What would I have done? That was the question I found myself pondering when I heard the 911 call made in early March from a California independent-living facility about an 87-year-old resident who had collapsed. In the recording, which gained national attention, an emergency dispatcher implores the caller -- an employee of the facility -- to give cardiopulmonary resuscitation to the barely breathing woman. The caller refused, saying, "We can't do CPR at this facility." The woman died, and public outrage ensued. Pundits wondered: How could anyone be so callous?But was it callous? Are there times when not performing CPR is the right choice?The decision to administer CPR should not be automatic, says David Newman, director of clinical research in the department of medicine at Mount Sinai School of Medicine in New York. "We should think of CPR as an invasive, burdensome, punishing procedure," he says. It's appropriate in some cases but not in others. Think of it like surgery, Newman says: "We would never say that a bed-bound, chronically ill, debilitated person should immediately go into surgery if they have an emergency."Many elderly or chronically ill people have advance directives stating that they don't want end-of-life interventions. The family of the California woman released a statement saying that she had not wanted any medical interventions to prolong her life: "We understand that the 911 tape of this event has caused concern, but our family knows that mom had full knowledge of the limitations of Glenwood Gardens and is at peace."The right conditionsCPR was developed to rescue what the technique's pioneers called "the hearts too good to die." The idea was that a person whose heart has a temporary problem -- such as a heart attack or sudden cardiac arrest caused by an arrhythmia -- has a good chance of recovery if you can just get the heart started again, Newman says.Say you're walking through the airport and an apparently healthy person collapses in front of you and shows no signs of life. In this situation, there's no question -- start CPR if you know how, says Jeffrey Pellegrino, an emergency medical technician and member of the American Red Cross Scientific Advisory Council. Cases where someone is struck by lightning or appears to drown or becomes hypothermic -- those are other scenarios where CPR might be effective, he says.With recent studies showing that chest-compression-only CPR is nearly as effective as traditional CPR given with mouth-to-mouth breaths, the American Heart Association and Red Cross both advise people unable or unwilling to perform full CPR to perform chest compressions.Here's how to do them: Put your hands in the center of the victim's chest, one on top of the other, and push hard and fast, says Michael Sayre, immediate past chairman of the heart association's Emergency Cardiovascular Care Committee and a professor of emergency medicine at the University of Washington in Seattle. Don't worry about how deep you're pushing -- unless you're an NFL linebacker, you probably can't push too hard. Aim for 100 compressions per minute (that's the basic pace of the Bee Gees song Stayin' Alive) and keep going until emergency personnel arrive or you're too tired to continue.For CPR to save a life, several things must happen. First, someone responds to the victim very quickly and calls 911. Then the responder provides CPR, and -- ideally -- the victim quickly receives treatment with an automated external defibrillator to shock the heart and restore a normal rhythm. Finally, the victim gets transported to advanced life support, especially if an AED isn't available. "Is CPR effective? It's effective when these pieces are together," Pellegrino says.A 2000 study found that a program that trained casino security guards to do CPR in conjunction with AEDs resulted in 53 percent of the patients surviving to be discharged from the hospital. Speed was an important factor: 74 percent of those who started AED treatment within three minutes of their collapse survived.But those are ideal conditions with trained, fast responders (the casino had video cameras everywhere) and the best equipment, says George Lundberg, a former editor of the Journal of the American Medical Association who is now editor at large for MedPage Today, a publication for doctors."If an average adult keels over in the street, is found unresponsive and pulseless by a bystander, and is administered CPR while a 911 call is made, the odds that such a person will emerge from the eventualities of the resuscitation effort healthy and with a normally functioning brain are about 2 percent," he wrote last year in a MedPage Today editorial. The other 98 percent, he says, die on the scene, die within 30 days after expensive treatments and much suffering, or get discharged from a hospital alive but mentally impaired.CPR is useful only when the heart has stopped beating because of a temporary problem that can be reversed or treated; it can't save someone whose heart has stopped due to trauma, such as a car accident or a fall, Newman says. "Cardiac arrest from traumatic injuries is not a survivable condition. Even when we say it 'works,' what that means is that the heart is back and instead of dying then, they die later in ICU with many tubes and wires."Moral expectationYet automatically giving CPR to any person found unresponsive has become a societal norm, in part perhaps because of unrealistic depictions presented on television, Lundberg says. A 1996 New England Journal of Medicine study found that television CPR had a success rate of 75 percent -- far greater than even the most ideal real-life situations."We've gotten confused about where CPR is good -- almost a miracle-maker -- and where CPR is a use of resources that would be unwanted by the patient and unduly invasive and burdensome to both provider and patient," Newman says. "It should be done when there's a substantial chance of full recovery and the patient can safely be presumed to have wanted it." CPR was never meant to revive hearts that stop when a long-term chronic condition comes to a culmination.Which brings us back to the retirement-home case. If you find an elderly person unresponsive, do you give CPR? If you're certain that the person has a do-not-resuscitate order, Lundberg says, then the answer is easy: no. He notes that most of the physicians who responded to his MedPage Today editorial said that they would not choose to receive CPR.But what if you don't know the person's wishes? "When in doubt, you try. Barring definitive evidence that would make CPR beyond reason, ethically you ought to intervene," says Robert Veatch, professor of medical ethics at the Georgetown University Kennedy Institute of Ethics. A practicing physician or nurse trained in CPR has an ethical duty to respond, Veatch says, but for the layperson, it's ethically noble but not morally obligatory to give CPR.Legally, you are generally protected whether you give CPR or not. "Lay responders don't have a duty to act -- that's a legal term," says Pellegrino, but if you do administer CPR, Good Samaritan laws protect lay responders in most states.