Risk Score Predicts Long-term Mortality After PCI

Download this article's Factoid (PDF & PPT for Gold Subscribers)

Long-term mortality after percutaneous coronary intervention (PCI) can be predicted using a simple score that employs 12 risk factors, according to a paper published online January 14, 2014, ahead of print in Circulation: Cardiovascular Interventions.

Researchers led by Edward L. Hannan, PhD, of State University of New York (Albany, NY), developed the model based on 11,897 patients who underwent PCI from October through December 2003 in New York State. Mortality was tracked using the National Death Index through the end of 2008. Death rates in the data set were 4.0% at 1 year, 9.8% at 3 years, and 16.1% at 5 years.

Preprocedural Factors Still Matter at 5 Years

Using a Cox proportional hazards model, the researchers identified 12 separate risk factors for mortality:

• Older age
• BMI < 25 or ≥ 40 kg/m²
• Multivessel disease
• Lower EF
• Unstable hemodynamic state or shock
• Cerebrovascular disease
• Peripheral vascular disease
• Congestive heart failure
• COPD
• Diabetes
• Renal failure
• History of CABG

Each of these factors was assigned a point value ranging from 0 to 11, with the highest value given to shock. For individual patients assessed using the risk score, point totals ranged from 0 to 39. Predicted mortality rates at 1, 3, and 5 years were calculated for the various point totals. For example, for a patient with a score of 0, the risk of death was 0.58% at 1 year and 2.98% at 5 years. Meanwhile a patient with a score of ≥ 19 had estimated risks higher than 60% and higher than 99%, respectively.

A validation sample used to test the discriminatory power of the model showed “good agreement between patients’ observed and predicted risks of death,” Dr. Hannan and colleagues write, reporting C statistics of 0.787, 0.785, and 0.773 at 1, 3, and 5 years, respectively.

Not a Stand-alone Score

“Although it is true that statistical models can now be programmed using computers to calculate the predicted risk of death, a simple clinical risk score remains an attractive risk stratification tool for both clinicians and patients,” the investigators note, adding that such risk scores are more likely than complex algorithms to be adopted by clinicians.

Dr. Hannan told TCTMD in an e-mail communication that the new score could be used to inform patient management but preferably in combination with other tools. “Ideally [it] would be examined in conjunction with a long-term CABG risk score and short-term CABG and PCI risk scores for purposes of informed consent and procedure referral for patients with extensive coronary artery disease [who are not appropriate for medical therapy alone],” he said, citing several papers where such scores appear.

Patient preferences and other factors not included in the risk scores should also be taken into account, he advised.

Clinicians May be Hard to Convince

But Kishore J. Harjai, MD, of Geisinger Wyoming Valley (Wilkes-Barre, PA), told TCTMD in an e-mail communication that PCI risk prediction scores do not influence his management strategy.

“Even in the highest risk cohort of this study (ie, those with a score ≥10), about 80% of patients were alive at 1 year and 50% were still alive at 5 years. So, I would not use this score to turn down a high-risk patient for PCI that is otherwise appropriate. We have to remember that patients at high risk for PCI are generally also at high risk of dying from medical management or CABG,” he stressed.

“I find the current PCI risk prediction scores difficult to use in a busy clinical practice,” Dr. Harjai continued, noting that many factors can help identify high-risk patients without being combined into a score. “Yet good risk prediction models, if made available as apps on hand-held devices, might be useful in some elective patients who have multiple treatment options. Some clinicians may use them in the informed consent process. Unfortunately, robust models that predict long-term outcomes are difficult to find.”

He does, however, use the Society of Thoracic Surgeons (STS) score for prediction of 30-day mortality after CABG.

“This extensively validated score is based on decades of experience with hundreds of thousands of patients; uses demographic, clinical and angiographic variables; and represents contemporary data,” Dr. Harjai noted. “The ideal PCI risk prediction score will have to meet these criteria, look at long-term outcomes, and be a well-funded, ongoing, professional society-sponsored effort.”

The reluctance of clinicians to use risk scores may stem from habit, he acknowledged. “As cardiologists, we deal with lots of emergent life-threatening situations, including STEMI and shock, where other options are limited. We accept these challenging cases without knowledge of coronary anatomy or even basic data like CBC, [basic metabolic panel], or chest X-ray, relying upon our intuition of risk rather than complex algorithmic models,” Dr. Harjai concluded. “For good or bad, this ‘take it as it comes’ approach also permeates our behavior in less emergent situations.”

Wu C, Camacho FT, King SB III, et al. Risk stratification for long-term mortality after percutaneous coronary intervention. Circ Cardiovasc Interv. 2014;Epub ahead of print.

Related Stories:

• Long-term Mortality Higher After Incomplete Revascularization
• Risk Score Predicts Stroke or Death in Carotid Stenting Candidates
• Depression After PCI Linked to Long-term Mortality