Physicians were provided with a computerized decision support system that alerted them to potential adverse drug events. In one arm of the trial, the number and type of alerts could be customized by the physician. However, neither method resulted in a reduction in prescribing problems, partly because physicians ignored most alerts.