Introduction: According to the Dutch guidelines, severity of community acquired pneumonia (CAP) (mild, moderate-severe, severe) should be based on either PSI, CURB65 or a 'pragmatic' classification. In the last mentioned, the type of ward of admission, as decided by the treating physician, is used as classifier: no hospital admission is mild, admission to a general ward is moderate-severe and admission to an intensive care unit (ICU) is severe CAP. Empiric antibiotic recommendations for each severity class are uniform. We investigated, in 23 hospitals, which of the three classification systems empirical treatment of CAP best adhered to, and whether a too narrow spectrum coverage (according to each of the systems) was associated with a poor patient outcome (in-hospital mortality or need for ICU admission).
Patients and methods: Prospective observational study in 23 hospitals.
Results: 271 (26%) of 1047 patients with CAP confirmed by X-ray were categorised in the same severity class with all three classification methods. Proportions of patients receiving guideline-adherent antibiotics were 62.9% (95% CI 60.0-65.8%) for the pragmatic, 43.1% (95% CI 40.1-46.1%) for PSI and 30.5% (95% CI 27.8-33.3%) for CURB65 classification. 'Under-treatment' based on the pragmatic classification was associated with a trend towards poor clinical outcome, but no such trend was apparent for the other two scoring systems.
Conclusions: Concordance between three CAP severity classification systems was low, implying large heterogeneity in antibiotic treatment for CAP patients. Empirical treatment appeared most adherent to the pragmatic classification. Non-adherence to treatment recommendations based on the PSI and CURB65 was not associated with a poor clinical outcome.