Performance of the TIMI and GRACE scores in identifying the risk of death from acute coronary ischemic syndrome: a retrospective study in patients with chest pain at a Peruvian Hospital

Introduction

Cardiovascular diseases (CVD) are a group of disorders that affect the heart and blood vessels, classified as coronary disease, heart failure, congenital heart disease, and cardiomyopathy.¹ Approximately 17.9 million people died from CVD in 2019, representing 32% of deaths.² The acute form of coronary disease is acute coronary syndrome (ACS), caused by decreased coronary artery perfusion due to stenosis, distal embolization, or occlusion of a coronary artery.³ The types of ACS include unstable angina, ST-elevation MI (STEMI), and Non-ST-Elevation MI (NSTEMI).⁴ The incidence of ACS in emergency rooms represents between 5 and 20% of chest pain, frequently present in patients older than 60 (ratio of men and women of 3:2).⁵ ACS mortality is between 10 and 80 deaths per 100,000 people yearly; global mortality is estimated to be 800,000 yearly.⁶

Patients with ACS present angina pectoris that radiates to the left upper limb (arm) and jaw, dyspnea, diaphoresis, nausea, and vomiting. Due to the diversity of the symptoms presented by patients with ACS, it is necessary to carry out a detailed anamnesis and request additional tests, such as an electrocardiogram, biomarkers (serum troponin I and T), and chest X-Ray.⁷ Likewise, the initial assessment of the patient’s risk is essential. For this reason, different predictive models have been developed to assess the risk of ACS or associated events. Among the best-known models suggested in clinical practice guidelines are Thrombolysis in Myocardial Infarction (TIMI) and the Global Registry of Acute Coronary Events (GRACE), which are helpful in initial patient care.⁸ The TIMI risk score predicts mortality in 30 days and one year after an MI with ST-segment elevation; the AUC value is 0.7. A 7-point TIMI score should be considered a high risk of mortality during the patient’s hospitalization time after suffering a STEMI since it has greater sensitivity and specificity.⁹ The GRACE score provides good discrimination for death from type 1 myocardial infarction. The AUC incorporating death or future myocardial infarction for GRACE was 0.76 and 0.81, respectively. Likewise, for type 2 myocardial infarction, the GRACE score provided moderate discrimination in predicting death or future infarction, with an AUC of 0.70 and 0.72, respectively (p = 0.007 and p = 0.042 compared to type 1 myocardial infarction).²

According to the Pan American Health Organization, in 2019, there were 2 million deaths due to cardiovascular diseases. The mortality rate varies between countries. For example, Haiti registered 428.7 deaths per 100,000 people, while Peru had 73.5 deaths per 100,000 inhabitants.¹⁰ The Ministry of Health of Peru reports that CVDs are the second-place cause of mortality between 2014 and 2016, representing 17.8%. Similarly, ischemic heart disease represents 6.2% of deaths.¹¹ In the Tumbes Region, CVDs rank first, with 11.4% of all non-communicable diseases. Likewise, within the subcategories of non-communicable diseases, ischemic heart disease ranks fifth among this region’s top 10 causes of morbidity and mortality.¹² According to what has been evidenced, the GRACE and TIMI predictive models are little used in clinical practice, and they are not considered part of epidemiological surveillance programs in populations at risk of cardiovascular diseases, mainly ACS. This situation generates a limitation in the response capacity for the timely approach of patients at risk of ACS, as well as shortcomings in strengthening strategies for primary prevention. This study aimed to determine the performance of the TIMI and GRACE predictive models in identifying the risk of death from ACS in patients with chest pain at a Hospital in Tumbes. We believe that identifying the risk of death is crucial in people at risk of CVD, and using the predictive model with the best performance would make it possible to strengthen prevention and therapeutic actions by the medical staff.

Methods

Results

A total of 116 medical records were found in the study period, of which four were excluded due to a lack of EKG reports and six without troponin results (Figure 1). Hence, 106 ACS cases were analyzed. The average age of the patients analyzed was 65 years (interquartile range [IQR]: 55-73), with 63.2% being male. The median hospitalization time since admission for emergencies was two days (IQR: 1-4). Table 1 shows that chest pain (97.2%), pain in the left arm (73.6%), and dyspnea (57.5%) were the three most frequent clinical manifestations among people with ACS. [se podría mencionar la incidencia de Muerte y las categorías de los scores de TIMI y GRACE.]

Figure 1. Flow chart for participant selection.

Table 2 shows the frequencies of each predictor used in assessing the risk of death by the TIMI scale. The median and interquartile range scores were 2,^1–4 with a range from 0 to 6. The most frequent predictor was having at least three cardiovascular risk factors (97.2%). Among these, we obtained frequencies for hypertension (56.5%), hypercholesterolemia (58.5%), DM2 (64.2%), family history of CAD (20.8%), and current smoking (22.6%). Troponin elevation (79.2%) was the second most frequent predictor. On the other hand, we found patients without the presence of any predictor (2.8%), as well as with one (19.8%), two (30.2%), three (46.2%), and four predictors (0.94%).

Table 3 shows the frequencies of each predictor used in assessing the risk of death by the GRACE scale. Unlike the TIMI scale, we observed median heart rate, systolic pressure, and blood creatinine of 97 bpm, 130 mmHg, and 1.2 mg/dL, respectively. Identifying rales at the base of the lung (13.2%) was the most frequent in the Killip categories.

The medians for the TIMI and GRACE scores were 8% in both cases, but their IQQ was 5-20 and 2-14, respectively, these differences being significant (p=0.028, Wilcoxon non-parametric test). Figure 2 shows the AUC values for the TIMI and GRACE scores with 95.1% (CI95: 90.9-99.2%) and 95.2% (85.7-100.0%), respectively. We did not find significant differences (p=0.982, Pearson’s Chi-square) between both scores. On the other hand, the mortality rate due to ACS was 8.5% (CI95: 4.0-15.5%).

