GRACE Performs Poorly in Ethnic Minorities With NSTEMI, Study Finds

The GRACE score’s ability to predict the in-hospital mortality risk of patients with NSTEMI is suboptimal in Black and Asian populations compared with white patients, a new study shows.

“This is a score advocated by the European Society of Cardiology as a class I indication for deciding when patients should receive invasive therapy and a class IIa indication for prognosis,” said senior study author Mamas Mamas, BMBCh, DPhil (Keele University/Royal Stoke University Hospital, Stoke-on-Trent, England). “It is a score that has been validated in many populations. But the problem is that it was derived and validated in predominantly Caucasian populations without consideration as to whether it performs equally well in different ethnic groups of patients that make up our society.”

Few risk scores have looked at performance by race and ethnicity, Mamas noted. Coronary artery calcium scoring has been validated in a broad spectrum of races and ethnicities, and concentrations of lipoprotein(a) have been shown to vary markedly between ethnicities, but teasing out true differences in outcomes remains complicated and imperfect. 

In their study of patients from England and Wales, Mamas and colleagues, led by Saadiq M. Moledina, MBBChir (Keele University/Royal Stoke University Hospital), found that compared with white patients, ethnic minority patients with NSTEMI were younger and had more comorbidities, but were more likely to have overestimated risk of mortality when the GRACE score was used.

Mamas told TCTMD that while the risk overestimation is actually contrary to what he initially suspected the study would show, the implications are far-reaching.

“The wider impact that this work highlights is that for any risk score, we should be assessing the performance of different racial groups,” he added. “We could have done it for any one of the scores, and I suspect that the findings would have been very similar. These scores are derived primarily from white populations. Performance will probably not be as good in an ethnically diverse population, meaning that we are making decisions based on incorrect assessments of risk, therefore placing these patients at a disadvantage in the healthcare that we provide.”

We could have done it for any one of the scores, and I suspect that the findings would have been very similar. Mamas Mamas

Commenting on the study for TCTMD, Khadijah Breathett, MD (Indiana University School of Medicine, Indianapolis), stressed that while race and ethnicity should not, in and of themselves, be a part of these risk-stratifying tools, “our calculators have to move forward in addressing major factors that contribute to racial differences in care and outcomes.” Specifically, that could include things such as whether the patient has received care from a cardiologist, is being appropriately treated for comorbid conditions, and is being seen regularly by a primary care physician. It might also include aspects of social determinants of health such as access to fresh food and open spaces for exercise.

“What this study demonstrates is that we have to do better with these calculators,” she said. “They’re only working for certain populations, and just adding race or ethnicity to the calculator is absolutely not the answer. [We need to] add those different factors that contribute to the differences in care and outcomes for racial ethnic groups.”

Good Discrimination, but Poorly Calibrated in Nonwhite Patients

For the study, published online last week in the European Heart Journal, the researchers calculated the GRACE score in 299,184 white and 26,976 ethnic minority patients from the Myocardial Infarction National Audit Project (MINAP), a registry of patients with a NSTEMI diagnosis who were admitted to hospitals in England, Wales, and Northern Ireland between 2010 and 2017. The ethnic minority population included those identifying as Black, Asian, or other nonwhite ethnicity. Ethnic minority patients were an average of 4 to 5 years younger than white patients and had significantly higher rates of diabetes, hypercholesterolemia, previous MI, and hypertension.

Patients were categorized by GRACE score as high risk (> 140), intermediate risk (109-140), and low risk (< 109). While there was no difference between white and ethnic minority patients in the low- and intermediate-risk categories in angiography rates, within the higher-risk category, angiography use was more common among ethnic minorities than whites (76% vs 63%; P < 0.001). The situation was the same for PCI, with similar rates at low and intermediate risk, but greater likelihood of PCI in high-risk ethnic minorities than high-risk whites (49% vs 42%; P < 0.001). Regardless of risk category, more ethnic minorities underwent CABG than did white patients.

The rates of in-hospital mortality were 5.2% in white patients and 2.9% in ethnic minority patients. The GRACE score showed good discriminating ability in the group as a whole, with an area under the curve (AUC) of 0.87 (95% CI 0.86-0.87), and individually in the white (AUC 0.87; 95% CI 0.86-0.87) and minority groups (AUC 0.87; 95% CI 0.86-0.88). The GRACE score was well-calibrated in the white group, but not in ethnic minority patients either overall or by individual ethnicity, overestimating risk in all nonwhite populations.

In subgroup analyses, the poorest calibration was in Black patients, with a slope of 0.84, followed by other nonwhite ethnic minorities, with a slope of 0.96. Looking further at ethnic minorities by sex, overestimated was more apparent in women than men, with slopes of 0.86 versus 0.98.

Asking the Right Questions

To TCTMD, Mamas said ethnic minorities typically present with NSTEMI about 5 years earlier than do white patients, while white patients often have comorbid conditions including frailty that are not accurately captured by the GRACE score. Both of these factors could explain some of the findings.

“It also could be related to other socioeconomic aspects,” he added. “It may be that some minority ethnic patients that have the highest risk perhaps don't come to hospital, because we know that there are ethnic and racial differences in health-seeking behaviors. There could be any number of reasons that explain what we found.”

There is a dark history of the treatment of patients of color, and we, as healthcare professionals, have to recognize that this is a systemic issue that we must be willing to address. Khadijah Breathett

Importantly, Mamas noted, despite increased emphasis on the underperformance of risk scores in stratifying women who present with an acute CV event, it’s time to ask why the same thing isn’t happening for race.

Breathett agreed. “There is a dark history of the treatment of patients of color, and we, as healthcare professionals, have to recognize that this is a systemic issue that we must be willing to address,” she said. “I think we have to go through each [risk calculator] one by one and figure out what can we do to change it, to make it better and to make sure that all of the appropriate stakeholders are at the table to make these decisions. That includes patients, community leaders, executives for the hospital, clinicians, nurses, front-desk individuals.

“These calculators are not serving our patients accurately. We have to get more diverse populations in the initial data sets and we have to make sure that we're asking the right questions,” she stressed.