• Pneumonia
• respiratory infection

Ewig et al are to be commended for their very large study of 388 406 patients admitted with community-acquired pneumonia (CAP) in German hospitals from 2005 to 2006.1 Using the CRB-65 tool (confusion, respiratory rate ≥30 min, low blood pressure (either systolic <90 mm Hg or diastolic ≤60 mm Hg) and age ≥65 years), the authors found 30-day mortality rates of 2.4, 13.4 and 34.4% in those with 0 points, 1–2 points and 3–4 points, respectively. As a result, the authors promote this tool as being accurate for predicting CAP-related deaths.

However, while this appears impressive, it is notable that of the >54 700 deaths, only 29.0% were classed as high risk, whereas 68.1% were only intermediate risk and 2.8% were low risk. In addition, many of those patients who died had treatment limitations applied and only 15.7% of the patients who died received ventilatory support. These two points raise the question of how clinically useful this tool really is. If over two-thirds of deaths were classed as having clinically ‘moderate’ CAP, then the tool cannot really be described as being accurate for this purpose. Furthermore, if the vast majority of people who died did so after active treatment was withdrawn, then the identification of such patients does not appear to serve much purpose. It would be more relevant to assess such a tool for its ability to identify those patients in whom every effort is made to save their lives—that is, those admitted to the intensive care unit.