Background The conventional method to assess stable chest pain of recent onset is to classify the nature of the chest pain, and then apply a risk model to predict the probability that it is caused by obstructive coronary artery disease (CAD), as recommended in American, European and NICE guidelines. The European and NICE guidelines are derived from the 1979 Diamond-Forrester risk model (DFRM), although this has been criticised for overestimating the risk of CAD. We hypothesised that recent studies would be more consistent and useful than earlier studies in diagnosing CAD.

Methods We performed a systematic literature search on studies published on MEDLINE and EMBASE until Nov 2016. Searched terms were Diamond Forrester and coronary artery disease. Overlapping studies and review articles were excluded. Data on the nature of chest pain and presence of CAD was independently extracted by both authors. Crude relative risks (CRR) of CAD were calculated by comparing typical angina and atypical angina respectively to non-anginal chest pain or pain free as the reference, and not taking into account demographics or cardiovascular risk factors.

Results 10 studies (n=31,528) were eligible for analysis (mean age 59±10, 54% male), as shown in Table 1; these used a variety of different methods to diagnose CAD. Table 2 shows that compared to the original DFRM, more recent studies tended to use cohorts that had larger of patients with atypical angina and non-anginal chest pain with positive diagnoses of CAD varying dramatically; such as of those with typical angina the%age with CAD ranged from 9%–88%. There was also wide and dramatically varying crude relative rates of CAD (Table 2), such that in some studies, having atypical angina conferred a lower crude RR of CAD.

Conclusions Many of the contemporary studies assessing the DFRM constitute a lower risk cohort and we opine that classification of the nature of chest pain in several of them was very different to that used in the DFRM, such that produced striking results not consistent with other data. Although many studies have criticised the DFRM for over-estimating the rate of CAD, we opine that some contemporary studies were intrinsically flawed, which was the opposite of what we had expected.