Since 2008, the American Heart Assn. has recommended a "hands only" approach to CPR, emphasizing the importance of performing rapid chest compressions on victims of sudden cardiac arrest. The group decided to nix the mouth-to-mouth portion of cardiopulmonary resuscitation in part because studies show that it doesn't improve overall survival, and in part to increase the odds that a bystander would perform any kind of CPR at all.

But a new study finds that the old-fashioned version of CPR is more effective at resuscitating children in cardiac arrest.

Japanese researchers examined the medical records of 5,170 minors (ages 17 and younger) who were treated by emergency medical personnel for an out-of-hospital cardiac arrest in the years 2005, 2006 and 2007. Unfortunately, only 9 percent of those children survived, and even fewer - 3 percent - had a "favorable neurological outcome." But the ones who got CPR from a bystander stood a much better chance of preserving their neurological function than those who didn't - 4.5 percent vs. 1.9 percent, according to a report being published online by the journal Lancet.

The researchers also found that conventional CPR was more likely to result in a "favorable neurological outcome" than compression-only CPR. In their analysis, 7.2 percent of children given chest compressions and mouth-to-mouth had a good outcome, compared with 1.6 percent of kids who got compressions only.

In a commentary accompanying the study, Spanish researchers say the reason is probably that most cases of sudden cardiac arrest in children - 71 percent in the Japanese study and more than 90 percent in other studies _ are probably caused by non-cardiac events. (Only about a third of cases in adults are thought to have non-cardiac origins.) In such cases, mouth-to-mouth resuscitation is helpful. When cardiac arrest has a cardiac cause, either type of CPR works equally well.

They conclude that bystanders should continue to provide traditional CPR to children in cardiac arrest.