Risk stratification in chest pain patients

Abstract

Yearly, a total of approximately 200,000 patients in The Netherlands present to the emergency department with chest pain. Only 20% of these patients will have an acute coronary syndrome (ACS), and need prompt admission and treatment. However, differentiating between ACS and other, mostly non-life-threatening disease, remains difficult since symptoms and results of laboratory tests and ECG can be aspecific. As a result, most chest pain patients are hospitalized for observation and additional testing. This is not only a time-consuming and costly strategy, but also puts many low risk patients for ACS at risk of complications of these diagnostic procedures. This emphasizes the importance of research to improve the risk stratification in chest pain patients, in particular the identification of patients at low risk for ACS.

In this thesis we have evaluated and compared several strategies for risk stratification in acute chest pain patients. In the first three chapters, we describe the rationale, design and findings of the HEART-Impact trial. The HEART score is a validated clinical prediction rule, especially developed to stratify risk in chest pain patients. In this trial with a stepped wedge design we compared “usual care” with “HEART care” in terms of safety, use of medical resources and costs. The HEART care included a calculation of the HEART score in every individual patient and a recommendation for further management. Particularly, this meant early discharge in low-risk patients (HEART score of 3 or below). Our findings were that active use of the HEART score during the work-up of chest pain patients at the emergency department is just as safe as usual care, since non-inferiority was demonstrated: six-week incidence of MACE during HEART care was 1.3% lower than during usual care (upper limit 95% CI: +2.0%, not exceeding the non-inferiority margin of 3%). The decrease in use of health care resources following the initial assessment was small, although cost-effectiveness was formally demonstrated, with the probability of 99% that HEART is cost-effective for a Willingness-to-Pay threshold of €20,000/QALY. Furthermore, we compared the performance of the HEART score with the well-known GRACE score and TIMI score. Our findings were that the HEART score is the most efficient score, since it identified the largest number of patients as low risk, without compromising safety. Also, we show that on top of risk stratification by the HEART score, bicycle stress testing has only a modest contribution to clinical decision-making. Lastly, when assessing patients with chest pain suspected of ACS, none of the investigated markers (copeptin, myoglobin, PlGF, s-Flt-1, NT-proBNP, GDF-15) yielded any additional diagnostic value when combined with high-sensitive cardiac troponin T.

In conclusion, based on the results presented in this thesis, we would advise the use of the HEART score in the work-up of chest pain patients in more hospitals. However, the limited impact on health care resources also underline the importance to identify possible barriers inhibiting the acceptance of the recommended management in the low-risk HEART patients. The question whether patients and society truly benefit from the use of the HEART score will hopefully be answered in the coming years, with increasing body of evidence on the HEART score and increasing acceptance and adherence to the score in clinical practice.