HEART vs TIMI vs GRACE score for chest pain

What HEART, TIMI and GRACE were built for

The HEART, TIMI and GRACE scores all estimate cardiac risk, but they were built for different patients and different outcomes. The TIMI risk score for unstable angina and non-ST elevation MI, published by Antman and colleagues in JAMA in 2000, came from patients in two randomized trials who already had that diagnosis, and predicted death, new or recurrent MI or severe recurrent ischaemia needing urgent revascularization within 14 days. The GRACE model, from Granger and colleagues in 2003, used 11,389 patients with acute coronary syndromes, with and without ST elevation, in the Global Registry of Acute Coronary Events, and predicted death in hospital. The HEART score, from Six, Backus and Kelder in 2008, was designed for every patient who comes to the emergency department with chest pain, and predicts major adverse cardiac events within 6 weeks. So TIMI and GRACE grade risk once the diagnosis is made, and HEART helps decide who needs further workup at all.

What each score counts

TIMI and HEART are simple sums, while GRACE weights continuous measurements. TIMI gives 1 point each for 7 findings: age 65 or over, three or more coronary risk factors, a known coronary stenosis of 50% or more, ST deviation on the presenting ECG, two or more anginal episodes in the past 24 hours, aspirin use in the past 7 days and raised cardiac markers, for 0 to 7. In its derivation cohort, the 14-day event rate rose from 4.7% at 0 or 1 to 40.9% at 6 or 7. GRACE combines 8 variables, namely age, Killip class, systolic blood pressure, ST deviation, cardiac arrest at presentation, serum creatinine, raised cardiac enzymes and heart rate, which carried 89.9% of the prognostic information in the registry. HEART scores history, ECG, age, risk factors and troponin at 0, 1 or 2 points each, for 0 to 10, and is the only one of the three that grades how suspicious the history is.

Which score did best in emergency chest pain?

In emergency department chest pain patients, the HEART score outperformed both TIMI and GRACE in the two large Dutch comparisons. Backus and colleagues scored all three in 2,388 patients at 10 hospitals (International Journal of Cardiology, 2013). The c-statistic for major adverse cardiac events within 6 weeks was 0.83 for HEART, 0.75 for TIMI and 0.70 for GRACE, and HEART was significantly better than both. A HEART score of 0 to 3 covered 36.4% of patients with a 1.7% event rate, a TIMI score of 0 or 1 covered 34.0% with 2.8%, and a GRACE score of 60 or less covered 14.0% with 2.9%. Poldervaart and colleagues repeated the comparison in 1,748 patients at nine hospitals (2017), with areas under the curve of 0.86 for HEART, 0.80 for TIMI and 0.73 for GRACE. At a sensitivity of at least 98%, HEART identified 381 low-risk patients with 0.8% of events missed, GRACE 231 with 2.2%, and TIMI none.

Both studies come from the Dutch group that developed the HEART score, and the 2013 sites read troponin against their own conventional assays, not high-sensitivity troponin. The high-risk ends differed too: only HEART and TIMI reached a group with an event risk above 40% in 2013, at HEART 7 to 10 (50.1%) and TIMI 6 or 7 (42.5%).

Where TIMI and GRACE still fit

TIMI and GRACE remain useful once an acute coronary syndrome is established, which is the setting they were built for. Antman and colleagues designed TIMI to be calculated at presentation without a computer and to identify patients whose response to treatment differed with risk. GRACE predicted death in hospital across the whole spectrum of acute coronary syndromes, including ST elevation MI, with c-statistics of 0.79 to 0.84 across its derivation and validation sets. Used on undifferentiated chest pain, both lose ground. Backus found that age accounted for about half of all GRACE points in emergency patients, and noted that neither score uses the history that clinicians rely on. The authors concluded that HEART is the best of the three for chest pain in the emergency department, and that TIMI and GRACE should be reserved for patients with an acute coronary syndrome in the coronary care unit.

Worked example: a patient the scores split on

One patient shows how HEART and TIMI can disagree. A 58-year-old woman has had two episodes of chest pain in the past 24 hours, with a moderately suspicious history. She has hypertension and smokes, takes aspirin, has no known coronary disease, a normal ECG and a first troponin below the normal limit. HEART gives 1 for history, 0 for the ECG, 1 for age over 45 and under 65, 1 for one or two risk factors and 0 for troponin: a score of 3, the low-risk group, where Backus found a 1.7% six-week event rate. TIMI gives 1 for two anginal episodes in 24 hours and 1 for aspirin in the past 7 days, and nothing else: a score of 2, in the TIMI 2 to 5 group where Backus recorded events in 23.4%. TIMI counts her episodes and her aspirin; HEART weighs how typical her pain sounds.

Score the same patient in the HEART score calculator to see the risk group and its 6-week event rate.