-hospital car-diac arrest a continuous process. “Weare excited about it growing,” he said.“We think we are in a position to see itgrow even more and become a sustain-

able model.”Mr. Willingham is even more opti-

mistic. “The American Heart Associationconsiders CARES to be the nationalmodel to collect and analyze data fromcardiac arrest events. We are assisting inthe implementation of CARES nation-wide. Dr. McNally and the CARES staffhave been critical in the improving sur-vival rates in Atlanta and the future sitesof the AHA Restart program,” he said.

Comments about this article? Write to

EMN at [email protected].

C ARESContinued from page 10

N e w s

“We want to develop amodel for a nationalcardiac death registry sowe can compare appleswith apples.”

Dr. Bryan McNally

Emergency Medicine News

encourages an open exchange ofideas, and welcomes your com-ments. Please address correspon-dence to Lisa Hoffman, Editor,Emergency Medicine News, 333Seventh Ave., 19th Fl., New York,NY 10001; [email protected].

Letters to the Editor

Atlanta becomes a template for improving EMSBy Robert Davis, USA TODAY When Atlanta's emergency medical system needed rescuing, Mayor Shirley Franklin started performing CPR — in more ways than one.

In 2003, she started issuing orders that have resulted in lifesaving changes to Atlanta's emergency system. She began looking for ways to hold emergency crews more accountable, and last year she ordered all 8,000 city employees — including herself — to be trained in cardiopulmonary resuscitation.

"All you need to do is save one life and it's worth it," Franklin says. "It's miraculous."

Thanks to those efforts and a program created in Atlanta by Emory University and the Centers for Disease Control and Prevention, the city is saving more residents who collapse of sudden cardiac arrest. Since September 2005, the survival rate for such patients in Atlanta has jumped from less than 3% to 15%. That's well above the 6% to 10% survival rate for most cities that was identified in a 2003 analysis by USA TODAY.

Atlanta's success has made it, and the program it's following, a template for cities trying to improve cardiac-arrest survival rates, an often murky set of figures complicated by communication problems among government agencies.

Several cities — including Houston, Anchorage, Austin, Cincinnati, Kansas City, Mo., Raleigh, N.C., and Tucson — are following in Atlanta's footsteps by signing up for the Emory/CDC program. It allows cities to use its Internet database to combine data from 911 dispatch centers, paramedic run reports and hospital discharge records to reveal more about the performance of EMS units — widely viewed as a key step in improving cardiac survival rates. Many cities have no system to effectively track such rates.

A few cities in the program or planning to join it were identified in the USA TODAY report four years ago as having particularly good systems for tracking emergency crews' performance. Those include Houston, Kansas City, Tucson, Boston, Nashville and San Francisco. Other cities taking part or planning to — such as Atlanta, Austin and Columbus, Ohio — were identified as having less-than-stellar systems for tracking cardiac survival rates.

The program — known as Cardiac Arrest Registry to Enhance Survival, or CARES — is a five-year, $1.5 million CDC project launched three years ago. It was partly inspired by the USA TODAY investigation, which found that emergency medical systems in most of the nation's 50 largest cities were fragmented, inconsistent and slow.

Why the focus on cardiac arrest survival rates and not those from something else, such as car accidents or cancer? Cities use cardiac arrest survival rates as a key measure of EMS performance because such victims typically live or die depending on the care they get in the first minutes after collapse, unlike other emergencies in which survival hinges more on hospital care.

"The system has to deliver in order to save a cardiac arrest victim," says Arthur Kellermann, an emergency physician at Emory University School of Medicine. "If it can deliver in a consistent manner for cardiac arrest victims, there is every reason to expect that it will deliver for trauma victims, asthma victims, women in labor."

More than 250,000 people die outside of hospitals each year when their hearts stop beating. Many are reaching the natural end to battles with disease, but others are healthy when struck by an electrical short circuit of the heart called ventricular fibrillation. "V-fib" can be caused by anything from a blocked coronary artery, to a ball striking the chest, to changes in the heart muscle from an infection.

In 2003, USA TODAY found disparities in emergency medical care across the nation, and said cities that carefully track their EMS performance save many more lives. In most cases, such cities also make a point of teaching residents CPR by, among other things, sending firefighters into homes, churches or businesses to train people.

The reason: If a bystander or acquaintance can quickly perform CPR when a person is stricken with cardiac arrest, they can buy the victim precious time before emergency personnel arrive. Businesses also are encouraged to have defibrillators and people trained to use them so victims can be shocked if rescue crews can't arrive quickly.

Bryan McNally, the emergency physician from Emory Healthcare who heads CARES, told an EMS conference in February that the impetus for the program included USA TODAY's finding that a lack of data regarding EMS responses to cardiac arrest victims is a "major obstacle to improving pre-hospital emergency cardiac care."

Atlanta's huge challenge

Franklin says she learned from USA TODAY's report that Atlanta was losing more than 10 times as many cardiac arrest victims as cities such as Boston, Seattle and Rochester, Minn. The newspaper's analysis ranked cities' EMS efforts in three tiers, with Atlanta's in the bottom tier of cities that had no idea how many lives their rescue units were — or were not — saving.

"Shirley Franklin was furious to see Atlanta as a 'Class C' city," Kellermann says. "She felt it should be in the first tier."

When Franklin took a closer look, what she saw was grim.

From August 2005 through March 2006, her city saved only one person considered by doctors to be among the "most saveable" victims of sudden cardiac arrest. They were deemed saveable because people saw them collapse, and what the victims needed was to be treated quickly with a defibrillator shock to restore their heart's rhythm.

USA TODAY found that in such cases, life and death usually is decided within six minutes of an attack. If the heart is not restarted by then, brain damage can be so severe that the victim is not likely to wake up, even if he or she survives.

"It became really clear to us when we looked at the statistics that the availability of trained personnel close by when somebody is experiencing cardiac arrest can save a life," Franklin says.

She vowed to do more to help the city improve, including enrollment in CARES.

It's paying off. From September 2005 through July 2007, months in which the city has tracked its performance using CARES, 10 of 66 cardiac arrest victims in the "most saveable" category survived with normal brain function.

Atlanta's 15% survival rate is a dramatic improvement, but still well behind leading cities such as Boston, where the survival rate for such cardiac patients is 38%.

One patient's good fortune

The response to save 69-year-old Ronald Williams on May 21 shows how the Atlanta area's system is still moving too slowly to save a life without help from bystanders.

Williams, of Tucker, Ga., was undergoing a stress test in his cardiologist's office when his heart went into V-fib. The medical staff called for help, began CPR and delivered a shock with a defibrillator.

The call for help went first to a 911 center, then to fire department rescuers from Sandy Springs, an Atlanta suburb. By the time paramedics reached Williams and delivered a second shock with their defibrillator, nine minutes had passed since he had gone into arrest.

Williams says he's lucky he was in his doctor's office. "It could have happened anyplace," says the retired aerospace technician, whose blocked arteries were cleared in a hospital after he was revived.

Jing Fang, a physician and researcher in CDC's Division for Heart Disease and Stroke Prevention who is technical director for CARES, says the program ultimately should help save more people like Williams.

Using the system's database, city leaders can track how many cardiac arrest victims their crews tried to save, how many of the victims had their hearts restarted in the field, and how many went home from the hospital with good brain function. The leaders also can see how many victims got help before rescuers arrive. By seeing how each part of the system performed, EMS leaders say they can determine what improvements are needed.

The CARES program allows cities to tell how their crews are performing compared with others in their region and, soon, to the other cities participating nationally.

Some cities that are struggling to determine their cardiac arrest survival rates are not in CARES. In Dallas, officials see CARES as "valuable and laudable," but they are creating their own system to track cardiac arrest survival, says Marshal Isaacs, medical director for the city's fire and rescue units. He says the system could be in place next year.

In Chicago, Philadelphia, El Paso and San Diego, medical directors report having problems getting hospitals to share data on patient survival rates. Jim Dunford, medical director for San Diego's EMS, says a law is needed to force cooperation.

"How can it be that the No. 1 killer of Americans remains heart disease and we still can't accurately measure outcome from cardiac arrest?" he asks.

El Paso's EMS medical director, James "Randy" Loflin, says his city is unable to track survival rates because "hospitals tell us they can't share survival data due to HIPAA," a federal law that protects patient privacy. CARES was designed to share data while complying with the law, McNally says.

'Community response' is key

When Atlanta started crunching its cardiac arrest survival numbers, it became clear that when rescue crews reached a patient, there often were people standing around, unsure how to help.

Only 7% of the city's cardiac arrest victims were getting CPR from bystanders when the CARES program was introduced. Houston, Tucson and other cities that save the most lives in such situations have raised their CPR rates for bystanders through training programs and by having 911 dispatchers give simplified CPR instructions over the phone. Chest compressions alone — even without mouth-to-mouth breathing — can buy minutes for a cardiac arrest victim until rescuers arrive.

"It's not just about streamlining or improving the professional response, it's also about the community response," McNally says. "What is happening before the ambulance or first responders get there? Are people doing CPR?"

When Franklin told city employees to get CPR training, she says, "each of us took a pledge that we would train others."

Atlanta's bystander CPR rate has more than doubled to more than 17%. To give an idea of how far Atlanta has to go to catch up with cities that save the most lives, McNally cites bystander CPR rates of 30% to 40% and higher in places such as Seattle and Boston.

"A lot of us think … the only solution is a doctor," Franklin says. "Having a trained workforce is part of the solution."

current as of November 7, 2008. Online article and related content

http://jama.ama-assn.org/cgi/content/full/300/12/1432

. 2008;300(12):1432-1438 (doi:10.1001/jama.300.12.1432) JAMA

Comilla Sasson; A. J. Hegg; Michelle Macy; et al.

Refractory Out-of-Hospital Cardiac ArrestPrehospital Termination of Resuscitation in Cases of

Correction Contact me if this article is corrected.

Citations Contact me when this article is cited.

This article has been cited 1 time.

Topic collections

Contact me when new articles are published in these topic areas.Emergency Medicine Prognosis/ Outcomes; Arrhythmias; Cardiovascular Disease/ Myocardial Infarction; Critical Care/ Intensive Care Medicine; Adult Critical Care; Cardiovascular System;

the same issueRelated Articles published in

. 2008;300(12):1462.JAMAArthur B. Sanders et al. Surviving Cardiac Arrest: Location, Location, Location

. 2008;300(12):1423.JAMAGraham Nichol et al. Regional Variation in Out-of-Hospital Cardiac Arrest Incidence and Outcome

http://pubs.ama-assn.org/misc/[email protected]

http://jama.com/subscribeSubscribe

[email protected]/E-prints

http://jamaarchives.com/alertsEmail Alerts

at Emory University on November 7, 2008 www.jama.comDownloaded from

http://jama.ama-assn.org/cgi/content/full/300/12/1432

http://jama.ama-assn.org/cgi/alerts/ctalert?alertType=correction&addAlert=correction&saveAlert=no&correction_criteria_value=300/12/1432

http://jama.ama-assn.org/cgi/external_ref?access_num=jama%3B300%2F12%2F1432&link_type=ISI_Citing

http://jama.ama-assn.org/cgi/alerts/ctalert?alertType=citedby&addAlert=cited_by&saveAlert=no&cited_by_criteria_resid=jama;300/12/1432

http://jama.ama-assn.org/cgi/alerts/collalert

http://jama.ama-assn.org/cgi/content/short/300/12/1423

http://jama.ama-assn.org/cgi/content/short/300/12/1462

http://jama.com/subscribe

http://pubs.ama-assn.org/misc/permissions.dtl

http://jamaarchives.com/alerts

mailto:[email protected]

http://jama.ama-assn.org

ORIGINAL CONTRIBUTION

Prehospital Termination of Resuscitationin Cases of Refractory Out-of-HospitalCardiac ArrestComilla Sasson, MD, MSA. J. Hegg, MDMichelle Macy, MDAllison Park, MPHArthur Kellermann, MD, MPHBryan McNally, MD, MPHfor the CARES Surveillance Group

