Routine Use of GRACE Score Doesn’t Improve ACS Management: AGRIS

In the setting of high-quality clinical care, routinely risk-stratifying with the GRACE score does not have a major impact on the management or outcomes of patients with ACS compared with standard care, results of the cluster-randomized AGRIS trial show.

Patients in the intervention arm were no more likely to receive all three performance measures—early invasive treatment, discharge prescription of guideline-recommended medications, and referral for cardiac rehabilitation—than were those in the control arm, either in high-risk patients or the overall trial population, researchers led by Derek Chew, MBBS, PhD (Flinders University, Bedford Park, Australia), report.

Among the individual measures, there was no impact in terms of medication use or cardiac rehab referral, although use of the GRACE score was associated with an increase in early invasive treatment in the overall trial cohort.

In terms of clinical outcomes, there were no significant differences between trial arms in terms of the 1-year rate of all-cause death or MI.

To TCTMD, Chew pointed to three explanations for the lack a major effect of implementing routine use of the GRACE score: improved quality of care overall from the time the study was conceived to when it was completed; the participation of relatively high-performing hospitals in Australia; and a masking of potential benefit by the inclusion of three measures in the primary endpoint rather than studying use of invasive treatment alone.

The study showed that the GRACE score has good performance for stratifying risk, consistent with prior studies, but how that tool is used in practice needs to be more nuanced, Chew said. It’s not clear, he added, what to do differently for a patient who is deemed high or low risk. “If we’re going to rely on objective measures of risk, we need better structured guidance about what to do in response to that. And then whether or not that improves outcome needs further prospective study.”

Jacqueline Tamis-Holland, MD (Icahn School of Medicine at Mount Sinai, New York, NY), believes there is likely still a role for use of tools like the GRACE score in practice.

“While the AGRIS study did not demonstrate a benefit from the routine use of formal risk score assessment for treating patients with ACS as defined by the primary endpoint, the use of risk scores remains important for prognostication,” she commented to TCTMD. “Additionally, this study did not specifically look at the use of risk scores to help guide the decision for noninvasive versus invasive evaluation in low-risk patients with ACS, and there may be a role for risk scores to help guide assessment and management of the low-risk patient with ACS who may benefit from noninvasive testing.”

The AGRIS results were published recently online in JAMA Cardiology. Some preliminary data were previously reported at the European Society of Cardiology Congress 2019.

Australian GRACE Risk Score Intervention Study

Prior research, Chew pointed out, has shown that clinicians’ assessment of patient risk doesn’t always line up well with assessments by objective risk tools like the GRACE score. Use of the GRACE score is recommended in ACS management according to international guidelines, but that recommendation is without supporting evidence. “There was no prospective study,” said Chew. “It just seems like a good idea.”

The investigators sought to address that evidence gap with the Australian GRACE Risk Score Intervention Study (AGRIS), a cluster-randomized trial nested within CONCORDANCE, a multicenter ACS registry. The trial was stopped for futility after 24 Australian hospitals had been randomized to routine implementation of the GRACE score or standard care. Of the 2,318 patients included (median age 65; 29.5% women), 62.9% were considered high risk.

The primary endpoint consisted of receipt of all three of the following measures:

• Early invasive treatment during the index hospitalization
• Discharge prescription of at least four of five guideline-recommended drugs (aspirin, statins, beta-blockers, P2Y12 inhibitors, and ACE inhibitors/ARBs)
• Referral for cardiac rehabilitation

In the high-risk patients, routine use of the GRACE score did not result in a higher proportion of patients getting all three measures compared with standard care (59.9% vs 55.2%; OR 1.04; 95% CI 0.63-1.71). The findings were similar in the overall trial population.

Patients in the GRACE arm were more likely to receive early invasive treatment (91.8% vs 83.6%; OR 2.26; 95% CI 1.30-3.96), but not guideline-recommended drugs (76.7% vs 77.5%; OR 0.97; 95% CI 0.68-1.38) or referral for cardiac rehab (75.1% vs 72.8%; OR 0.68; 95% CI 0.32-1.44).

At 1 year, the rate of all-cause death or MI was numerically, but not significantly, lower in the GRACE arm (9.2% vs 13.4%; OR 0.66; 95% CI 0.38-1.14).

Tamis-Holland said that the study was underpowered to show an impact on clinical outcomes, pointing to the low event rates and the high number of patients who received early invasive treatment in both arms of the trial. In addition, she said, every patient with ACS will go home on guideline-directed therapy, regardless of risk, rendering it difficult to show an effect of routine risk stratification in this setting.

“I don’t think [calculating a risk score is] going to help recruit the number of patients you’re more aggressive with, because we’re pretty aggressive with our patients nowadays,” Tamis-Holland said.

Can Technology Help?

In an accompanying editorial, Robert Harrington, MD (Stanford University, CA), and E. Magnus Ohman, MBBS (Duke University, Durham, NC), say a closer look at the data suggests risk stratification may still be useful. They note that mortality was numerically lower in the GRACE group in the overall cohort (5.2% vs 8.0%; P = 0.12) and significantly lower in the high-risk patients (4.5% vs 8.9%; P = 0.03).

“The limited sample size and early termination prevent us from making a conclusive statement on mortality, but the findings are intriguing,” they write.

Because of the limitations of the trial, which also include the mixing of patients with and without ST-segment elevation, the question of whether risk stratification using a tool like the GRACE score is useful remains unanswered, Harrington and Ohman state. Newer technology might help in the task of risk assessment and delivery of effective treatments.

“Chew and colleagues have made the first attempt to clarify the science of risk stratification and implementation,” write the editorialists. “We look forward to future investigations in which technology can assist us in performing much larger trials in risk stratification.”

Chew said his group is planning a new 9,600-patient trial that will incorporate diagnostic and risk algorithms using artificial intelligence to see if enhanced technology support will improve the utility of risk assessment.

Future trials, too, need to provide specific guidance to clinicians about what to do in response to various levels of estimated risk. “We need to have those decision points in the risk score and then from those decision points we need to prospectively validate whether that . . . actually leads to better care and better outcomes, either in reducing complications or in reducing future events,” Chew said.

The takeaway from the paper is that “it’s not good enough to publish a risk score,” Chew added. “You need to demonstrate that it truly impacts care and outcome in prospective studies.”