Risk scores prognostic implementation in patients with chest pain and nondiagnostic electrocardiograms

30 day predicted outcome in undifferentiated chest pain: multicenter validation of the HEART score in Tunisian population

BACKGROUND

Chest pain remains one of the most challenging serious complaints in the emergency department (ED). A prompt and accurate risk stratification tool for chest pain patients is paramount to help physcian effectively progrnosticate outcomes. HEART score is considered one of the best scores for chest pain risk stratification. However, most validation studies of HEART score were not performed in populations different from those included in the original one.

OBJECTIVE

To validate HEART score as a prognostication tool, among Tunisian ED patients with undifferentiated chest pain.

METHODS

Our prospective, multicenter study enrolled adult patients presenting with chest pain at chest pain units. Patients over 30 years of age with a primary complaint of chest pain were enrolled. HEART score was calculated for every patient. The primary outcome was major cardiovascular events (MACE) occurrence, including all-cause mortality, non-fatal myocardial infarction (MI), and coronary revascularisation over 30 days following the ED visit. The discriminative power of HEART score was evaluated by the area under the ROC curve. A calibration analysis of the HEART score in this population was performed using Hosmer-Lemeshow goodness of test.

RESULTS

We enrolled 3880 patients (age 56.3; 59.5% males). The application of HEART score showed that most patients were in intermediate risk category (55.3%). Within 30 days of ED visit, MACE were reported in 628 (16.2%) patients, with an incidence of 1.2% in the low risk group, 10.8% in the intermediate risk group and 62.4% in the high risk group. The area under receiver operating characteristic curve was 0.87 (95% CI 0.85-0.88). HEART score was not well calibrated (χ² statistic = 12.34; p = 0.03).

CONCLUSION

HEART score showed a good discrimination performance in predicting MACE occurrence at 30 days for Tunisian patients with undifferentiated acute chest pain. Heart score was not well calibrated in our population.

Does an imaging stress-test adds information to prognostic scores in patients with chest pain in the emergency department?

We evaluated the ability of a stress-test (Str-T) to improve the risk stratification based on prognostic scores in patients presenting to the ED with chest pain. Between 2008, June and 2013, December, 1082 patients with chest pain were evaluated with an imaging Str-T. With a retrospective analysis, patients were stratified according to: (1) Florence Prediction Rule as low (0-1, LR-FPR), intermediate (2-4, IR-FPR), high risk (5-6, HR-FPR), respectively, 26, 50 and 24% of patients; (2) HEART score as LR-HEART, (0-3) and HR-HEART (≥4), respectively, 36 and 64%; (3) likelihood of CAD according to NICE guidelines, 10-29% LR-NICE, 30-60% IR-NICE and > 60% HR-NICE, respectively, 12, 18 and 70%. Scores' diagnostic performance was calculated with Str-T as reference. One-month follow-up by a phone call was performed, to investigate the occurrence of new cardiovascular events. In LR and HR patients, FPR and NICE score showed sensitivity 66 vs 93%, specificity 59 vs 19% (both p < 0.001), Positive Predictive Value (PPV) 36 vs 31%, Negative Predictive Value (NPV) 83 vs 87%. Among LR-HEART patients, Str-T was positive for inducible ischemia in 53 (14%) patients and 12 (4%) of them underwent a percutaneous coronary revascularization. The Str-T was negative for inducible ischemia in 760 (70%) patients, positive in 272 (25%), inconclusive in 50 (5%); among patients in the LR and IR subgroups, incidence of CAD (1.3 and 1.6%) and the cumulative incidence of significant events at 1-month follow-up (both 1%) was very low Str-T improved prognostic scores' diagnostic performance in LR- and HR-subgroups.

Comparison of nine coronary risk scores in evaluating patients presenting to hospital with undifferentiated chest pain

INTRODUCTION

We compared performance of nine risk scores for coronary heart disease (CHD) among patients presenting to an emergency department (ED) with undifferentiated chest pain of possible coronary origin.

METHODS

A retrospective study was undertaken of adult patients presenting with chest pain to atertiary hospital ED with no electrocardiographs or troponin results diagnostic of ischemic chest pain (ICP) or acute coronary syndrome at ED presentation, and no clearly evident noncoronary diagnosis. Risk scores were applied using cut-points distinguishing low- from high-risk patients according to discharge diagnosis of noncardiac chest pain (NCCP) or ICP, respectively. A lower odds ratio (OR) for ICP denoted lower risk for ICP. Score performance was compared using area under receiver-operator characteristic curves (AUC) and predictive values.