CARDIAC ARREST IS PRIMARILY A

fatal event. It is estimated that166 200 out-of-hospital car-diac events occur each year in

the United States, with approximately60% of these events treated by emer-gency medical services (EMS).1 Re-ported rates of survival following out-of-hospital cardiac arrest (OHCA) varywidely, from 0.2% (Detroit [2007])2 to23% (London, England [2005]).3 Na-tionwide, the median reported sur-vival rate is 6.4%.4 The vast majority ofpatients who survive OHCA are resus-citated at the scene of the cardiac ar-rest and subsequently transported to thehospital for definitive care.5,6

Nevertheless, thepracticeofEMSsys-tems in cases of refractory OHCA varywidely fromagency toagency.Althoughmost systems generally follow the basiclife support (BLS) and advanced lifesupport(ALS)generalresuscitationguide-linesoutlinedby theAmericanHeartAs-sociation,7 there is widespread variabil-ity in their application. In one study,adherencetoAmericanHeartAssociation

guidelines for theout-of-hospital careofcardiac arrest was only 40%.8

During the past 30 years, several re-search teams have sought to define ob-jective clinical criteria to identify pa-tients who likely will not benefit fromrapid transport to the hospital for fur-ther resuscitative efforts.9-18 Despite thisresearch, many EMS systems still ur-gently transport patients with refrac-tory cardiac arrest to the hospital forcontinued resuscitative efforts.19-21

Rapid transport with lights and siren

may pose hazards for EMS personneland the public and should occur onlywhen the risks of high-speed trans-

See also pp 1423 and 1462.

Author Affiliations: Department of Emergency Medi-cine (Drs Sasson and Macy) and Robert Wood JohnsonClinical Scholars Program (Dr Sasson), University ofMichigan, Ann Arbor; Department of Emergency Medi-cine/Internal Medicine, Henry Ford Health System, De-troit, Michigan (Dr Hegg); and CARES Project (Ms Park)and Department of Emergency Medicine (Drs Keller-mann and McNally), Emory University, Atlanta, Georgia.Additional Members of the CARES Surveillance Groupare listed at the end of this article.Corresponding Author: Comilla Sasson, MD, MS, 6312Medical SciencesBldg I,1150WMedicalCenterDr,SPC5604,AnnArbor,MI48109-5604([email protected]).

Context Identifying patients in the out-of-hospital setting who have no realistic hopeof surviving an out-of-hospital cardiac arrest could enhance utilization of scarce healthcare resources.

Objective To validate 2 out-of-hospital termination-of-resuscitation rules devel-oped by the Ontario Prehospital Life Support (OPALS) study group, one for use byresponders providing basic life support (BLS) and the other for those providing ad-vanced life support (ALS).

Design, Setting, and Patients Retrospective cohort study using surveillance dataprospectively submitted by emergency medical systems and hospitals in 8 US cities tothe Cardiac Arrest Registry to Enhance Survival (CARES) between October 1, 2005,and April 30, 2008. Case patients were 7235 adults with out-of-hospital cardiac ar-rest; of these, 5505 met inclusion criteria.

MainOutcomeMeasures Specificityandpositivepredictivevalueofeach termination-of-resuscitation rule for identifyingpatientswho likelywill not survive tohospital discharge.

Results The overall rate of survival to hospital discharge was 7.1% (n=392). Of 2592patients (47.1%)whometBLScriteria for terminationof resuscitationefforts,only5 (0.2%)patients survived to hospital discharge. Of 1192 patients (21.7%) who met ALS criteria,nonesurvivedtohospitaldischarge.TheBLSrulehadaspecificityof0.987(95%confidenceinterval [CI], 0.970-0.996) and a positive predictive value of 0.998 (95% CI, 0.996-0.999)forpredicting lackof survival.TheALSrulehadaspecificityof1.000(95%CI,0.991-1.000)andpositivepredictivevalueof1.000 (95%CI,0.997-1.000) forpredicting lackof survival.

Conclusion In this validation study, the BLS and ALS termination-of-resuscitationrules performed well in identifying patients with out-of-hospital cardiac arrest whohave little or no chance of survival.JAMA. 2008;300(12):1432-1438 www.jama.com

1432 JAMA, September 24, 2008—Vol 300, No. 12 (Reprinted) ©2008 American Medical Association. All rights reserved.



port are justified by the potential ben-efits to the patient.22

The Ontario Prehospital AdvancedLife Support (OPALS) study group hasproposed 2 termination-of-resuscita-tion rules for EMS personnel. Both ruleswere derived from the OPALS group’slarge registry of cardiac arrest cases. Onerule is intended for use by respondersproviding BLS who are equipped withan automated external defibrillator.17

The other, more conservative rule, is in-tended for use by responders provid-ing ALS.23 A patient must meet all of thecriteria included in either rule to war-rant termination of resuscitation in theout-of-hospital setting.

The BLS rule has 3 criteria, whereasthe ALS rule adds 2 additional criteria(BOX). The BLS rule was derived froman initial cohort of 662 patients17 andvalidated in a second cohort of 1240 pa-tients.17,23,24 Of the 776 patients who metBLS criteria for termination of resusci-tation efforts in the out-of-hospital set-ting, only 4 (0.5%) survived to hospi-tal discharge. The researchers calculatedthat had the rule been in effect, nearlytwo-thirds of their patients with OHCAwould have been pronounced dead inthe out-of-hospital setting. High-speed EMS transports of patients withcardiac arrest would have decreasedfrom 100% of cases to 37.4%.17,23,24

The ALS rule was devised to reduceand ideally eliminate the small misclas-sification rate associated with the BLSrule and was derived from a cohort of4673 patients.23 Adding the 2 addi-tional criteria—cardiac arrest not wit-nessed by a bystander and no bystander-administered cardiopulmonaryresuscitation (CPR)—reduced the mis-classification rate to zero. No patientwho met all 5 ALS criteria survived tohospital discharge.23 The authors esti-mated that had this rule been applied,approximately 30% of their patientswith OHCA would have been pro-nounced dead in the out-of-hospital set-ting, and emergency transports of pa-tients with cardiac arrest would havebeen reduced from 100% of cases to70%. Unlike the BLS rule, the ALS rulehas not been validated in a second co-

hort of patients with cardiac arrest. De-spite their respective names, it is im-portant to note that either rule can beapplied by ALS personnel or by BLS per-sonnel equipped with an automated ex-ternal defibrillator.

To independently assess the valid-ity of the BLS and ALS rules for iden-tifying individuals with refractoryOHCA who likely will not benefit fromrapid transport to a hospital for fur-ther attempts at resuscitation, we per-formed a retrospective cohort studybased on data from a large, preexist-ing surveillance registry of 7235 casesof OHCA drawn from 8 US cities.

METHODSThe data used in this analysis were ob-tained from the Cardiac Arrest Regis-try to Enhance Survival (CARES). Thisregistry is designed to help local offi-cials determine how well their com-munity is performing in each link of theAmerican Heart Association “chain ofsurvival.” Detailed information aboutthis registry is published elsewhere.25

From October 1, 2005, to April 30,2008, 8 cities submitted data to CARES:Anchorage, Alaska; metropolitan At-lanta, Georgia; Boston, Massachu-setts; Raleigh, North Carolina; Cincin-nati, Ohio; Columbus, Ohio; Austin,Texas; and Houston, Texas. Becausemetropolitan Atlanta was the first com-munity to report data, a plurality(50.5%) of the cases in this analysis arefrom Georgia. Percentages of cases inthe other states submitting data were32.1% (Texas), 7.4% (Ohio), 4.8%(North Carolina), 3.1% (Massachu-setts), and 2.1% (Alaska). The registryis designed to capture all cardiac ar-rest events in a defined geographic area(city or county) for which the 911 sys-tem was activated. Data analysts con-firmed the capture of all cardiac arrestevents at each city’s 911 center monthlyduring the data review process.

All cases submitted to the registryduring the study interval were eligiblefor inclusion. A cardiac arrest case wasexcluded if (1) EMS personnel deter-mined that arrest was due to a noncar-diac etiology (eg, trauma, electrocu-

tion, drowning, or respiratory); (2)out-of-hospital resuscitation was notattempted based on local EMS proto-cols (eg, obvious signs of death such asrigor mortis, decomposition, lividity);or (3) the patient was younger than 16years.

Data Collection and Processing

Eight US cities prospectively submit-ted data in accordance with the CARESuser agreement. The registry collectsand links a limited standard set of dataelements from 3 sources: 911 call cen-ters, EMS personnel, and receiving hos-pitals. It is designed to provide EMS sys-tem managers with a simple andstandard way to identify cases of OHCA,measure key response-time intervals,document the delivery of important in-terventions, and ascertain outcomes.The registry uses a secure, Health In-surance Portability and Accountabil-ity Act–compliant Web site so that hos-pitals can report each patient’s outcome,length of stay, and cerebral perfor-mance category (CPC) score at the time

Box. Basic Life Supportand Advanced Life SupportTermination-of-ResuscitationRules

Basic Life Support

Event not witnessed by emergencymedical services personnel

No automated external defibrillatorused or manual shock applied in out-of-hospital setting

No return of spontaneous circula-tion in out-of-hospital setting

Advanced Life Support

Event not witnessed by emergencymedical services personnel

No automated external defibrillatorused or manual shock applied in out-of-hospital setting

No return of spontaneous circula-tion in out-of-hospital setting

Arrest not witnessed by bystander

No bystander-administered cardio-pulmonary resuscitation

PREHOSPITAL TERMINATION OF RESUSCITATION IN REFRACTORY CARDIAC ARREST

©2008 American Medical Association. All rights reserved. (Reprinted) JAMA, September 24, 2008—Vol 300, No. 12 1433



of hospital discharge; the registry alsocomplies with the Utstein criteria forcardiac arrest reporting.26

All submitted reports are reviewed bya data analyst. Once essential data ele-ments pass this quality check, indi-vidual identifiers are stripped from therecord and the case is permanently en-tered into the registry database. Once in-dividual identifiers are removed, a casecan be distinguished only on the basisof demographics (age, sex, and race/ethnicity) and the location of the event.Race/ethnicity is determined by the pa-tient, family, or EMS personnel. This in-formation was collected to better char-acterize any potential disparities that mayexist in rates of bystander-administeredCPR, location and types of arrest, or sur-vival to hospital discharge.

A data set containing all events sub-mitted to the registry database be-tween October 1, 2005, and April 30,2008, was secured from the Sansio Cor-poration server, which hosts the data-base. The data set was provided in Ex-cel (Microsoft, Redmond, Washington)and subsequently transferred to Stataversion 10.0 (StataCorp, College Sta-tion, Texas) for statistical analysis.

Data Analysis

The main outcome measures were thespecificity and positive predictive valueof the BLS rule and the ALS rule foridentifying patients with OHCA wholikely would not survive to hospital dis-charge. We estimated the potentialeffect each rule would have on rates ofout-of-hospital pronouncement ofdeath and EMS transports if the rule hadbeen implemented.

Descriptive statistics of the baselinepopulation, including counts, means,and standard deviations, were calcu-lated. The data set was used to calcu-late the sensitivity, specificity, posi-tive predictive value, and negativepredictive value of the BLS and ALSrules for identifying patients who likelywill not survive to hospital discharge.

