Major CPR and emergency-response program finds no significant survival gain

More than 200,000 people received CPR training in a North Carolina effort that also sought to strengthen emergency dispatch and first response. Yet the five-year RACE-CARS trial found no statistically significant improvement in patients leaving hospital alive with favorable brain function, according to results released November 8.

For anyone who might one day help a person who collapses, the distinction matters: this was a test of an added community program. Patients were not assigned to receive CPR or go without it. The comparison was whether extra education and emergency-response support could improve outcomes beyond the care communities already provided.

Researchers randomly assigned 60 counties: 31 received the program and 29 continued existing practices. The study ran from July 2021 through June 2026 and included 22,459 adult cardiac arrests. It excluded arrests witnessed by emergency medical services personnel and those in hospitals, nursing homes and other healthcare facilities. The American Heart Association’s account describes the trial population and design.

Assigning whole counties lets researchers test changes to a shared rescue system. It also makes the question demanding: can a program reach enough people and change enough real emergencies to improve patient outcomes across a community?

The package combined free community training in CPR and automated external defibrillator use with instruction for emergency dispatchers and responders. Dispatchers were trained to recognize cardiac arrest and guide callers through CPR. The American College of Cardiology summarizes those components.

The results describe different stages of that effort:

Training reach and early defibrillation figures come from the AHA report; ACC also reports the CPR and survival percentages.

The training percentage counts residents; the response and survival percentages count cardiac arrests. They cannot be read as successive steps involving the same people. Nor does 4.8% mean everyone else had never learned CPR—it measures this program’s reach.

The survival result needs equal care. Failure to demonstrate a statistically significant improvement does not establish that the approaches are equivalent. The slightly lower reported survival percentage also does not, by itself, establish harm.

There was one potentially encouraging later signal. In the final year, patients in program communities had 33% higher odds of receiving bystander CPR. But the AHA identifies this as an exploratory analysis. It raises a question worth following, rather than demonstrating a survival benefit or proving that a longer program would work. ACC reports the final-year finding and researchers’ interpretation.

Implementation varied between counties. Some training and engagement information was self-reported, and emergency-call-center staffing shortages and turnover complicated delivery. These are limitations, not proven explanations for the outcome. The AHA details those difficulties.

Independent specialists made a similar distinction in comments to STAT. University of Michigan cardiologist Amrish Deshmukh noted that interventions can increase CPR or defibrillator use without demonstrating improved survival. Cardiologist Brahmajee Nallamothu emphasized the unresolved challenge of helping people retain training and apply it during an emergency. Neither participated in the trial. Their observations provide context, not another test of the program.

For bystanders, current guidance still emphasizes acting promptly. Resuscitation Council UK’s 2025 guidelines advise calling emergency services immediately for an unresponsive person, starting CPR when breathing is abnormal and following dispatcher and defibrillator instructions. The emergency number is 999 in the UK and 911 in the US.

Those guidelines also make a practical point about the system around the rescuer: defibrillators should be clearly marked, unlocked and accessible around the clock. Knowing how to help and being able to reach the equipment are separate requirements when minutes matter.