CHICAGO — Researchers have devised and validated a scoring system to predict a patient's risk of dying while undergoing percutaneous coronary intervention.
“We have developed a user-friendly model, without need for angiography, to use in the decision making process,” Dr. Eric D. Peterson said at the annual meeting of the American College of Cardiology.
One anticipated use is to calculate a patient's risk-prediction score as part of the informed consent process. Having a score that is reliably accurate without the need for angiographic data is vital because once catheterization and angiography is underway, it's often “hard to stop the train” that ends up as a percutaneous coronary intervention (PCI), said Dr. Peterson, a cardiologist and professor of medicine at Duke University, Durham, N.C.
Another potential use is to give individualized feedback to interventional cardiologists by comparing their expected procedural mortality rate, based on their patients' characteristics, with their actual mortality rate.
The project was sponsored by the American College of Cardiology's National Cardiovascular Data Registry (NCDR), and it used data collected throughout the United States by the registry.
The scoring system was devised based on 60% of the 302,958 cases in the registry from January 2004 to March 2006, collected from 470 U.S. PCI sites that were voluntary NCDR participants. The system underwent an initial validation using the remaining 40% of cases from this period, and then had a second validation test with data from 285,440 cases done at 608 sites during April 2006-April 2007. For both score derivation and testing, cases were excluded if they were not the patients' first PCI, if the mortality data were questionable because the patients had transferred early, or if data on two or more of the tested clinical variables were missing.
Thirty-four candidate variables were initially considered, and this list was eventually narrowed to eight factors that made it into the scoring system (see table). The goal was a simple scoring formula that could easily be summarized on a card or programmed into an electronic device, Dr. Peterson said.
The result was a score that can range from 0 to 117. The periprocedural mortality rates presented by Dr. Peterson ranged from zero, for patients with a score of 0, to 98%, for patients with a score of 100 (see figure).
The two validations showed that the predicted scores were highly correlated with actual mortality, and that this held up regardless of patients' gender, age, risk level, whether or not they had diabetes, and whether or not they had an ST elevation MI. The only limitation to the scoring system is that it has only been validated with data collected through the NCDR, Dr. Peterson said.
ELSEVIER GLOBAL MEDICAL NEWS
ELSEVIER GLOBAL MEDICAL NEWS