Consistentwithpreviousstudies,17,24,27

we considered a misclassification rate ofless than1%asanacceptablecriterionfora termination-of-resuscitation rule. Our

sample sizeof1192 for theALSrulepro-vided more than 80% power (1-sided�=.05) to detect a misclassification ratesignificantly lower than 1%; our samplesizeof2592fortheBLSruleprovidedmorethan 90% power to detect that rate.

To determine the diagnostic accu-racy of a decision rule, it must be com-pared against a gold standard. The goldstandard in this study was survival tohospital discharge, as documented inhospital records. A sensitivity analysiswas performed to exclude those pa-tients not transported to the emer-gency department and for whom re-suscitation efforts were terminated inthe out-of-hospital setting as well asthose patients whose clinical out-comes could not be determined.

Human Research Considerations

The CARES database is a continuousquality-improvement tool and surveil-lance registry providing deidentifieddata to help local officials monitor andimprove their provision of out-of-hospital emergency cardiac care. Ev-ery patient received the standard careavailable in his or her community, andno patient received an experimentalintervention. In light of these safe-guards, the institutional review boardsof all participating sites approved thisstudy and determined that this regis-try was exempt from the requirementto secure oral or written consent.28

RESULTSMain Study Results

The full registry included 7235 cases col-lected from 19 EMS agencies and 111hospitals located in 8 US cities. Of the7235 cases, 406 were excluded becauseresuscitation was not attempted by para-medics, 1150 because the arrest was dueto a noncardiac etiology, and 123 be-cause the patients were younger than 16years; an additional 51 were excluded be-cause data documenting clinical out-come were missing. Therefore, a total of5505 cardiac arrest cases met criteria forinclusion in the study. The overall rateof survival to hospital discharge of theremaining cases was 7.1% (n=392). Atotal of 947 (17.2%) were pronounced

dead in the out-of-hospital setting, basedon local EMS agency protocols. TABLE 1displays the demographic, clinical, andEMS characteristics for the study sample.

The FIGURE summarizes the perfor-mance of the BLS and ALS termination-of-resuscitation rules in our studysample. The BLS rule would have rec-ommended out-of-hospital termina-tion of resuscitation in 2592 patients(47.1%). Seventy patients who met BLScriteria for termination of care in theout-of-hospital setting were resusci-tated in the emergency department andadmitted to the hospital, and 5 (0.2%)who met BLS criteria survived to hos-pital discharge. Four of these patientswere documented as having a good CPCscore (score of 1 on a scale of 1-5, where1=good cerebral performance [con-scious, alert, able to work and lead anormal life]; 2=moderate cerebral dis-ability [conscious and able to func-tion independently—dress, travel, pre-pare food—but may have hemiplegia,seizures, or permanent memory ormental changes]; 3=severe cerebraldisability [conscious, dependent on oth-ers for daily support, functions only inan institution or at home with excep-tional family support]; 4=coma or veg-etative state; and 5=death) at the timeof hospital discharge, and 1 had se-vere disability (CPC score of 3). If theBLS rule had been applied, EMS per-sonnel would have terminated out-of-hospital resuscitation efforts in 1645 ad-ditional cases, increasing their rate ofout-of-hospital pronouncement ofdeath from 17% to 47%.

ThemoreconservativeALSrulewouldhave recommended out-of-hospital ter-mination of resuscitation in 1192 cases(21.7%).Twenty-fourof thepatientswhomet ALS criteria for termination of carein the out-of-hospital setting were re-suscitated in the emergency depart-ment, but none survived to hospital dis-charge. If the ALS rule been applied, EMSpersonnel would have terminated resus-citative efforts in 245 additional cases, in-creasing the rate of out-of-hospital pro-nouncement of death from 17% to 22%.

The BLS and ALS rules were bothhighly specific and had high positive





predictive values (TABLE 2). A posi-tive diagnostic result was considered tobe fulfillment of all 3 BLS or all 5 ALScriteria. The rules performed well inpredicting which patients would ben-efit from further efforts at resuscita-tion (specificity) and which would not(positive predictive value). In our study,the BLS rule misclassified 0.2% of cases;the ALS rule misclassified no cases.

Sensitivity Analyses

In our study, 947 patients had resuscita-tion efforts terminated in the out-of-hospital settingbasedon localprotocols.Ofthesepatients,144(15%)didnotmeet1ormoreBLScriteriaand564(59%)didnotmeet1ormoreALScriteria.Exclud-ing these 947 patients would not appre-ciably alter our results. The positive pre-dictivevaluefortheBLSrulewouldchangefrom0.998(95%confidenceinterval[CI],0.996-0.999) to 0.997 (95% CI, 0.993-0.999). The positive predictive value forthe ALS rule would change from 1.000(95%CI,0.997-1.000) to1.000(95%CI,0.995-1.000).

Fifty-one patients who otherwise metinclusion criteria (0.2%) were lost tofollow-up. Eighteen of these 51 pa-tients would have met BLS rule crite-ria, and 9 would have met ALS rule cri-teria. If all of these patients hadsurvived, the BLS rule would have mis-classified a total of 23 patients (0.4%),and the ALS rule would have misclas-sified 9 (0.2%). However, under the as-sumption that the survival rate in thesepatients was similar to that of the over-all study sample (ie, 7.1%), inclusionof these patients would not apprecia-bly alter our findings.

COMMENTWe sought to assess the usefulness of theBLS and ALS termination-of-resuscita-tion rules using data collected in 8 cit-ies across the United States. We deter-mined that both rules accuratelyidentified patients with OHCA who wereunlikely to benefit from rapid transportto the hospital for further attempts at re-suscitation. Had either rule been used toguide decisions to cease resuscitative ef-forts in the out-of-hospital setting, EMS

systems could have substantially re-duced the rate of emergency EMS trans-ports without appreciably worsening therate of cardiac arrest survival.

Given the choice, it is likely that manyEMS medical directors and administra-tors would opt for the more conserva-tive ALS rule, because that rule did notmisclassify any survivors. However,adopting this rule would force someEMS systems to transport more pa-tients than they currently do without ap-

preciably improving patients’ odds ofsurvival. Because receiving bystander-administered CPR warrants transportunder the ALS rule, EMS systems thatoffer dispatcher-assisted CPR wouldhave to transport nearly every patient.

The BLS rule misclassified 5 survi-vors in our study, resulting in a posi-tive predictive value of 99.8%. If therule had been consistently applied inthe cities reporting to our registry, itwould produce a higher rate of out-of-

Table 1. Characteristics of Patients With Out-of-Hospital Cardiac Arrest (N = 5505)

Characteristic No. (%)

Age, mean (SD), y (n = 5470) 64.4 (16.4)

Men (n = 5504) 3286 (60.0)

Race (n = 5446)White 1940 (35.6)

Black 1542 (28.3)

Unknown 1435 (26.3)

Other 529 (9.8)

Location of arrestHome 3557 (64.6)

Nursing home/assisted living 797 (14.5)

Public building 350 (6.3)

Street/highway 251 (4.6)

Other 550 (10.0)

Type of initial rhythm (n = 5503)Ventricular fibrillation/ventricular tachycardia 1006 (18.3)

Unknown shockable rhythm 294 (5.3)

Unknown unshockable rhythm 715 (13.0)

Asystole 2486 (45.2)

Idioventricular/pulseless electrical activity 999 (18.2)

Other 3 (0.01)

Arrest witnessed byBystander (n = 5501)a 2056 (37.4)

EMS/first responder (n = 5503)b 665 (12.1)

CPR attempted by (n = 5479)Bystandera 1123 (20.5)

Medical provider not part of EMS/first responder teama,c 677 (12.4)

Responding EMS personnel 3657 (66.7)

Other 22 (0.4)

Return of spontaneous circulation in out-of-hospital setting (n = 5503)b 1687 (30.7)

Pronounced dead in out-of-hospital setting 947 (17.2)

SurvivalTo hospital admission 1208 (21.9)

To hospital discharge 392 (7.1)

Discharged with good CPC scored 190 (3.5)Abbreviations: CPC, cerebral performance category; CPR, cardiopulmonary resuscitation; EMS, emergency medical

services.aCriteria for advanced life support rule only.bCriteria for basic life support as well as advanced life support rules.cCardiac arrests in which CPR was initiated by medical personnel other than the EMS team/first responder were con-

sidered bystander-administered CPR in application of the termination-of-resuscitation criteria.dConsidered a score of 1 on a scale of 1-5, where 1 = good cerebral performance (conscious, alert, able to work and

lead a normal life); 2 = moderate cerebral disability (conscious and able to function independently [dress, travel, pre-pare food] but may have hemiplegia, seizures, or permanent memory or mental changes); 3 = severe cerebral dis-ability (conscious, dependent on others for daily support, functions only in an institution or at home with exceptionalfamily support); 4 = coma or vegetative state; and 5 = death.





hospital pronouncement of death(47.1%) than the ALS rule (21.7%).However, use of either rule would in-crease out-of-hospital pronounce-ment rates above the aggregate rate of17.2% currently reported by the citiesparticipating in the registry.

In a recent assessment of the BLS rule,researchers from Arizona29 reported thatapplying the rule to 2180 cases of adultnontraumatic cardiac arrest would haveproduced a rate of out-of-hospital pro-nouncement of death of 54%, with a mis-classification rate of 0.09%. A single pa-tient who met criteria for termination ofresuscitation was misclassified; this pa-tient eventually survived to hospital dis-charge and had moderate cerebral dis-ability (CPC score of 2). These findings

are similar to our findings and to thoseof the OPALS study group.24

National organizations such as theNational Association of EMS Physi-cians and the American Heart Associa-tion have promoted guidelines that al-low for termination of futile cardiacresuscitation efforts in the out-of-hospital setting. Both organizations rec-ognize that rapid transport of patientswho have little or no chance of survivalposes risks and generates needless costs.However, reports published as recentlyas 2008 show that adherence to theseguidelines is less than 50%.8,21,30 In a pre-vious survey of EMS personnel, 40% ofemergency medical technicians re-ported that they had provided resusci-tation in situations in which they per-

sonally would not have wanted anythingdone.31 In the same study, almost a quar-ter of all respondents stated that cur-rent guidelines for termination of resus-citation are inadequate.

Without specific, evidence-basedguidance, many EMS personnel prob-ably err on the side of caution by rap-idly transporting patients with refrac-tory OHCA to the hospital. However,rapid transport of patients with lightsand siren, while EMS personnel are at-tempting cardiac resuscitation in theback of the moving vehicle, may poserisks to the personnel. In addition to therisk of injury from motor vehiclecrashes or vehicle-pedestrian colli-sions, this practice may increase the riskof occupational exposure to contami-nated blood and body secretions.32

The decision to transport a patientwith refractory cardiac arrest for addi-tional efforts at the hospital entails sub-stantial costs. These transports removeambulance units from service for an ex-tended period, limiting availability ofEMS resources to other communitymembers with treatable medical emer-gencies. Moreover, the arrival of a patientwith cardiac arrest affects the work-flow of the receiving emergency de-partment, typically resulting in nurses,technicians, and physicians leaving thebedsides of other patients to engage inresuscitation efforts.33 If the emergencydepartment staff initially succeed at re-suscitating the patient, in most cases itis still unlikely that the patient will sur-vive and leave the hospital with goodneurologic status.

Our study has several limitations.First, we used a multicommunity car-diac arrest registry in which case infor-mation was collected prospectively,

Figure. Estimated Potential Outcomes Associated With Application of Basic Life Support (BLS) vsAdvanced Life Support (ALS) Rules and Rates of Out-of-Hospital Termination of Resuscitation

2592 No out-of-hospital returnof spontaneous circulation

3362 No AED used or manualshock applied out of hospital

4840 Arrest not witnessed byEMS personnel

5505 Adults with primary cardiac arrest

1192 Terminate resuscitation/notransport to hospital

2592 Terminate resuscitation/notransport to hospital

1763 Arrest not witnessed bybystander

1192 No CPR initiated bybystander

BLS Rule

BLSandALS

Rules

ALS Rule

AED indicates automated external defibrillator; CPR, cardiopulmonary resuscitation; EMS, emergency medicalservices.

