Haasenritter et al1 performed an external validation of the Marburg Heart Score (MHS), a clinical prediction rule to rule out coronary heart disease (CHD) in patients presenting with chest pain in primary care. We read this potentially important article with great interest because ruling out CHD in primary care is of special concern. The authors concluded that, according to its generalisability, ease of application, and accuracy, its use in clinical practice is recommended.
However, we have some doubts about their outcome measure and conclusion. The outcome measure, the reference diagnosis, was established using a delayed-type reference standard and an expert panel. Our main concern was that this expert panel was not blinded to the results of the index test. The authors acknowledged this problem, but stated that blinding of this panel would have led to fewer available data for this study. In addition, another study showed a ‘substantial and satisfying’ agreement (kappa = 0.62) between a blinded and unblinded panel. We think that having used two independent experts without knowledge of the MHS, blinding without loss of data would have been possible without risk of bias. Furthermore, the reported agreement was derived from another study, and is therefore not generalisable to this study. We would be inclined to rate a kappa of 0.62 at best as moderate rather than ‘substantial and satisfying’.
The authors report an impressive negative predictive value of 97.9%. Nevertheless, still one in 50 patients with CHD would have been missed using the MHS. Moreover, four of 21 patients with acute coronary syndrome (ACS) were falsely classified as ‘CHD-negative’. In our opinion, missing almost one in five patients with ACS does not justify recommending the MHS for use in clinical practice. Besides, the low positive predictive value may lead to more unnecessary investigations and costs.
Lastly, the authors did not demonstrate that there is a strong need for the MHS, nor published data that the MHS performs better than a GP’s own judgment based on common practice. Apart from statistical evidence, do GPs feel that the MHS will positively contribute to their diagnostic practice?
Therefore, it is hard for us to see the diagnostic accuracy of the MHS in the right perspective and estimate its clinical relevance. In our opinion, it is premature to recommend the MHS. Nevertheless, we would like to encourage the authors to continue validating the MHS, for example, in a prospective cohort study, and demonstrate its surplus value.
- © British Journal of General Practice 2012