Introduction The CURB65 score was developed to predict mortality in community acquired pneumonia (CAP) but is often used in pneumonia complicating acute exacerbations of COPD (pAECOPD). We have previously shown that CURB65 underestimates in-hospital mortality in pAECOPD, particularly in low risk patients (observed mortality 11.2%, CURB65-predicted 1.5%). Of importance, CURB65 was derived in a population with significant exclusions, notably admission from nursing home, and few patients with dementia were included, whereas in our DECAF AECOPD cohort  such patients were included. The higher than predicted mortality in pAECOPD may reflect additional risk conferred by co-existent COPD, a less selected population and/or clinical outcomes in participating hospitals. We have therefore investigated whether the mortality of an equivalent population with CAP, but without COPD, is similar to that found previously in pAECOPD.
Methods Patients admitted with a primary diagnosis of CAP were identified from coding records. Patients with confirmed or suspected COPD were excluded; selection criteria and time frame otherwise matched the DECAF cohort. Demographic, clinical and mortality data were gathered from clinical notes. Categorical variables were compared using Fisher’s exact test.
Results 115 patients with CAP were included: mean (SD) age 72.1 (16.4) years, 29.6% were admitted from institutional care and 21.7% had dementia. Median (IQR) CURB65 score was 2 (1–3) and in-hospital mortality 16.5%. Compared to the earlier cohort with pAECOPD, mortality in patients with low or intermediate risk CURB65 scores was lower.
In the present study, 74% of deaths occurred in patients admitted from institutional care (mortality 35%, non-institutional care 9% p = 0.002) and/or those with dementia (mortality 36%, without dementia 11% p = 0.006).
Conclusions Compared to the BTS national audit, the proportion of patients with severe pneumonia is higher (49% v 30%) and mortality lower (16.5% v 21.2%). Both dementia and admission from institutional care were associated with high mortality rates. Among patients with low or intermediate risk CURB65 scores the mortality of those with CAP without COPD was lower than we previously found in pAECOPD, confirming that the underestimation of mortality risk by CURB65 in pAECOPD was not attributable to less effective clinical care.
Steer. The DECAF score. Thorax, 2012;67:970–6.