Introduction and objectives An accurate assessment of severity of community acquired pneumonia (CAP) on admission is pivotal in early identification of patients who are critically ill. CURB-65 is recommended by BTS, but is a poor predictor of ICU admissions and often underestimates severity in young patients. We compared this with REA-ICU and SMART-COP in predicting severity and mortality.

Methods The notes of all adult patients admitted with a diagnosis of CAP in June 2016 were reviewed. Inclusion criteria consisted of consolidation on chest radiograph and raised inflammatory markers. Scores were calculated from results obtained within 24 hours of admission. The patients were followed up to ascertain length of stay, complications (effusions, empyema) antibiotic escalation, delivery of non-invasive ventilation (NIV), ICU escalation and death.

Results 43 patients identified with CAP were included in our analysis. 76.7% of patients were ≥65 years old. 24 hours after admission, 39.5% had ward-based ceilings of care in place and 27.9% had no escalation plan documented. 11.6% were still inpatients at the time of analysis. No patients were escalated to ICU.

CURB-65 was 0 to 1 in 23.3%, 2 in 18.6%, and ≥3 in 58.1%. In the low risk group, 50% developed complications, 10% required NIV and there were no inpatient deaths. Amongst the moderate and high risk patients, NIV was administered in 25% and 16% respectively, primarily as the patient’s ceiling of care. Inpatient deaths occurred in 12.5% of moderate risk and 16% of high risk patients.

As shown in Table 1, there was variation in REA-ICU and SMART-COP scores amongst moderate and high risk CURB-65 scores. A high risk CURB-65 score did not correlate with high REA-ICU and SMART-COP scores. We found that higher REA-ICU and SMART-COP scores did not correlate with increased mortality. However, length of stay and antibiotic escalation was increased with higher SMART-COP scores, particularly in those with low CURB-65 scores.

Conclusions CURB-65 score correlates well with mortality, particularly in the elderly group of patients studied. REA-ICU and SMART-COP are better at identifying younger, morbidly ill patients with misleadingly low CURB-65 scores requiring early decisions regarding escalation of care.