Validity of severity scores in hospitalized patients with nursing home-acquired pneumonia.

BACKGROUND

Several severity scores have been advanced to predict a patient's outcome from community-acquired pneumonia (CAP). The purpose of this study is to compare the accuracy of confusion, urea, respiratory rate, BP (CURB); CURB plus age ≥ 65 years (CURB-65); CURB-65 minus urea (CRB-65); and systolic BP, oxygenation, age, and respiratory rate (SOAR) scoring systems in predicting 30-day mortality and ICU admission in patients with nursing home-acquired pneumonia (NHAP).

METHODS

A retrospective analysis of a prospectively collected database of 457 nursing home residents hospitalized with pneumonia at two university-affiliated tertiary care facilities. Clinical and laboratory features were used to compute severity scores using the British Thoracic Society severity rules and the SOAR criteria. The sensitivity, specificity, and positive and negative predictive values were compared for need for ICU admission and 30-day mortality.

RESULTS

The overall 30-day mortality and ICU admission rates were 23% and 25%, respectively. CURB, CURB-65, and CRB-65 performed similarly in predicting mortality with areas under the receiver operating characteristic curves (AUCs) of 0.605 (95% CI, 0.559-0.650), 0.593 (95% CI, 0.546-0.638), and 0.592 (95% CI, 0.546-0.638), respectively, whereas SOAR showed superior accuracy with an AUC of 0.765 (95% CI, 0.724-0.803) (P < .001). The need for ICU care was also better identified with the SOAR model compared with the other scoring rules.

CONCLUSIONS

All three British Thoracic Society rules had lower performance accuracy in predicting 30-day mortality of hospitalized NHAP than SOAR. SOAR is also a superior alternative for better identification of severe NHAP. An improved rule for severity assessment of hospitalized NHAP is needed.