Ability of high-resolution computed tomography to distinguish Mycoplasma pneumoniae pneumonia from other bacterial pneumonia: Significance of lateral bronchial lesions, less air bronchogram, and no peripheral predominance.

ABSTRACT

BACKGROUND: No study has investigated the capability of high-resolution computed tomography (HRCT) to detect a lateral bronchus abnormality, degree of air bronchogram, and distribution of affected lesions in the diagnosis of Mycoplasma pneumoniae pneumonia (MPP). METHODS: We prospectively enrolled patients with serologically-confirmed MPP or culture-confirmed other bacterial pneumonia (OBP). The distribution of affected areas, abnormalities in lateral bronchial lesions, the degree of air bronchogram, and previously reported findings on HRCT were evaluated for MPP and OBP. Predictive HRCT findings for MPP were determined by logistic regression analysis. We provisionally designed our HRCT criteria (negative, probable, or highly suspected) for diagnosing MPP and investigated the diagnostic yield of the HRCT criteria. RESULTS: Sixty-three MPP and 126 OBP patients were included in this study. Logistic regression analysis showed that the absence of peripheral predominance, bronchial wall thickening, lateral bronchial wall thickening, intralobular or lobular ground-glass opacities, intralobular ground-glass opacities connected to a lateral bronchus, and less air bronchogram in infiltrates were significant predictors of MPP. Our HRCT criteria showed that the sensitivity and specificity in negative, probable, and highly suspected MPP were 0.0 and 0.33, 1.0 and 0.69, and 0.5 and 0.98, respectively. CONCLUSIONS: HRCT had considerable ability to detect a lateral bronchial abnormality and to diagnose or rule out MPP based on the distribution of affected areas, abnormalities in lateral bronchial lesions, and the degree of air bronchogram in the infiltrates.