Table 2. Survival to Hospital Discharge Among Patients Meeting Criteria for the Basic Life Support (BLS) vs Advanced Life Support (ALS)Termination-of-Resuscitation Rules

Died Survived Sensitivity (95% CI) Specificity (95% CI) PPV (95% CI) NPV (95% CI)

BLS ruleMet 3/3 criteria 2587 5

0.506 (0.492-0.520) 0.987 (0.970-0.996) 0.998 (0.996-0.999) 0.133 (0.121-0.146)Did not meet criteria 2526 387

ALS ruleMet 5/5 criteria 1192 0

0.233 (0.222-0.245) 1.000 (0.991-1.000) 1.000 (0.997-1.000) 0.091 (0.083-0.100)Did not meet criteria 3921 392

Abbreviations: CI, confidence interval; NPV, negative predictive value; PPV, positive predictive value.





based on the Utstein criteria for stan-dardized reporting of OHCA. However,because the CARES database de-couples the data set from any indi-vidual identifiers, we were unable to ac-cess the original patient records todetermine what factors, if any, contrib-uted to the outcomes of the 5 patients(0.2%) who survived to hospital dis-charge despite meeting all 3 BLS crite-ria for termination of resuscitation. Nev-ertheless, with these cases included, theBLS rule had a positive predictive valueof 99.8% for predicting lack of sur-vival, which is within the acceptablerange used by medical ethicists for de-fining futility.27

Second, although our registry-based data set included cases from 8 dif-ferent US cities, half of all cases weresubmitted by EMS systems operating inmetropolitan Atlanta, Georgia. It istherefore possible that our findingsmight not be generalizable to other partsof the United States. However, the BLSand ALS termination-of-resuscitationrules both performed as well with ourregistry data set as they did with casescollected in Ontario and Arizona.

Third, neither the BLS rule nor theALS rule consider patient age, the exis-tence of comorbid or terminal condi-tions, collapse-to-treatment intervals,or the length of time a resuscitation ef-fort is pursued before it is determinedfutile. These variables were filtered outin the statistical process that derivedeach rule. Research has shown that agealone is a weak predictor of survival.34

Collapse-to-treatment intervals are im-portant but are difficult for bystandersto estimate accurately.9,35,36 One guide-line recommends 20 to 30 minutes ofaggressive out-of-hospital advanced car-diac life support before attempting totransport a patient with OHCA.9 How-ever, most other guidelines do notspecify a time interval or the manda-tory elements of an “adequate” resus-citation effort.

Once a clinical rule has been de-rived and validated, the next step is toassess its implementation. A prospec-tive study is warranted to assess howwell EMS systems implement either the

BLS rule or the ALS rule in clinical prac-tice. To succeed, this will require theunderstanding and commitment of EMSpersonnel, system administrators, andEMS physicians, as well as the sup-port of the community. Because adop-tion of either protocol will shift morepronouncements of death from the hos-pital to the out-of-hospital setting, in-cluding homes, it will be advisable forEMS researchers to evaluate the poten-tial psychological and emotional rami-fications of this practice on EMS per-sonnel and bystanders, including familymembers of the deceased. EMS sys-tems that routinely cease unsuccess-ful cardiac resuscitation efforts in theout-of-hospital setting reportedly havenot encountered problems.37,38 Find-ings from 2 studies have shown thatmany family members are grateful to bespared the additional ordeal of follow-ing their loved one to the emergencydepartment when it is obvious that thepatient has already died.39,40 Educat-ing the public about the appropriate-ness of out-of-hospital termination offutile resuscitation efforts may also helpmanage the community’s sometimesunrealistic expectations about the prob-ability of surviving OHCA.41 More-over, EMS systems adopting a policy ofout-of-hospital termination of resusci-tation for the first time must ensure thatcertain logistical procedures are place,such as the proper procedures to fol-low in pronouncing death and howEMS personnel will notify the policeand mortuary services of the death.

Based on our findings and those ofother research groups,23,24,29 we sug-gest that the BLS rule can be ethicallyapplied by EMS systems in the UnitedStates. The BLS rule identified, with ahigh specificity and high positive pre-dictive value, patients with OHCA whohave a very low likelihood of survivalto hospital discharge. Although someof these patients were resuscitated in theemergency department and spent sev-eral hours or days in the intensive careunit, only 5 (0.2%) identified by the BLSrule survived to hospital discharge.

The combined studies assessing thesetermination-of-resuscitation rules17,23,24,29

included more than 10 000 patients. Ap-plication of the BLS rule would haveincreased rates of out-of-hospital pro-nouncement of death to an estimatednearly 50% in all 3 cohorts, with a 0.1%misclassification rate. Now that the BLSrule has been independently validated inmultiple settings across both the UnitedStates and Canada, it may be time to con-sider standardizing the termination-of-resuscitation guidelines for OHCA.Widespread implementation of eitherrule could materially reduce the riskposed to EMS personnel during high-speed transports, decrease pressure onoverburdened EMS systems, allow emer-gency department staff to focus on pa-tients who have greater odds of sur-vival, and decrease admissions to theintensive care unit of patients with out-of-hospital cardiac arrest who have littleor no chance of surviving to discharge.

Author Contributions: Dr Sasson had full access to allof the data in the study and takes responsibility for theintegrity of the data and the accuracy of the data analysis.Study concept and design: Sasson, Hegg, Kellerman,McNally.Acquisition of data: Park, Kellerman, McNally.Analysis and interpretation of data: Sasson, Hegg,Macy, Kellerman.Drafting of the manuscript: Sasson, Hegg, Macy, Park,Kellerman, McNally.Critical revision of the manuscript for important in-tellectual content: Sasson, Macy, Kellerman, McNally.Statistical analysis: Sasson, Park.Obtained funding: Kellerman, McNally.Administrative, technical, or material support: Sasson,Park, Kellerman, McNally.Financial Disclosures: None reported.Funding/Support: The Cardiac Arrest Registry to En-hance Survival (CARES) is funded by grant support fromthe Centers for Disease Control and Prevention (CDC)and Emory University and is also sponsored by the South-eastern affiliate of the American Heart Association.Role of the Sponsors: The CDC assisted in the designand funding of the surveillance database. The CDC andthe American Heart Association had no role in the de-sign and conduct of the study; the collection, manage-ment, analysis, and interpretation of the data; or thepreparation, review, or approval of the manuscript.Additional Members of the CARES Surveillance Group:Mike Levy, MD (Anchorage Fire Department/EMS [An-chorage, Alaska]); Ian Greenwald, MD, Earl Grubbs,MD, and Eric Ossman, MD (Metropolitan Atlanta,Georgia [Clayton, Cobb, Douglas, Fulton, Gwinnett,Newton, and Rockdale Counties]); Ed Racht, MD (Aus-tin-Travis County EMS [Austin, Texas]); Peter Moyer,MD (Boston EMS [Boston, Massachusetts]); DonaldLocasto, MD (Cincinnati Fire Department–EMS [Cin-cinnati, Ohio]); Michael Keseg, MD (Columbus FireDepartment–EMS [Columbus, Ohio]); David Persse,MD (Houston Fire Department–EMS [Houston,Texas]); and Brent Myers, MD (Wake County EMS [Ra-leigh, North Carolina]).Previous Presentations: This work was presented inpart at the Society for Academic Emergency Medi-cine Annual Meeting; Washington, DC; May 29-June 1, 2008.





Additional Contributions: We thank the 19 EMS agen-cies and 111 hospitals from 8 US cities that comprisethe CARES Surveillance Group. We also thank the Rob-ert Wood Johnson Clinical Scholars Program for its as-sistance.

REFERENCES

1. Rosamond W, Flegal K, Furie K, et al. Heart dis-ease and stroke statistics—2008 update: report fromthe American Heart Association Statistics Committeeand Stroke Statistics Subcommittee [published onlineahead of print December 17, 2007]. Circulation. 2008;117(4):e25-e146. doi:10.1161/CIRCULATIONAHA.107.187998.2. Dunne RB, Compton S, Zalenski RJ, Swor R, WelchR, Bock BF. Outcomes from out-of-hospital cardiacarrest in Detroit. Resuscitation. 2007;72(1):59-65.3. Davies CS, Colquhoun MC, Boyle R, ChamberlainDA. A national programme for on-site defibrillationby lay people in selected high risk areas: initial results.Heart. 2005;91(10):1299-1302.4. Nichol G, Stiell IG, Laupacis A, Pham B, De MaioVJ, Wells GA. A cumulative meta-analysis of the ef-fectiveness of defibrillator-capable emergency medi-cal services for victims of out-of-hospital cardiac arrest.Ann Emerg Med. 1999;34(4, pt 1):517-525.5. Eisenberg M, Bergner L, Hallstrom A. Paramedicprograms and out-of-hospital cardiac arrest, I: fac-tors associated with successful resuscitation. Am J Pub-lic Health. 1979;69(1):30-38.6. Eisenberg MS, Horwood BT, Cummins RO,Reynolds-Haertle R, Hearne TR. Cardiac arrest and re-suscitation: a tale of 29 cities. Ann Emerg Med. 1990;19(2):179-186.7. International Liaison Committee on Resuscitation.2005 International Consensus on Cardiopulmonary Re-suscitation and Emergency Cardiovascular Care Sci-ence with treatment recommendations, part 2: adultbasic life support. Resuscitation. 2005;67(2-3):187-201.8. Kirves H, Skrifvars MB, Vahakuopus M, Ekstrom K,Martikainen M, Castren M. Adherence to resuscitationguidelines during prehospital care of cardiac arrestpatients. Eur J Emerg Med. 2007;14(2):75-81.9. Bailey ED, Wydro GC, Cone DC; National Asso-ciation of EMS Physicians Standards and Clinical Prac-tice Committee. Termination of resuscitation in the pre-hospital setting for adult patients sufferingnontraumatic cardiac arrest. Prehosp Emerg Care. 2000;4(2):190-195.10. Bonnin MJ, Pepe PE, Kimball KT, Clark PS Jr. Dis-tinct criteria for termination of resuscitation in the out-of-hospital setting. JAMA. 1993;270(12):1457-1462.11. Eisenberg M, Hallstrom A, Bergner L. The ACLSscore: predicting survival from out-of-hospital car-diac arrest. JAMA. 1981;246(1):50-52.

