Background: Community-acquired pneumonia (CAP) is the major manifestation of Q fever, an emerging disease in French Guiana. Consequently, the empirical antibiotherapy used for the treatment of CAP combines doxycycline and the recommended amoxicillin. Our objectives were to estimate the prevalence of Q fever pneumonia and to build a prediction rule to identify patients with Q fever pneumonia for empirical antibiotic guidance.
Methods: A retrospective case-control study was conducted on inpatients admitted with CAP in the Department of Infectious Diseases of Cayenne Hospital from 2004 to 2007. Serodiagnosis for Coxiella burnetii was performed for all patients. Risk factor analysis was performed using multivariate logistic regression, and a prognostic score was computed using bootstrap procedures. The score performance characteristics were used to choose the best prediction rule to identify patients with Q fever pneumonia.
Results: One hundred thirty-one patients with CAP were included and the Q fever pneumonia prevalence was 24.4% (95% confidence interval [CI], 17.1-31.9). In multivariate analysis, male sex, middle age (age, 30-60 years), headache, leukocyte count <10 × 10(9)/L and C-reactive protein level >185 mg/L were independently associated with Q fever pneumonia. Patients with a predictive score ≤3 had a low risk of Q fever pneumonia with a negative predictive value of 0.97 (95% CI, .90-1) and a sensitivity of 0.97 (95% CI, .89-1).
Conclusions: The prediction rule described here accurately identifies patients with low risk of Q fever pneumonia and may help physicians to make more rational decisions about the empirical use of antibiotherapy. Further prospective studies should be performed to validate this score.