Risk assessment after acute coronary syndrome

Abstract

Lots of potential but will it end up being yet another risk score?

A range of presentations of ischaemia is seen in acute coronary syndromes, from unstable angina at one end of the risk spectrum to myocardial infarction (with or without ST elevation) at the other. In all these disorders the risk of death is highest before admission to hospital, with mortality rates of up to 20%. Risk remains high after admission to hospital, and although mortality rates have fallen greatly in recent years,¹ up to 7% of patients die before discharge, and risk continues to be high for six months after the ischaemic event.² Minimising the risk of complications relies on identifying and treating patients at higher risk early on. In this week's BMJ, a multinational study by Fox and colleagues assesses the effectiveness of a risk prediction tool in estimating cumulative six month risk of death or myocardial infarction in people presenting with acute coronary syndrome.³

On presentation to hospital, patients are often triaged on the basis of concentrations of biomarkers and ST segment changes into three groups—those who need immediate reperfusion therapy (ST elevation myocardial infarction), early angiography with a view to revascularisation (non-ST elevation myocardial infarction), or medical treatment only without routine catheter laboratory intervention (troponin negative unstable angina). This triage system also provides a crude basis for risk stratification as patients with myocardial infarction have a greater risk of death than those with unstable angina.⁴ However, risk stratification based simply on biomarkers and changes in electrocardiography fails to reflect the diversity of risk within these diagnostic subgroups. Many attempts have been made to refine the process of risk assessment further, but none of these newer tools has been adopted in clinical practice.

Will the global registry of acute coronary events (GRACE) scoring system described in this week's BMJ prove more useful in clinical practice?³ Fox and colleagues reassuringly confirm what doctors already recognise—high risk patients are elderly, have left ventricular failure, and have documented ST changes. These variables are components of other scoring systems, however, and the value added by each of the additional clinical factors incorporated into the GRACE score is not known. In other words, how many patients who would have been identified as low risk could theoretically benefit from treatment and how many patients with an inflated risk could be spared potentially harmful treatment if the GRACE scoring system were implemented? The stated goal of the new score is to provide a simple but robust basis for guiding care in hospital and after discharge for patients with acute coronary syndrome but stops short of suggesting how. Doctors are left to ponder at what point in the scoring system the large number of patients with acute coronary syndrome at intermediate risk justify a different management approach than is used for patients at the extremes.

Another potential weakness relates to the cohort of patients recruited. The score was developed and validated in selected cohorts of patients from voluntarily participating hospitals, the response rates are unclear, and patient recruitment was not explicitly consecutive in all cases—so was not strictly random. This may explain why hospital mortality rates seem lower than would be expected from comparable UK data.⁵ Despite this, the score appears more representative of patients across the spectrum of acute coronary syndromes than those developed previously.^{6 7}

Other factors independent of the risk score determine which treatment patients will receive. Interventions such as secondary prevention drugs and lifestyle advice are (or should be) given to all patients before discharge from hospital regardless of risk—a management policy that will be unaffected by risk score. Timely revascularisation may reduce risk in selected patients with acute coronary syndromes, yet patients at highest risk such as elderly patients⁸ and those with left ventricular failure are least likely to be referred. Indeed, such referral patterns in combination with angiographic findings often drive decisions on revascularisation in high risk patients,⁹ and provide a recipe for inequity that risk scores have the potential to remedy. If this treatment bias can be corrected by using the risk calculator of Fox and colleagues this might prove to be its greatest asset. If patients were treated according to their risk of events rather than referral patterns, older patients, for example (who benefit from coronary artery bypass grafting but are less likely to be referred for it) might get treated.

Fox and colleagues conclude that risk prediction tools may potentially guide therapeutic interventions in the future. This highlights how useful risk scores could be in clinical practice; for example, in predicting long term outcomes and alternative endpoints to death, which occurs in a minority of patients after an acute coronary event. The authors stress that their score may not be appropriate for low risk patients with non-specific chest pain without acute coronary syndrome. This emphasises the need of accurate risk stratification upstream from tertiary care; that is, primary and secondary care and interfaces such as rapid access chest pain clinics, where a surprising number of such patients develop cardiac endpoints.¹⁰ This forces the question of whether the Fox and colleagues' risk score really is so different from that proposed for patients with myocardial infarction in the late 1960s,¹¹ or whether we should take a closer look at why risk assessment in general is so difficult to adopt in clinical practice.