12. Kellermann AL, Hackman BB, Somes G. Predict-ing the outcome of unsuccessful prehospital ad-vanced cardiac life support. JAMA. 1993;270(12):1433-1436.13. Lindholm DJ, Campbell JP. Predicting survival fromout-of-hospital cardiac arrest. Prehosp Disaster Med.1998;13(2-4):51-54.14. Marsden AK, Ng GA, Dalziel K, Cobbe SM. Whenis it futile for ambulance personnel to initiate cardio-pulmonary resuscitation? BMJ. 1995;311(6996):49-51.15. Pepe PE, Swor RA, Ornato JP, et al. Resuscita-tion in the out-of-hospital setting: medical futility cri-teria for on-scene pronouncement of death. PrehospEmerg Care. 2001;5(1):79-87.16. Petrie DA, De Maio V, Stiell IG, Dreyer J, MartinM, O’Brien JA. Factors affecting survival after pre-hospital asystolic cardiac arrest in a Basic Life Support-Defibrillation system. CJEM. 2001;3(3):186-192.17. Verbeek PR, Vermeulen MJ, Ali FH, MessengerDW, Summers J, Morrison LJ. Derivation of atermination-of-resuscitation guideline for emer-gency medical technicians using automated externaldefibrillators. Acad Emerg Med. 2002;9(7):671-678.18. Herlitz J, Engdahl J, Svensson L, Young M, AngquistKA, Holmberg S. Can we define patients with no chanceof survival after out-of-hospital cardiac arrest? Heart.2004;90(10):1114-1118.19. Jaslow D, Barbera JA, Johnson E, Moore W. Ter-mination of nontraumatic cardiac arrest resuscitativeefforts in the field: a national survey. Acad Emerg Med.1997;4(9):904-907.20. Lockey AS. Recognition of death and termina-tion of cardiac resuscitation attempts by UK ambu-lance personnel. Emerg Med J. 2002;19(4):345-347.21. O’Brien E, Hendricks D, Cone DC. Field termina-tion of resuscitation: analysis of a newly imple-mented protocol. Prehosp Emerg Care. 2008;12(1):57-61.22. Maguire BJ, Hunting KL, Smith GS, Levick NR. Oc-cupational fatalities in emergency medical services: ahidden crisis. Ann Emerg Med. 2002;40(6):625-632.23. Morrison LJ, Verbeek PR, Vermeulen MJ, et al.Derivation and evaluation of a termination of resus-citation clinical prediction rule for advanced life sup-port providers. Resuscitation. 2007;74(2):266-275.24. Morrison LJ, Visentin LM, Kiss A, et al. Valida-tion of a rule for termination of resuscitation in out-of-hospital cardiac arrest. N Engl J Med. 2006;355(5):478-487.25. SoRelle R CARES program working to improvecardiac arrest survival. Emerg Med News. 2008;30(2):1, 10, 15.26. Jacobs I, Nadkarni V, Bahr J, et al. Cardiac arrestand cardiopulmonary resuscitation outcome reports:update and simplification of the Utstein templates for

resuscitation registries: a statement for healthcare pro-fessionals from a task force of the International Liai-son Committee on Resuscitation (American Heart As-sociation, European Resuscitation Council, AustralianResuscitation Council, New Zealand ResuscitationCouncil, Heart and Stroke Foundation of Canada, In-terAmerican Heart Foundation, Resuscitation Coun-cils of Southern Africa). Circulation. 2004;110(21):3385-3397.27. Schneiderman LJ, Jecker NS, Jonsen AR. Medicalfutility: its meaning and ethical implications. Ann In-tern Med. 1990;112(12):949-954.28. CARES IRB approval and modification. Cardiac Ar-rest Registry to Enhance Survival Web site. https://www.mycares.net/cares_info.jsp. Accessed April 6,2008.29. Richman PB, Vadeboncoeur TF, Chikani V, ClarkL, Bobrow BJ. Independent evaluation of an out-of-hospital termination of resuscitation (TOR) clinical de-cision rule. Acad Emerg Med. 2008;15(6):517-521.30. Eckstein M, Stratton SJ, Chan LS. Termination ofresuscitative efforts for out-of-hospital cardiac arrests.Acad Emerg Med. 2005;12(1):65-70.31. Marco CA, Schears RM. Prehospital resuscita-tion practices: a survey of prehospital providers. J EmergMed. 2003;24(1):101-106.32. Marcus R, Srivastava PU, Bell DM, et al. Occu-pational blood contact among prehospital providers.Ann Emerg Med. 1995;25(6):776-779.33. Cheung M, Morrison L, Verbeek PR. Prehospitalvs. emergency department pronouncement of death:a cost analysis. CJEM. 2001;3(1):19-22.34. Swor RA, Jackson RE, Tintinalli JE, Pirrallo RG. Doesadvanced age matter in outcomes after out-of-hospital cardiac arrest in community-dwelling adults?Acad Emerg Med. 2000;7(7):762-768.35. Isaacs E, Callaham ML. Ability of laypersons toestimate short time intervals in cardiac arrest. AnnEmerg Med. 2000;35(2):147-154.36. Hallstrom AP. Should time from cardiac arrest un-til call to emergency medical services (EMS) be col-lected in EMS research? Crit Care Med. 2002;30(4)(suppl):S127-S130.37. Delbridge TR, Fosnocht DE, Garrison HG, AubleTE. Field termination of unsuccessful out-of-hospitalcardiac arrest resuscitation: acceptance by familymembers. Ann Emerg Med. 1996;27(5):649-654.38. Schmidt TA, Harrahill MA. Family response to out-of-hospital death. Acad Emerg Med. 1995;2(6):513-518.39. Feder S, Matheny RL, Loveless RS Jr, Rea TD. With-holding resuscitation: a new approach to prehospitalend-of-life decisions. Ann Intern Med. 2006;144(9):634-640.40. Hick JL, Mahoney BD, Lappe M. Factors influ-encing hospital transport of patients in continuing car-diac arrest. Ann Emerg Med. 1998;32(1):19-25.41. Marco CA, Schears RM. Societal opinions regard-ing CPR. Am J Emerg Med. 2002;20(3):207-211.





EMERGENCY MEDICAL SERVICES/CONCEPTS

CARES: Cardiac Arrest Registry to Enhance Survival

Bryan McNally, MD, MPHAllen Stokes, BS, EMT-PAllison Crouch, MPHArthur L. Kellermann, MD, MPHFor the CARES Surveillance

Group*

From the Department of Emergency Medicine, Emory University School of Medicine, Atlanta, GA.

Despite 3 decades of scientific progress, rates of survival from out-of-hospital cardiac arrest remain low. TheCardiac Arrest Registry to Enhance Survival (CARES) was created to provide communities with a means to identifycases of out-of-hospital cardiac arrest, measure how well emergency medical services (EMS) perform key elementsof emergency cardiac care, and determine outcomes through hospital discharge. CARES collects data from 3sources—911 dispatch, EMS, and receiving hospitals—and links them to form a single record. Once data entry iscompleted, individual identifiers are stripped from the record. The anonymity of CARES records allows participatingagencies and institutions to compile cases without informed consent. CARES generates standard reports that canbe used to characterize the local epidemiology of cardiac arrest and help managers determine how well EMS isdelivering out-of-hospital cardiac arrest care. After pilot implementation in Atlanta, GA, and subsequent expansionto 7 surrounding counties, CARES was implemented in 22 US cities with a combined population of 14 millionpeople. Additional cities are interested in joining the registry. CARES currently contains more than 13,000 casesand is growing rapidly. [Ann Emerg Med. 2009;xx:xxx.]

0196-0644/$-see front matterCopyright © 2009 by the American College of Emergency Physicians.doi:10.1016/j.annemergmed.2009.03.018

RATIONALEOut-of-hospital cardiac arrest is a leading cause of death in

the United States.1 Successful resuscitation depends on rapidperformance of 4 critical actions: notification of emergencymedical services (EMS), rapid provision of cardiopulmonaryresuscitation (CPR), immediate defibrillation of victims foundin ventricular fibrillation or pulseless ventricular tachycardia,and prompt access to definitive care.2 The collective effect ofthese actions is so important that the American HeartAssociation (AHA) coined the term “the chain of survival” morethan 15 years ago.3 Despite universal acceptance of this conceptand more than 3 decades of scientific progress in understandingthe pathophysiology of cardiac resuscitation, the overall rate ofsurvival from out-of-hospital cardiac arrest in the United Statesremains low. For reasons that are not always clear, community-specific rates of survival vary dramatically.4 Reported rates ofsurvival from witnessed ventricular fibrillation range from 2% inChicago to 46% in Seattle.5,6 However, many EMS systems donot collect data, so their managers have no idea how well orhow poorly they are doing.7

In 2004, the AHA called on researchers to developintegrated methods of data collection that will allow for“[U]niform data collection and tracking of data to facilitate

better continuous quality improvement in hospitals,emergency medical service (EMS) systems, andcommunities.” The AHA noted that this will “. . .enablecomparison across systems for clinical benchmarking toidentify opportunities for improvement.”8 In 2006, theInstitute of Medicine (IOM) Committee on the Future ofEmergency Care in the US Health System observed thatmany EMS agencies cannot document their effect on thecommunities they serve. To strengthen accountability, theIOM challenged EMS to “collect, analyze and useperformance improvement data.”7 In 2008, the AHA calledfor out-of-hospital cardiac arrest to be made a reportableevent. It recommended that any out-of-hospital cardiac arrestreporting system include data on hospital outcomes.9

The Cardiac Arrest Registry to Enhance Survival (CARES)was developed to serve as a central repository of data aboutcardiac arrests from EMS systems throughout the United States.Through its use, it is hoped that EMS systems of any size will beable to review key performance indicators about their ownresponses to out-of-hospital cardiac arrest, including responseintervals, important aspects of care, and patients’ outcomes.Thoughtful evaluation of the information derived from CARES,including comparing local data with similar systems elsewhere,may identify opportunities to improve out-of-hospital care andachieve better outcomes.*All members are listed in the Appendix.

Volume xx, . x : Month Annals of Emergency Medicine 1

ARTICLE IN PRESS

DESIGNCore Data Elements

CARES represents a collaboration of Emory University, theCenters for Disease Control and Prevention (CDC), theSoutheastern Affiliate of the AHA, and the Sansio Corporation(Duluth, MN). It collects out-of-hospital cardiac arrest–relateddata from 3 sources that help define the continuum of care: 911dispatch centers, EMS providers, and receiving hospitals. Thenumber of mandatory data elements is limited to the minimumrequired to characterize an out-of-hospital cardiac arrest eventand document its outcome. Candidate variables were drawnfrom 3 existing data sets that focus on out-of-hospital cardiacarrest (as opposed to inhospital arrests)—the Utstein template,developed by an international consensus panel in 1991 andupdated in 2004,8,10 the National EMS Information System,created by a panel of US experts in 2001,11 and theResuscitation Outcomes Consortium, created by a network ofNational Institutes of Health–funded research institutions in2006.12 Each variable was vetted by our team, with input froman ad hoc panel of national EMS experts (Appendix E1,available online at http://www.annemergmed.com). Fourcriteria were used: Is the variable necessary to characterize anout-of-hospital cardiac arrest event? Can it be clearly defined?Can it be objectively measured and reliably reported? Is itrelevant to influencing or documenting the outcome? For thevariable to be included in CARES, the answer to all 4 questionshad to be yes.

This iterative process produced a core data set of 37elements, which is fewer than the elements required by Utstein,Resuscitation Outcomes Consortium, or National EMSInformation System (Table 1). For the sake of consistency,National EMS Information System data definitions were usedwherever possible. The CARES data dictionary is available athttps://mycares.net. A sample form is depicted in Figure 1.

Data CollectionOnce we defined our core data set, we devised a streamlined

process to collect it. Because out-of-hospital cardiac arresttreatment spans a continuum of care, 3 sources are required tofully characterize an out-of-hospital cardiac arrest event: (1) 911call center data (to document incident address and dispatch andunit arrival times), (2) EMS data (to capture presenting cardiacrhythm and the performance of key interventions) and (3)receiving hospital data (to document outcome at hospitaldischarge). To compile and collate data, CARES uses a secure,Health Insurance Portability and Accountability Act (HIPAA)-compliant, Internet-based data collection system developed bySansio, Inc. and managed by CARES staff. This system providesparticipating 911 call centers, EMS agencies, and hospitals withready access to their own data and the aggregate output of theregistry, but users cannot scrutinize another agency’s orhospital’s data. Unauthorized parties and outside entities cannotgain access to individual records or the performance profiles ofparticipating organizations.

Because 911 call centers, EMS agencies, and hospitalscapture and compile data differently, special procedures wereneeded to collect information from each source.