RESULTS

A total of 401 patients were studied, of whom 123 (30.7%) had ICP as final diagnosis. Among the nine risk scores, those with greatest ability to detect low-risk patients were The North American Chest Pain Rule (NACPR) score (OR=0.35, 95% CI=0.27-0.46); History, ECG, Age, Risk Factors, and Troponin (HEART) score (OR=0.43; 95% CI=0.35-0.52); and Thrombolysis in Myocardial Infarction (TIMI) score (OR=0.49; 95% CI=0.41-0.58). Discrimination between patients with NCCP and those with ICP was greatest for HEART score (AUC=0.82; 95% CI=0.78-0.86) and lowest for Accelerated Diagnostic Protocol to Assess Patients with Chest Pain Symptoms Using Contemporary Troponins (ADAPT) score (AUC=0.63; 95% CI=0.58-0.69). In excluding ICP, ADAPT had negative predictive value (NPV) 100% (miss rate 0%) but classified only 1.7% of patients as low risk, compared to NACPR with NPV 98% (miss rate 2%), classifying 10.2% as low risk, and HEART with NPV 94% (miss rate 6%), classifying 32.4% as low risk.

CONCLUSION

The NACPR risk score maximized yield of low-risk patients with lowest miss rate for ICP, while HEART score classified highest proportion of low-risk patients but with a higher miss rate.

A new score for the diagnosis of acute coronary syndrome in acute chest pain with non-diagnostic ECG and normal troponin

BACKGROUND

Acute coronary syndrome (ACS) represents a difficult diagnostic challenge in patients with undifferentiated chest pain. There is a need for a valid clinical score to improve diagnostic accuracy.

OBJECTIVES

To compare the performance of a model combining the Thrombolysis in Myocardial Infarction (TIMI) score and a score describing chest pain (ACS diagnostic score: ACSD score) with that of both scores alone in the diagnosis of ACS in ED patients with chest pain associated with a non-diagnostic ECG and normal troponin.

METHODS

In this observational cohort study, we enrolled 809 patients admitted to a chest pain unit with normal ECG and normal troponin. They were prospectively evaluated in order to calculate TIMI score, chest pain characteristics score and ACSD score. Diagnosis of ACS was the primary outcome and defined on the basis of 2 cardiologists after reviewing the patient medical records and follow-up data. Mortality and major cardiovascular events were followed for 1 month for patients discharged directly from ED. Discriminative power of scores was evaluated by the area under the ROC curve.

RESULTS

ACS was confirmed in 90 patients (11.1%). The area under the ROC curve for ACSD score was 0.85 (95% CI 0.80 to 0.90) compared with 0.74 (95% CI 0.67 to 0.81) for TIMI and 0.79 (95% CI 0.74 to 0.84) for chest pain characteristics score. A threshold value of 9 appeared to optimise sensitivity (92%) and negative predictive value (99%) without excessively compromising specificity (62%) and positive predictive value (23%).

CONCLUSIONS

The ACSD score showed a good discrimination performance and an excellent negative predictive value which allows safely ruling out ACS in ED patients with undifferentiated chest pain. Our findings should be validated in a larger multicentre study.

External validation of new risk prediction models is infrequent and reveals worse prognostic discrimination

OBJECTIVES

To evaluate how often newly developed risk prediction models undergo external validation and how well they perform in such validations.

STUDY DESIGN AND SETTING

We reviewed derivation studies of newly proposed risk models and their subsequent external validations. Study characteristics, outcome(s), and models' discriminatory performance [area under the curve, (AUC)] in derivation and validation studies were extracted. We estimated the probability of having a validation, change in discriminatory performance with more stringent external validation by overlapping or different authors compared to the derivation estimates.

RESULTS

We evaluated 127 new prediction models. Of those, for 32 models (25%), at least an external validation study was identified; in 22 models (17%), the validation had been done by entirely different authors. The probability of having an external validation by different authors within 5 years was 16%. AUC estimates significantly decreased during external validation vs. the derivation study [median AUC change: -0.05 (P < 0.001) overall; -0.04 (P = 0.009) for validation by overlapping authors; -0.05 (P < 0.001) for validation by different authors]. On external validation, AUC decreased by at least 0.03 in 19 models and never increased by at least 0.03 (P < 0.001).

CONCLUSION

External independent validation of predictive models in different studies is uncommon. Predictive performance may worsen substantially on external validation.