The Authors' reply

To the Editor The article by Gray et al is a succinct and useful summary of the National Institute for Health and Clinical Excellence (NICE) clinical guidance 94 (CG94) on the management of unstable angina and non-ST elevation myocardial infarction. The NICE document is itself a major piece of work that synthesises vast bodies of evidence. The majority of the document and the resulting summary to which we respond are to be applauded. However, we believe the risk stratification method for access to routine invasive management is flawed. NICE CG94 recommends management in line with the estimated 6-month mortality according to the GRACE (Global Registry of Acute Coronary Events) registry outcomes prediction model, with some caveats in relation to bleeding risk. This model has very important limitations that invalidate its use for the stated purpose. First, all risk prediction models become increasingly inaccurate as the size of the population to which they are applied reduces. Consequently, there is too much emphasis on the percentage score and too little emphasis on clinical assessment of individual patients. Second, the model ignores the main benefits of a routine invasive strategy over a selective invasive strategy, namely reductions in recurrent nonfatal myocardial infarction, admission rates, revascularisations and costs during followup. Third, the GRACE score does not take into account the underlying age-based actuarial survival rates. It therefore overestimates the risk of death in relation to the index acute coronary syndrome episode in older patients. The NICE-recommended approach places patients aged below 50 years with electrocardiographic ST depression and troponin elevation into a selective invasive group, while requiring routine invasive management for virtually all patients aged above 70 years presenting without these features; this is clearly unworkable and wrong. The cardiology community should not accept the oversimplified use of risk scores in this way. The algorithms recommended by NICE should be adapted to reflect the individual clinical picture and to take into account the risk of recurrent myocardial infarction. Otherwise, it is likely that individual units and cardiac networks will find their own ways to adapt CG94 recommendations for clinical use.