911 Data elements include incident address, the time each911 call was received, the time the first responder and thetransporting EMS unit were dispatched, and time each unitreached the scene. 911 Center managers can either upload datain batches or submit individual records for data entry. Toensure accurate reporting from communities with multiplepublic safety answering points, clocks are periodicallysynchronized. Once the relevant information is entered, CARESautomatically calculates call processing and response-timeintervals.

EMS providers initiate the reporting process because they arebest positioned to determine that an out-of-hospital cardiacarrest has occurred. To qualify as a case, an out-of-hospitalcardiac arrest must be “worked,” meaning CPR was performedor defibrillation was attempted. If the victim is obviously deador EMS personnel honor a “do not attempt resuscitation”request, the case is excluded. To be entered in the registry, anevent must be presumptively due to cardiac disease. Cases ofasphyxia, drowning, electrocution, drug overdose, trauma, or aprimary respiratory event are excluded.

EMS providers may submit data in any of 3 ways: A 1-pagepaper form can be fed into a special scanner that autopopulatesCARES data fields, EMS services equipped with tablet or laptopcomputers can directly upload data to the CARES Web site, orEMS providers can directly file a CARES report online, using aWeb-based application with embedded error checks to enhancethe accuracy of data entry.

To ensure that no cases are missed, each participating servicecompares the official records of “worked” resuscitations to casesentered in the CARES database. If a case was not reported orkey data are missing, the providers who treated the patient arecontacted and asked to resolve the problem.

Hospital outcome is widely regarded to be the criterionstandard for assessing the effectiveness of out-of-hospital cardiaccare.8 In most communities, the primary obstacle to securinghospital data appears to be unwarranted fears of violating federalpatient confidentiality laws.7 Some administrators resistproviding outcome data because they fear that the process willbe burdensome and costly. To allay these fears, we created asimple, HIPAA-compliant process that allows hospitals tosubmit data through the Web. Only 5 data elements arerequested: emergency department (ED) outcome (admitted,died, or transferred); receiving hospital outcome information(same); patient disposition (discharged home, transferred to asecond hospital or rehabilitation unit, released to the morgue);Cerebral Performance Score at discharge, a simple measure offunctional status; and hypothermia treatment (if provided).

To the degree possible, reporting is automated to expedite datacollection. For example, whenever paramedics notify CARES of anew case, the system automatically e-mails the designated liaison atthe receiving hospital that a CARES patient was transported to his

Cardiac Arrest Registry to Enhance Survival McNally et al

2 Annals of Emergency Medicine Volume xx, . x : Month

ARTICLE IN PRESS

http://www.annemergmed.com

https://mycares.net

Table 1. Data elements comparison: CARES, National EMS Information System, Utstein, and Resuscitation Outcomes Consortiumdata sets related to cardiac arrest information.10

Data Element

National EMSInformation

System, NationalEMS Data Set

UtsteinCardiacArrest

NomenclatureCARES Surveillance

Registry

Resuscitation OutcomesConsortium, Clinical

Trial

Patient informationPatient’s first (given) name Yes No Yes* NoPatient’s middle initial/name Yes No No NoPatient’s last (family) name Yes No Yes* NoPatient’s age Yes Yes Yes YesAge units Yes No Yes YesPatient’s date of birth Yes Yes Yes* NoPatient’s sex Yes Yes Yes YesPatient’s race/ethnicity Yes No Yes Yes

Scene informationIncident address Yes No Yes NoIncident city Yes No Yes NoIncident county Yes No No NoIncident state Yes No Yes NoIncident ZIP code Yes No Yes NoScene zone number Yes No No NoScene GPS location Yes No No NoGeospatial location of event No No No Yes

†

Incident location type Yes Yes Yes Yes911 responder information

Date of cardiac arrest Yes Yes Yes YesEMS agency ID Yes No Yes YesEMS vehicle ID Yes No No YesNumber of EMS personnel Yes No No YesService level Yes No No YesEMS call number Yes No Yes NoFire/first responder service ID Yes No Yes NoOther EMS agencies at scene Yes No No NoOther services at scene Yes No No No

Situation informationComplaint reported by dispatch Yes No No NoCardiac arrest (yes/no) Yes Yes No NoCardiac arrest cause Yes Yes Yes YesContributing factors No No No YesEvidence of implantable defibrillator No No No YesArrest witnessed Yes “arrest

witnessed by”Yes Yes Yes

Arrest after arrival of 911 responder Yes (by abovequestion)

Yes Yes Yes

Resuscitation attempted by 911 responder Yes Yes Yes YesCPR information

Who initiated CPR Yes (by otherCPR questions)

No Yes Yes (by other CPRquestions)

Bystander CPR Yes (by “previousaid preformedby”)

Yes Yes (by “whoinitiated CPR”)

Yes

CPR performed Yes Yes Yes (by “whoinitiated CPR”)

Yes

Chest compressions No Yes No NoChest compressions (by EMS) No No No YesAssisted ventilation Yes (by below

question)Yes No No

Type(s) of airway intervention(s) used Yes No No Yes‡

Reason CPR discontinued Yes No No NoMonitor/defibrillator information

First monitored rhythm of the patient Yes No No NoFirst arrest rhythm of the patient No Yes Yes No

McNally et al Cardiac Arrest Registry to Enhance Survival


ARTICLE IN PRESS

Table 1, continued. Data elements comparison: CARES, National EMS Information System, Utstein, and Resuscitation OutcomesConsortium data sets related to cardiac arrest information.10 (continued)

Data Element





Registry


Trial

First arrest rhythm of the patient (monitored by EMS) No No No YesWho first applied AED or monitor/defibrillator Yes (by other

monitor/defibrillatorquestions)

No Yes Yes (by other monitor/defibrillator questions)

Bystander AED/defibrillator applied Yes (by “previousaid preformedby”)

Yes Yes (by “who firstapplied AED ormonitor/defibrillator”)

Yes

Defibrillation attempted Yes Yes Yes (by “who firstapplied AED ormonitor/defibrillator”)

Yes

Was an AED used during resuscitation No No Yes NoAdvanced monitor used Yes No No Yes#

Of AED shocks Yes No Yes No#

Of manual shocks Yes No Yes NoOther out-of-hospital treatment information

Previous aid preformed by (before EMS) Yes No No NoOutcome of previous aid (before EMS) Yes No No NoDrug therapies Yes Yes No YesIV/IO line used Yes No No YesHypothermia Yes No Yes

§Yes

ROSC Yes Yes Yes YesSustained ROSC No Yes Yes Yes

Out-of-hospital disposition informationOut-of-hospital disposition No No Yes Yes (by 4 questions

below)Died at scene or en route Yes No Yes (by “out-of-

hospitaldisposition”)

Yes

Reason not treated or why treatment halted Yes No No YesAlive and not transported by EMS to hospital/ED No No No YesTransported to hospital/mode of transport Yes No Yes (by “out-of-

hospitaldisposition”)

Yes

End of event/patient status at ED No Yes Yes YesHospital information

Date of ED/hospital arrival Yes No Yes YesDestination hospital ID Yes No Yes YesCardiac rhythm on arrival at destination Yes No No NoED outcome Yes No Yes NoInterhospital transfer to another acute hospital No No Yes (by “ED

outcome”)Yes

Date of interhospital transfer to another acutehospital

No No No Yes

Hospital outcome Yes Yes Yes YesDate of discharge or death No Yes No YesDischarge destination No No Yes NoNeurologic status at discharge Yes Yes Yes No

CAD time informationAgency CAD ID No No Yes NoCAD call number Yes No Yes NoEms call received time

�Yes Yes Yes Yes

EMS dispatched time Yes No Yes YesEMS unit notified by dispatch time Yes No No NoEMS en route time Yes Yes

¶No No

EMS on scene time Yes Yes¶

Yes YesEMS on scene time (for first ALS crew) No No No YesEMS arrived at patient time Yes Yes

#No Yes

EMS left scene time Yes Yes#

No Yes



ARTICLE IN PRESS

or her facility. When the liaison accesses the hospital’s site throughthe CARES portal, the case is there. It takes only a few clicks toreport the patient’s outcome. If no outcome data are reported by 15days after an out-of-hospital cardiac arrest, the system automaticallysends 2 additional e-mail reminders to the hospital during the nextfew days. If outcome data are still missing 30 days after the event,the system prompts a CARES administrator to call the hospitalcontact.

Data Linkage and StorageIf a community’s computer aided dispatch system

autogenerates a common case identifier, CARES uses thisnumber to link 911, EMS, and hospital reports to form acomplete record. If a community’s computer aided dispatchsystem lacks this capability, reports are linked by probabilisticmatching. In most instances, reports are matched by patientname and the incident’s time, date, and location. Once linkage

Table 1, continued. Data elements comparison: CARES, National EMS Information System, Utstein, and Resuscitation OutcomesConsortium data sets related to cardiac arrest information.10 (continued)

Data Element





Registry


Trial

Transfer to aeromedical time Yes No No YesEMS arrived at hospital/ED time Yes Yes

#No Yes

EMS back in service time Yes No No NoEMS cancelled time Yes No No NoFirst responder call received time

�No No Yes No

First responder dispatched time No No Yes NoFirst responder on scene time Yes “date/time

initialresponderarrived onscene”

No Yes No

Treatment time informationIncident or onset date/time Yes No No NoEstimated time of arrest before EMS arrival Yes No No NoTime of arrest (if EMS witnessed) No Yes No YesTime arrest confirmed No Yes

#No No

Time of first rhythm analysis/assessment of needfor CPR

No Yes No No

Time of first CPR No Yes No NoTime of first CPR (by EMS) Yes No No YesTime when vascular access achieved No Yes

¶No No

Time tracheal intubation achieved No Yes#

No NoTime of EMS shock assessment Yes No No YesTime of first defibrillation No Yes No NoTime of first defibrillation (by EMS) Yes No No YesTime when medication given Yes Yes

#No No

Time of ROSC No Yes¶

No YesTime of end of ROSC No Yes

#No No

Time of awakening No Yes#

No NoTime CPR stopped/terminated Yes “time

resuscitationdiscontinued”

Yes No Yes

Time of death No Yes No No

IV, Intravenous; IO, intraosseous, ROSC, return of spontaneous circulation.This table does not represent a full, comprehensive summary of all the fields and subfields in the National EMS Information System or Resuscitation Outcomes Con-sortium. It is intended only for general comparison between the data sets.*CARES only temporarily collects patient identifiers (name and date of birth) to ensure accuracy in matching the EMS records with the hospital outcomes. After therecord is deemed complete by the CARES staff, the record is deidentified of name and date of birth.†The Resuscitation Outcomes Consortium uses a combination of census tract, latitude/longitude, and universal transverse Mercator (UTM) to capture the location ofthe cardiac arrest incident.‡The Resuscitation Outcomes Consortium collects data for bag-valve-mask, esophageal-tracheal twin-lumen airway device (Combitube)/laryngeal mask airway/esopha-geal obturator airway, oral endotracheal, cricothyrotomy, ventilator, nasal endotracheal, continuous positive airway pressure, and rapid sequence intubation.§CARES will start collecting hypothermia data starting summer 2008.�Call received time (for EMS and first responder) may vary by organization, depending on local CAD structure. This may be a primary or secondary public safety answer-ing point call time, depending on what times are locally recorded or available.¶Utstein: Indicates recommended as supplemental data element.#Utstein: Indicates in original data set but removed in newer version.



ARTICLE IN PRESS

is complete, the record is reviewed for accuracy. At this point,the case’s individual identifiers are permanently stripped fromthe record. Once a record is permanently entered in the CARESdatabase, no further edits can be made.

The anonymity of CARES data is a major advantage.Because the registry contains only deidentified reports and usesHIPAA-compliant procedures, Emory University’s institutionalreview board ruled that participating agencies and institutionsmay collect data without informed consent.

