A randomized trial of an automated cardiopulmonary resuscitation machine approved by the Food and Drug Administration suggests that it produces worse outcomes than CPR performed manually.

In the trial, there was no difference between machine and manual methods in the survival of patients for the first four hours after a heart attack. But the rate of survival until hospital discharge was 9.9 percent in the manual CPR group, compared with 5.8 percent in the machine group.

The results with the machine were so poor that the study, originally planned for 12 to 18 months, was halted after 9 months. Al Hallstrom, a professor of biostatistics at the University of Washington, was the lead author of the report, which was published in the June 14 issue of The Journal of the American Medical Association.

The authors said they could find no good explanation for the results. Another study, published in the same issue of the journal, found better results for the machine, but that study was conducted through observations and was not a randomized trial.

Dr. Roger J. Lewis, a co-author of an editorial accompanying the papers, said the skill with which both manual and machine CPR were used may have been a factor in the different findings. "I think the take-home lesson here is that the details of resuscitation — the quality of manual CPR and the timing of deployment of any device — are much more important than whether the CPR is done manually or with a device," Dr. Lewis said. Still, he urged caution in using new medical technologies.

"It's an uneasy balance between forces promoting early adoption and the forces advocating for more gradual incorporation of new technology to make sure it actually works," he said.