Background The NICE guideline for chest pain of recent onset recommends diagnosis of angina based on clinical assessment and estimated likelihood of coronary artery disease (CAD). Pre-test likelihood (PTL) estimates are provided in a simplified table based on the Pryor risk equation. If the PTL of CAD is <10% or >90%, further diagnostic testing is not required. If 10%–29% cardiac CT should be offered, if 30%–60% functional imaging, and if 61%–90% coronary angiography. We compared risk estimation methods to determine how much use of the table in an uncritical manner vs use of the full risk equation led to a different referral pattern. We also investigated whether interpolation for risk factor profiles would improve assessment.
Methods Data were collected prospectively for all patients referred to the Rapid Access Chest Pain Clinic, Newcastle upon Tyne, between February 2002 and August 2011. For each patient, PTL of CAD was assessed by three methods: (1) Reference to the NICE table based on chest pain (typical or atypical), age in deciles, sex and risk factors—smoking, hyperlipidaemia and diabetes dichotomised into high and low risk. For high risk, all three risk factors needed to be present; otherwise the patient was assumed low risk. (2) As per method one, but risk estimates were interpolated between low and high risk values in the NICE table according to the number of risk factors. (3) Calculation by the Pryor equation which includes, in addition, age in years, prior MI, ECG Q waves, and ST/T changes.
Results Out of an initial 7022 patients, 1820 were excluded as they had non-anginal chest pain. This left 5202 patients, 2889 with atypical angina and 2313 with typical angina. The number (%) of patients in the five different risk groups by the three risk assessment methods is given in Abstract 140 table 1. Use of the risk equation compared to the table led to half as many people in the category <10%. Fewer patients would be referred for cardiac CT, more for functional imaging and more for invasive coronary angiography. Use of the adapted table correlated with the risk equation better, but there were still discrepancies, and the percentage of patients for coronary angiography would increase overall (Abstract 140 table 1). The NICE table would classify only 39% of patients with PTL <10% correctly, 60.3% of those with PTL 10%–29%, 44.9% of those with PTL 30%–60%, 62.9% of those with PTL 61%–90%, and 86.3% of those with PTL >90% (Abstract 140 figure 1). The corresponding figures correctly classifying people in the five risk categories using the adapted table were 92.9%, 86.4%, 65.2%, 71.2% and 85.9% respectively (Abstract 140 figure 2).
Conclusion Use of the NICE table uncritically for assessing PTL of CAD misclassifies a significant proportion of people for further investigation. It is important to include ECG data, interpolate risk factor information, and take into account actual age when using the table rather than the Pryor equation, use of which takes these factors into account automatically.
- Pre-test likelihood
- coronary artery disease
- chest pain