Figure 2. Comparison of areas under the curve (AUC) of the TIMI and GRACE scores.

Discussion

This study aimed to evaluate the performance of the TIMI and GRACE scores in identifying the risk of death from acute coronary ischemic syndrome in patients with chest pain. No differences between AUC values were found concerning the risk of death in patients with ACS that arrives with chest pain at a public hospital in the north of Peru.

Previous studies in other contexts have compared the TIMI and GRACE scores as predictors of mortality in patients with NSTEMI. One study reported that the GRACE score had a better prediction than TIMI in patients with NSTEMI (GRACE AUC vs. TIMI AUC [hospital: 0.82 vs. 0.62; long-term mortality: 0.89 vs. 0.68; long-term cardiac mortality: 0.91 vs. 0.67], all with a p<0.05).¹⁸ Another study in China reported in-hospital mortality of 6.0% in patients with NSTEMI, and the GRACE AUC of 0.7930 (95% CI: 0.767-0.818) was better than the TIMI AUC (0.5588; 95% CI 0.532-0.586), p<0.001.¹⁹ These results are in line with a meta-analysis comparing different risk scores for ACS, showing that in validation studies, pooled c-statistics of 0.54 (95% CI=0.52-0.57) and 0.67 (95% CI=0.62-0.71) for short and long term were found for studies using TIMI score, while 0.83 (95% CI=0.79-9.87) and 0.80 (95% CI=0.74-0.89) were found for short and long term for studies using GRACE. Similar results were found for validation studies comparing TIMI and GRACE in STEMI patients.²⁰ Regarding the GRACE score, it has been reported that it has a good predictive value for in-hospital death and death at six months in patients with NSTEMI. A previous study in 300 patients with NSTEMI with a six-month follow-up found that the high-risk category presented 10.5% of in-hospital deaths and 11.8% of deaths in the first six months (p=0.001 and p=0.013).²¹ The GRACE score, gender, diabetes mellitus, family history, and smoking were associated with hospital mortality. In contrast, the GRACE score, obesity, and lipid profile abnormalities were associated with death at six months.²¹ While the GRACE score is reported as a better risk predictor in ACS patients compared to TIMI, the results of this study do not find differences when using any of these scores. The non-identification of differences between these two scores in patients with ACS in the study population could be due to the small sample size, the fact that the patients come from a single health facility, and the possible inadequate recording of patient information in the medical record.

The literature also compares the GRACE, TIMI, and other scores for mortality in patients with ACS. For example, a study comparing the GRACE, TIMI, and RISK-PCI scores reported that the RISK-PCI score showed excellent potential for predicting 30-day mortality (AUC=0.96; 95%CI: 1.339-3.548), being non-inferior compared to the GRACE and TIMI scores.²² Another study, in patients with IMASTNE and diabetes, compared the Revised CADILLAC (Revised Controlled Abciximab and Device Investigation to Lower Late Angioplasty Complications) score with the TIMI, GRACE, and PAMI (Primary Angioplasty in Myocardial Infarction) scores. In the long-term findings, only the CADILLAC score achieved acceptable discrimination of the prediction of death at six (0.7261), 12 (0.7319), and 24 (0.7256) months.²³ Likewise, it has been reported that in patients with NSTEMI, the GRACE AUC was 0.81 versus 0.62 for the TIMI score, p<0.001, and the combination of TIMI and GRACE improved the predictive value of in-hospital events (AUC=0.71, 95% CI: 0.65-0.77), with a better PPV (37.1%) compared to that of the TIMI (33.3%) and GRACE (36.8%) scores independently.²⁴ Another study indicates that a TIMI score >4 (OR=3.42; 95%CI: 1.35-8.66) and a GRACE score ≥150 (OR=8.43; 95%CI: 3.33-21.38) are independent predictors of mortality after six months of an episode of NSTEMI.²⁵ What has been described indicates that the TIMI and GRACE scores are helpful for the prognosis of death in patients with ACS. However, other scores would be more useful in specific groups of patients, so future research could consider these scores and evaluate their usefulness and ease of use in the prognosis of death in patients with ACS.

The study’s limitations included the absence of variables (incorrectly localized pain, incorrectly filled out medical records, non-readable handwriting in medical records, misinterpretation of EKG) that could confuse the assessment of the risk of death from ACS. Likewise, some data used to calculate the scores obtained by both predictive models were obtained based on self-report rather than by measurement based on objective indicators. In Peru, a previous study of 234 cases of ACS found that the most significant proportion had a GRACE score with high risk (59% in ACS and 66% in IMA), with 9.4% of deaths. This study found an association between sex and the GRACE score with a high mortality risk.²⁶ Another study, also in the Peruvian population, found that patients who presented a TIMI score (≥3) had a high risk of a combination of events (heart attack or angina) compared to those with a score ≤2 (RR= 4.56; 95%CI: 1.97-10.54).²⁷ Given that the use of clinical scores depends on the health context in which they are used, we believe that the strength of this study is that it provides initial evidence regarding the comparability of these predictive scores for mortality in patients with ACS. This health problem leads the mortality in Peru.

Conclusions

There would be no difference in using the TIMI and GRACE scores to assess the risk of death in patients with ACS who attend a hospital for chest pain. Given that both scores are used in clinical settings and are easy to apply, using any of these as part of medical care in the Peruvian health system could help identify patients with a possibly worse prognosis. In the future, using larger sample sizes and a longer follow-up time will allow for more solid results regarding using these scores to assess the risk of death in patients with ACS.