Hospitalized patients, in contrast, typically have chronic illnesses, infections, or other conditions that lead to the shutdown of the heart. The risks and benefits of cooling such patients are unclear.
"I don't think we have the data or evidence to tell us which patients would benefit from it," said lead author Mark E. Mikkelsen, a Penn critical care physician and epidemiologist.
The study, published recently in the journal Critical Care Medicine, analyzed an American Heart Association database of 538 hospitals committed to following resuscitation guidelines, which say that cooling should be "considered" for in-hospital cardiac arrest.
The researchers found that only 2 percent of the 67,498 patients whose hearts stopped in those hospitals between 2003 and 2009 were cooled, with the rest receiving standard cardiopulmonary resuscitation.
Even when hypothermia was used, the target temperature - about 91 degrees Fahrenheit within 24 hours - was not reached in 44 percent of patients, and 18 percent were overcooled.
Mikkelsen believes cooling should be more common than it is. At Penn, about 20 percent of in-hospital cardiac arrest patients are cooled, and the vast majority reach the target temperature, he said.
"This is a potentially effective therapy that in general is not being used," he said.
Studies have clearly shown cooling helps when cardiac arrest occurs at home, work, or elsewhere in the community. Typically, the heart stops because of an abnormal rhythm, an unrecognized heart defect, or a clogged artery that triggers a heart attack.
Restoring blood flow is lifesaving, but it unleashes chemical reactions that are especially damaging to the brain.
Slowly cooling and rewarming the body is the only proven way to relieve this "reperfusion injury" and reduce the chance of neurological problems.
The cooling process is complex, and it has risks, including blood-sugar fluctuations and infection. Even so, most hospitals and many paramedic squads are now equipped to do it. In New Jersey, for example, a study found 68 percent of hospitals had a hypothermia program in 2011, and 14 percent were planning one.
The new study found that more than half of the 538 hospitals had cooled at least one in-hospital cardiac arrest patient. The study couldn't tell whether cooling helped, but the researchers speculated that outcomes were often unsatisfactory.
They suggest a clinical trial to compare cooled patients with those who get standard care. "We urgently need it," Mikkelsen said.
Contact Marie McCullough at 215-854-2720 or firstname.lastname@example.org.