Analysis and Production of ReportsCARES allows its users to generate standard reports or

conduct custom queries. Because CARES data are anonymous,aggregated reports can be shared without fear of compromisinga patient’s or service’s confidentiality. Each organization has24-hour access to its own profile but not individual access to

others. However, participating organizations are able tocompare their performance against that of their peers bybenchmarking against a regularly generated summary nationalreport. Outside parties cannot gain access to patient- or service-specific data. They can only examine reports at the aggregatelevel.

Standard CARES output includes histograms thatgraphically depict 911 call handling and response intervals.“Call to Dispatch” histogram provides insight into the efficiencyof a community’s 911 call center. The “Dispatch to UnitArrival,” histogram depicts the intervals required for the firstpublic safety provider to reach the scene (whether it is fire,police, or EMS). The “Call to Unit Arrival” interval depicts thesystem’s overall response time (Figure 2).

CARES also captures incident location data so that EMSsystem managers can assess where and in what type of locations

1 - Street Address (Where Arrest Occurred)

1 - City 1 - State 1 - Zip Code

Cardiac Arrest Registry

14 - Location Type

Farm

Mine / Quarry

Industrial Place

Recreation/Sport

Street/HwyPublic Building

Educational Inst. Airport

Home/ Residence

Nursing HomeResidence/InstitutionPhysician Office/Clinic

Hospital

Jail

Other

SH3001 (1 of 1), Rev 3, 04/06 Copyright 2008 Sansio (Page 1)

Arrest Information

First Responding Agency

Presumed Cardiac Etiology

Trauma

Respiratory

Drowning

17 - Presumed Cardiac Arrest Etiology

Electrocution

Other

Yes

No

16 - Arrest After Arrival of EMS

Witnessed Arrest

Unwitnessed Arrest

15 - Arrest Witnessed

3 - Last Name2 - First Name

5 - Date of Birth 6 - GenderMale

Female

YesNo

18 - Resuscitation Attempted by EMS

Ventricular Fibrillation

Ventricular Tachycardia

Asystole

Idioventricular/PEA

23 - First Arrest Rhythm of Patient

Unknown Shockable Rhythm

Unknown Unshockable Rhythm

Dead in Field

Pronounced Dead in ED

Ongoing Resuscitation in ED

27 - End of the Event

Resuscitation Information

First Cardiac Arrest Rhythm of Patient and ROSC Information

Resuscitation not initiated at scene due toobvious signs of death, DNR, resuscitationconsidered futile, or resuscitation is notrequired

Resuscitation terminated at scene due tomedical control order, protocol/policyrequirements completed

Transported to Hospital with or withoutROSC

26 - Out of Hospital DispositionYes

No

24 - ROSC

Yes

No

25 - Sustained ROSC

YesNoAED Present but not Used

21 - Was an AED Used During Resuscitation

Lay Person

Lay Person Medical ProviderFirst Responder

22 - Who First Applied Monitor/Defibrillator, AED

AED Malfunctioned

# 0f AED Shocks

# 0f Manual Shocks

Not Applicable

Lay Person


Lay Person Medical Provider

First Responder


Hospital Destination

Not Applicable


Part A : Dem ographic Inform ation

Part B : Run Inform ation9 - Call #8 - Date of Arrest

/ /

4 - AgeDays

Months

Years

11 - Fire/First Responder 12 - Destination Hospital

AsianBlack/African-American

American-Indian/AlaskaNative Hawaiian/Pacific IslanderWhite

Unknown28 - Race/Ethnicity


Hispanic/Latino

20 - Who Initiated CPR

Figure 1. CARES EMS data form.



ARTICLE IN PRESS

out-of-hospital cardiac arrest occurs. This can be used to guidecommunity-level interventions, such as neighborhood-specificCPR training programs, public access automated externaldefibrillators, and first-responding fire companies. In the nearfuture, geographic information system software may be added toenable participating cities to map events and display them inrelation to neighborhood landmarks and various physical andsocial characteristics.

PILOT IMPLEMENTATIONOnce the registry was created, we first implemented it in

Atlanta, GA, a city served by a single advanced life support EMSservice (Grady EMS), 1 first-responding fire department (theAtlanta Fire Department), a single 911 computer aided dispatchcenter (Fulton County 911), and a relatively small number ofhospitals. After a 12-month period of pilot testing andrefinement, during which CARES collected and linked data onmore than 600 cases of out-of-hospital cardiac arrest, theregistry was expanded to 7 surrounding counties, an areaencompassing 2,000 square miles and a population of roughly 3million. This phase provided a stringent test of the registry’scapacity to identify and link cases in a complex,multijurisdictional setting.

Historically, hospitals and EMS services in metro Atlanta werereluctant to share data. The sheer number and diversity of 911centers, EMS agencies, fire departments, and hospitals posed asignificant challenge. To satisfy the needs of participants, CAREShad to be flexible enough to accommodate multiple methods of

data submission and still link records from different jurisdictions.Nearly every EMS service that participated in this phase of theproject works with multiple first-responding services. Many receivecalls from more than one 911 center. Strategies were devised toovercome all of these difficulties (Appendix E2, available online athttp://www.annemergmed.com)

EARLY EXPANSIONFor many EMS systems, this is their first effort to

systematically collect and evaluate data related to the continuumof care for out-of-hospital cardiac arrest.

CARES was expanded to more than 21 communities outsideGeorgia. Additional communities have expressed an interest injoining. Currently, CARES is processing data from 32 computeraided dispatch centers, 108 first-responder agencies, 31 EMSsystems, and more than 200 receiving hospitals in 13 states(Figure 3). Approximately 14 million people reside incommunities that are participating in CARES.

CASE REPORTINGTo date, CARES has compiled data on more than 13,000

cases of out-of-hospital cardiac arrest. The typical time fromevent to EMS data submission is 1 week (daily in a few siteswith electronic patient care records). Delinquent reports andcases with missing data are requested at the end of each calendarmonth. According to initial audits and feedback from CARESsite directors in 15 participating cities, we estimate that theregistry is capturing at least 95% percent of eligible cases. With

Figure 2. A sample EMS response time report. The respective intervals are provided that allow for the dispatch processtime to be viewed separately from the ambulance response time. A report for first responder times using these sameintervals can also be generated.



ARTICLE IN PRESS


the exception of race, a social construct largely determined byself-report,13 every element of the CARES data set is beingconsistently reported. The proportion of missing data rangesfrom a high of 25% for victim race to a low of 1% for patientname (permanently stripped from each record once data linkageis completed). Hospitals are reporting outcome data in morethan 99% of cases. In 40 months of operation, there have beenno breaches of confidentiality and no patient or family membercomplaints. No call center, EMS service, hospital, orcommunity has withdrawn from the registry.

LIMITATIONSCARES’ greatest strength—simplicity—is also its greatest

weakness. It is designed to collect the minimum number ofvariables required to characterize the treatment and outcome ofan out-of-hospital cardiac arrest event. Participating systemsmay choose to collect additional data elements for their ownuse, but CARES is not intended to be all things to all people.

Determining that a cardiac arrest is due to heart disease issubjective. The AHA has conceded that “no uniformly applieddefinition of cardiac arrest exists.”9 In general, out-of-hospitalcardiac arrest is ascribed to heart disease unless there is anobvious alternate cause, such as major trauma, drowning,electrocution, drug overdose, asphyxia, or exsanguination.Because few victims are autopsied, it is often impossible toassign a definitive cause of death.

The anonymous nature of CARES records precludes theregistry from being used for patient-specific inquiries or casereviews. CARES cannot be used to contact survivors for long-

term follow-up. Had we desired this capability, we would havehad to secure written, informed consent from every patient orthe legally authorized representative, a prohibitive task.

DISCUSSIONThe CDC defines public health surveillance as “the ongoing,

systematic collection, analysis, interpretation, and disseminationof data about a health-related event for use in public healthaction to reduce morbidity and mortality and to improvehealth.”14 According to the CDC, surveillance serves a numberof public health functions, including supporting case detectionand public health interventions, estimating the effect of a diseaseor injury, portraying the natural history of a health condition,determining the geographic distribution and spread of illness,generating hypotheses and stimulating research, evaluatingprevention and control measures, and facilitating planning.15

CARES satisfies many aspects of this definition. We designed itto facilitate the timely collection, analysis, and interpretation ofout-of-hospital cardiac arrest data so it can be quickly sharedwith those who need to know.16

The concept of making a noncommunicable disease such asout-of-hospital cardiac arrest a “reportable event” may beperplexing to some. Out-of-hospital cardiac arrest is notimmediately preventable and, for individual victims, the event isfinal. But out-of-hospital cardiac arrest reporting can produceseveral benefits. First, consistent collection of performance andoutcome data can pinpoint opportunities to improve treatmentof the next cardiac arrest victim. Second, the system’sbenchmarking features, which allow communities to compare

CARES SITES 2009

Designates sites active prior to December 1, 2008 Designates sites added since December 1, 2008

Figure 3. CARES national map of current sites with active data collection.



ARTICLE IN PRESS

how well they perform relative to other communities, may fostergreater support for EMS. And third, ongoing monitoring of theincidence and outcomes of this particularly lethal form of heartdisease may help local public health officials focus the public’sattention on the value of primary prevention. Over time,CARES may allow participating communities to document thebeneficial effects of population interventions, such as enactmentof an indoor air ordinance or improved access to primary care.17

As a public health surveillance system, CARES differs fromtypical research databases (Appendix E3, available online athttp://www.annemergmed.com). Research databases aredesigned to answer hypotheses, rather than monitor populationhealth.18 They are generally more complex, costly, and laborintensive than a typical surveillance system.

The Resuscitation Outcomes Consortium is a prime exampleof a research database. 12 Created by a national network ofresearch institutions, with funding provided by the NationalInstitutes of Health, the Resuscitation Outcomes Consortium isdesigned to test promising treatments for out-of-hospital cardiacarrest and life-threatening trauma. CARES was not created forthis purpose. It was designed to help communities monitor theirsuccess at achieving each link in the AHA’s “chain of survival”3

to identify opportunities to improve rates of out-of-hospitalcardiac arrest survival. If and when the Resuscitation OutcomesConsortium identifies a new technique for treating out-of-hospital cardiac arrest, it will be important to quicklydisseminate it throughout the nation and assess its effect.CARES is ideally suited to this task. Its core data set can besupplemented at any time. For example, several cities currentlyparticipating in CARES have recently added a single dataelement to document out-of-hospital initiation of hypothermia.

There is a clear need for tools to help communities monitor andimprove delivery of emergency cardiac care. AHA has observed that“[t]here is extensive variation in reported outcome after the onset ofcardiac arrest. . .This disparity in survival rates reemphasizes that aneffective EMS system can decrease disability and death from acutecardiovascular events in the out of hospital setting.”9 Nichol et al4

observed that if out-of-hospital cardiac arrest survival could beincreased throughout the United States to the level reported by thehighest-performing community in their study of out-of-hospitalcardiac arrest outcomes, 15,000 premature deaths could beprevented each year.4

The National EMS Information System is not a researchdatabase, but it also differs markedly from CARES. UnlikeCARES, which focuses on a single condition, the National EMSInformation System is designed to capture the full spectrum ofEMS encounters. It is therefore larger and much more complex.And the National EMS Information System has an importantlimitation: unlike CARES and the Resuscitation OutcomesConsortium, it does not collect data on hospital outcomes.

