Purpose: To evaluate the performance of a rapid, bedside whole blood C-reactive protein test as a diagnostic test for pneumonia in adults.
Methods: We enrolled consecutive adults who presented with acute cough (duration < or =3 weeks). A fingerstick blood specimen for C-reactive protein level was obtained. Patients also provided information about demographic characteristics and symptoms. Physical examination findings, diagnoses, and treatments were abstracted from the medical record; illness duration and subsequent office visits were determined with follow-up telephone calls. A clinical prediction rule for pneumonia was calculated for each patient and compared with C-reactive protein levels.
Results: Twenty (12%) of the 168 patients in the study had radiographic evidence of pneumonia. Median C-reactive protein levels were significantly higher for patients with pneumonia than in the remaining patients (60 mg/L vs. 9 mg/L, P <0.0001). The area under the receiver operating characteristic (ROC) curve for C-reactive protein level as a predictor of pneumonia was 0.83. C-reactive protein level and the clinical prediction rule were independently associated with pneumonia, yielding a combined area under the ROC curve of 0.93. C-reactive protein level was not associated with hospitalization or resolution of symptoms.
Conclusion: C-reactive protein levels could be a valuable addition to clinical prediction rules for pneumonia. A C-reactive protein level > or =100 mg/L might be a useful indication for chest radiography or empiric antibiotic therapy when the diagnosis of pneumonia is in doubt.