Barlow et al1 have shown effectively that CURB-65 outperforms generic early warning scores in the prediction of 30-day mortality from community acquired pneumonia (CAP). We are concerned, however, that stratification of 30-day mortality is taken without reflection to be an indicator of requirement for higher levels of care. In our own institution 5 of 40 patients presenting with CAP and an initial CURB-65 score of 1 required admission to high dependency or intensive care, and it has been recognised previously that a CURB-65 score may be misleadingly low in the young and otherwise fit.2 Early warning scores were initially developed based on unexpected admissions to intensive care rather than mortality,3 and it could in fact be argued that the better an early warning score is applied, the less it will relate to mortality as more deaths will be prevented. We have shown that a modified early warning score may not compare with CURB-65 for mortality prediction but outperforms it significantly in terms of prospectively predicting the need for hospital admission and for a higher level of care,4 while Bynd et al have demonstrated the value of the original medical early warning score in predicting hospital admission.5 Using simple physiology, emergency department research has demonstrated the ability to identify at first presentation those patients who are likely to deteriorate.6

We would suggest that mortality predictors and early warning scores are in fact different entities, developed from different cohorts and with different aims, and that conflation of the two should be avoided.