Communities wary of tackling the National EMSInformation System may make their initial foray into electronicdata collection through CARES. A favorable experience mayencourage them to adopt the National EMS Information

System at a later date. In the meantime, many EMS managersmay opt to use out-of-hospital cardiac arrest as a sentinelcondition to assess their system’s capabilities to manage time-critical events. In Emergency Medical Services at the Crossroads,7

the IOM Committee on the Future of Emergency Careendorsed this approach:

“While a full-blown data collection and performancemeasurement and reporting system is the desired ultimate outcome,the committee believes a handful of key indicators of regionalsystem performance should be collected and promulgated as soonas possible. These could include, for example, indicators of 911 callprocessing times, EMS response times for critical calls, andambulance diversions. In addition, consensus measurement of EMSoutcomes could be applied to two to three sentinel conditions. Forexample, emergency and trauma care systems across the countrymight be tasked with providing data on conditions such as cardiacarrest, pediatric respiratory arrest, and major blunt trauma withshock. Data from different system components would allowresearchers to measure how well the system performs at each level ofcare (911, first response, and ED).”7

CONCLUSIONAn adage from the business world states, “You can’t manage

what you can’t measure.”19 This concept is equally applicable toEMS. The IOM Committee on the Future of Emergency Careenvisions a day when our nation will be served by “coordinated,regionalized and accountable emergency care systems.”7 Thiscannot happen without uniform procedures to collect and analyzeperformance improvement data. We created CARES to helpcommunities of every size assess their treatment of out-of-hospitalcardiac arrest, with the goal of improving rates of survival. For someof our participating systems, CARES represents their first effort tosystematically collect, link, and evaluate data related to care of out-of-hospital cardiac arrest. We hope that others will follow.

Technical assistance from the CDC staff was provided byMichael D. Matters, MD, MPH, and Jing Fang, MD, MPH.

Supervising editor: Theodore R. Delbridge, MD, MPH

Funding and support: By Annals policy, all authors are requiredto disclose any and all commercial, financial, and otherrelationships in any way related to the subject of this articlethat might create any potential conflict of interest. See theManuscript Submission Agreement in this issue for examplesof specific conflicts covered by this statement. Funding for theCardiac Arrest Registry to Enhance Survival (CARES) isprovided by cooperative agreement from the Centers forDisease Control and Prevention grant number MM-0917-05/05. The American Association of Medical Colleges is the grantadministrator for CARES.

Publication dates: Received for publication June 29, 2008.Revisions received December 30, 2008, and March 7, 2009.Accepted for publication March 11, 2009.



ARTICLE IN PRESS


Address for reprints: Bryan McNally, MD, MPH, Department ofEmergency Medicine, Emory University School of Medicine,531 Asbury Circle Annex, Suite N340, Atlanta, GA 30322;E-mail [email protected].

REFERENCES1. Gillum RF. Sudden coronary death in the United States.

Circulation. 1989;79:756-765.2. 2005 American Heart Association guidelines for cardiopulmonary

resuscitation and emergency cardiovascular care. Circ. 2005;112:iv-20.

3. Cummins RO, Ornato JP, Thies WH, et al. Improving survival fromsudden cardiac arrest: the “chain of survival” concept. Astatement for health professionals from the Advanced CardiacLife Support Subcommittee and the Emergency Cardiac CareCommittee, American Heart Association. Circulation. 1991;83:1832-1847.

4. Nichol G, Thomas E, Callaway C, et al. Regional variation in out-of-hospital cardiac arrest incidence and outcome. JAMA. 2008;300:1423-1431.

5. Rea TD, Eisenberg MS, Sinibaldi G, et al. Incidence of EMS-treated out-of-hospital cardiac arrest in the United States.Resuscitation. 2004;63:17-24.

6. Eisenberg MS, Horwood BT, Cummins RO, et al. Cardiac arrestand resuscitation: a tale of 29 cities. Ann Emerg Med. 1990;19:179-186.

7. Institute of Medicine of the National Academies Committee onthe Future of Emergency Care in the United States HealthSystem. Emergency Medical Services at the Crossroads.Washington, DC: National Academies Press; 2007:207-230.

8. Jacobs I, Nadkarni V, Bahr J, et al; International LiaisonCommittee on Resuscitation; American Heart Association;European Resuscitation Council; Australian Resuscitation Council;New Zealand Resuscitation Council; Heart and Stroke Foundationof Canada; InterAmerican Heart Foundation; ResuscitationCouncils of Southern Africa; ILCOR Task Force on Cardiac Arrestand Cardiopulmonary Resuscitation Outcomes. Cardiac arrest andcardiopulmonary resuscitation outcome reports: update andsimplification of the Utstein templates for resuscitation registries:a statement for healthcare professionals from a task force of theInternational Liaison Committee on Resuscitation (American HeartAssociation, European Resuscitation Council, AustralianResuscitation Council, New Zealand Resuscitation Council, Heartand Stroke Foundation of Canada, InterAmerican HeartFoundation, Resuscitation Councils of Southern Africa).Circulation. 2004;110:3385-3397.

9. Nichol G, Rumsfeld J, Eigel B, et al. Essentials features ofdesignating out-of-hospital cardiac arrest as a reportable event.Circulation. 2008;117:2299-2308.

10. Cummins RO, Chamberlain DA, Abramson NS, et al.Recommended guidelines for uniform reporting of data from out-of-hospital cardiac arrest: the Utstein Style. A statement forhealth professionals from a task force of the American HeartAssociation, the European Resuscitation Council, the Heart andStroke Foundation of Canada, and the Australian ResuscitationCouncil. Circulation. 1991;84:960-975.

11. National Emergency Medical Services Information System.NEMSIS Data Dictionary version 2.2.1. National EmergencyMedical Services Information System Web site. Available at:

http://www.nemsis.org/softwareDevelopers/downloads/datasetDictionaries.html. Accessed March 6, 2007.

12. National Institutes of Health News. New federally funded researchprogram aims to improve survival from cardiac arrest and severetrauma. US Department of Health and Human Services. ReleasedMarch 24, 2006.

13. Mays VM, Ponce NA, Washington DL, et al. Classification of raceand ethnicity: implications for public health. Annu Rev PublicHealth. 2003;24:83.

14. Centers for Disease Control and Prevention. Updated guidelinesfor evaluating public health surveillance systems:recommendations from the guidelines working group. MMWRMorb Mortal Wkly Rep. 2001;50:RR-1.

15. Teutsch SM, Churchill RE. Principles and Practice of Public HealthSurveillance. 2nd ed. Oxford, England: Oxford University Press;2000.

16. Teutsch S, Thacker S. Planning a public health surveillancesystem. Epidemiol Bull. 1995;16:21.

17. Cesaroni G, Forastiere F, Agabiti N, et al. Effect of the Italiansmoking ban on population rates of acute coronary events.Circulation. 2008;117:1183-1188.

18. Thacker SB, Berkelman RL. Public health surveillance in theUnited States. Epidemiol Rev. 1988;10:164-190.

19. Alter M. You can’t manage it if you can’t measure it. NationalFederation of Small Business. Available at: http://www.nfib.com/object/IO_23800.html. Accessed April 3, 2009.

APPENDIX. CARES Surveillance Group.Mike Levy, MD (Anchorage, Alaska (Anchorage Fire Depart-

ment/EMS, Anchorage, AK); Ian Greenwald, MD, Earl Grubbs,MD, Eric Ossmann, MD (Clayton, Cobb, Douglas, Fulton,Gwinnett, Newton, and Rockdale Counties, Metropolitan At-lanta, Georgia ); Louis Gonzales, BS, EMT-P (Austin TravisCounty EMS, Austin, TX); David Hall, MD (Baytown EMS,Baytown, TX); Peter Moyer, MD, Sophia Dyer, MD, (BostonEMS, Boston, MA); Donald Locasto, MD (Cincinnati Fire De-partment—EMS, Cincinnati, OH); David Ross, MD (Multi-agency EMS, Colorado Springs, CO); David Keseg, MD (Co-lumbus Fire Department–EMS, Columbus, OH); Joe Barger,MD (Contra Costa County EMS, Contra Costa County, CA);Chris Colwell, MD (Denver Health Paramedic Division, Denver,CO); James Leaming, MD (Penn State Life Lion EMS, Hershey,PA); David Persse, MD (Houston Fire Department—EMS,Houston, TX); Joseph Salomone, MD (Metropolitan AmbulanceServices Trust, Kansas City, MO); Dave Slattery, MD (Las VegasFire-Rescue, Las Vegas, NV); Corey Slovis, MD (Nashville FireDepartment—EMS, Nashville, TN); Bob Swor, MD (WilliamBeaumont Hospital, Oakland County, MI); Brent Myers, MD(Wake County EMS, Raleigh-Durham, NC); Joe Ornato, MD(Richmond Ambulance Authority, Richmond, VA); Karl Sporer,MD (San Francisco Fire Department—EMS, San Francisco,CA); Jeff Luther, MD (Sioux Falls REMSA, Sioux Falls, SD); BenOsborne, MD (Multi-agency EMS, Springfield, MA); AngeloSalvucci, MD(Multi-Agency EMS, Venture County, CA).



ARTICLE IN PRESS

mailto:[email protected]

http://www.nemsis.org/softwareDevelopers/downloads/datasetDictionaries.html

http://www.nemsis.org/softwareDevelopers/downloads/datasetDictionaries.html

http://www.nfib.com/object/IO_23800.html

http://www.nfib.com/object/IO_23800.html

APPENDIX E1. CARES data element ad hoc panel (blinded).

Mickey Eisenberg, MD, PhD, University of Washington School of MedicineRay Fowler, MD, University of Texas SouthwesternIan Greenwald, MD, Emory University School of MedicineRichard Hunt, MD, Center for Disease Control and PreventionGreg Mears, MD, University of North Carolina School of MedicinePeter Moyer, MD, PhD, Boston University School of MedicineEric Ossmann, MD, Emory University School of MedicineArthur Yancey, MD, MPH, Emory University School of Medicine

Appendix E2. Diagram: importance of data elements and linkage CARES.

Volume xx, . x : Month Annals of Emergency Medicine 10.e1

ARTICLE IN PRESS

Appendix E3. Distinctions between public health surveillance and epidemiologic research.

Public Health Surveillance Epidemiologic Research

Reason for initiating datacollection

Problem detectionProblem descriptionIdentify cases for epidemiologic studiesMay be legally requiredMonitor geographic and temporal

trends in disease occurrence

Hypothesis testingProblem description

Frequency of datacollection

Ongoing Usually time-limited

Method of data collection Established systems or proceduresMany persons involvedTraditionally depends on voluntary

participation

Special procedures tailored to hypotheses orquestions of interest

Fewer persons involvedDepends on paid, supervised employees

Amount of data collectedper case

Usually minimal Can be considerable and usually detailed

Completeness of datacollected

Often incomplete Usually complete

Analysis of data Traditionally simplePrimarily to detect change in incidenceUsually historical comparison groups

Can be complexHypothesis testing often requires statistical

methodsConcurrent controls

Dissemination of data TimelyRegularReview in public health agencyTargeted to public health and clinical

audience

Not timelySporadicExternal reviewTargeted to academic as well as public health

and clinical audienceUse of data Identifies a problem

Triggers interventionSuggest hypothesesCommonly used to evaluate programsEstimates magnitude of a problem

Describes a problem in detailProvides etiologic informationTests hypotheses, suggests additional

hypothesesLess often used to evaluate programs

Adapted from Table 1 in Thacker SB, Berkelman RL. Public health surveillance in the United States. Epidemiol Rev. 1988;10:164.

10.e2 Annals of Emergency Medicine Volume xx, . x : Month

ARTICLE IN PRESS

Documents

Cardiac Arrest Registry to Enhance Survival (CARES) Summary Doc 5.5.09.pdfDisaster Medicine. The CDC and the AHA are working together to reduce the death rate from heart disease and