ESC: Newer Risk Scores May Be Needed for Predicting Mortality in Valve Disease

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

The most widely used risk scores to aid decision making in patients with valvular heart disease are suboptimal for estimating operative mortality risk, according to a position paper by a working group of the European Society of Cardiology. In the paper, published online March 15, 2011, ahead of print in the European Heart Journal, the authors say newer scores with additional variables and conditions are needed.

The working group, led by Raphael Rosenhek, MD, of the Medical University of Vienna (Vienna, Austria), reviewed the 3 most common risk scores: EuroScore (additive and logistic), the Society of Thoracic Surgeons (STS) risk model, and the Ambler score.

According to Dr. Rosenhek and colleagues, all 3 scores have been tested in large populations of patients with valvular heart disease but primarily were defined and validated in patients undergoing CABG. Although their ability to predict perioperative morbidity and long-term outcome has been reported, most published data refer to perioperative mortality only.

Scores May Under or Overestimate Mortality Risk

When calibrating the scores in the same population of high-risk patients with aortic stenosis, the researchers report that the mortality risk in 1 large study was underestimated by STS and overestimated by the other 2 scores (table 1).

Table 1. Predicted and Observed Mortality Rates

“In summary, the currently available scores achieve acceptable discrimination but suboptimal calibration in estimating the operative mortality of contemporary heart valve surgery,” the working group writes.

They suggest that differences in predicted mortality between the scores may be explained by the following:

• Different risk factors selected
• Different weighting of risk factors
• Variable patient characteristics
• Differences in outcome data used for risk score development (partly related to caseload at enrolling centers)
• Changing patient characteristics (age, comorbidity, and repeat surgery) over time, resulting in a disproportionate number of high-risk patients
• Changing operative techniques and reduced mortality with time, resulting in discrepancies between estimated and observed outcome

Differences may not be as important for low-risk patients as they are for higher risk ones, they add.

“It is therefore important that high-risk patients are sufficiently represented in the derivation cohort,” the working group writes. “Inclusion of large numbers of low-risk patients results in distortion (or skewing) of the model, making it imprecise and inapplicable in high-risk patients.”

Dr. Rosenhek and colleagues also take issue with the variables incorporated into the scores, noting that additional factors such as frailty and cognitive and functional capacity should be included. At the same time, however, the group maintains that risk scores should be “simple, reliable, and reproducible” and “based on a limited number of variables.”

Areas for Future Research

Among the working groups’ other recommendations are:

• A multicenter, international approach including patients with a broad spectrum of operative risk
• Validation of scores in both high- and low-volume centers
• Frequent updating
• Focus on estimating 30-day mortality following surgery
• Focus on operative mortality (in-hospital or 30-day mortality)

In addition, they say scores should be developed based on standardized definitions of endpoints, such as the Valve Academic Research Consortium criteria, which provide a standardized definition for endpoints after transcatheter aortic valve implantation (TAVI). Future risk evaluation models should incorporate methods for predicting outcomes after TAVI and other percutaneous valve procedures given their increasing volume of use.

Inaccurate and Dismissive

But the current president of STS, Michael J. Mack, MD, of Baylor Health Care System (Dallas, TX), told TCTMD in a telephone interview that he was extremely disappointed by “multiple inaccuracies and misstatements of facts” in the position paper.

In particular, the suggestion that STS or the other scores were developed and validated primarily in CABG patients is false, he said.

“It is not true for the STS, and it is not true for EuroScore. There is a very specific aortic valve risk algorithm in the STS score, so dismissing these scores as inaccurate because they were designed for CABG surgery is totally not valid. The STS score was designed for all surgery and, as a matter of fact, was based on the outcomes of 68,000 aortic valve patients,” Dr. Mack asserted.

“I totally disagree with a whole host of statements they make in this paper,” he concluded. “I think they are being too dismissive of a large body of scientific evidence.”

Dr. Mack added that all of the variables mentioned in the position paper currently are included in the STS with the exception of ‘neurological impairment affecting daily activity,’ which has been added to the most recent version scheduled to be released July 1, 2011. In addition, a number of conditions for exclusion, which the working group says should be considered for future scores, also already have been added to the new STS score.

“Everything they are asking for or recommending is already in progress,” Dr. Mack said. “There is an online calculator for the STS that takes 2 minutes to do. There are 27 variables that have been proven to impact mortality and that are weighted appropriately based on this. You want it to be accurate, so you have to include every variable that has been demonstrated to impact mortality. Sixteen of the variables in the STS are related to valvular heart disease.”

Dr. Mack added that the working group’s recommendation for a multicenter, international approach is something STS would “love to do” but cannot, because there is no European registry.

The one assertion he did agree with, however, is that the risk scores were developed for conventional open-heart surgery and may not translate to transcatheter therapy.

“Already there are some signals that TAVI may be lower risk in women,” Dr. Mack said. “The new STS algorithm is capturing transcatheter valve procedures and will continue to do so going forward. The Europeans have done 30,000 valves in 5 years. Where are their data?”

Source:

Rosenhek R, Iung B, Tornos P, et al. ESC working group on valvular heart disease position paper: Assessing the risk of interventions in patients with valvular heart disease. Eur Heart J. 2011;Epub ahead of print.

Disclosures:

• The paper contains no disclosure information for Dr. Rosenhek.
• Dr. Mack reports being a member of the executive board of the PARTNER trial, which is sponsored by Edwards Lifesciences.

Related Stories:

• Consensus TAVI Endpoint Definitions Raise Standard of Research in New Field
• PARTNER EU Registry Shows Promising TAVI Results at 1 Year
• Most Anatomy Suitable for Contemporary TAVI Approaches