Development of a Model for Early Differentiation of Adenovirus Pneumonia from Pneumonia

Background: Adenovirus pneumonia (AVP) and pneumonia (MPP) have similar clinical manifestations such as a high prevalence of lung consolidation, making the differential diagnosis difficult before the etiology is reported. This study aimed to compare AVP and MPP, and to build a predictive model to differentiate them early.

Methods: We selected 198 cases of AVP and 876 cases of MPP. Clinical manifestations, computed tomography (CT) features, and biomarkers were compared. A logistic regression model was built to predict AVP. The area under the curve (AUC) of the receiver-operating characteristic was calculated to evaluate the discriminant ability of the prediction model.

Results: Patients in the AVP group were mainly infants and toddlers, while the MPP group had more pre-school age children. The rate of hypoxemia and severe pneumonia was 3- and 11-times higher, respectively, in the AVP group than in the MPP group (5.6% 1.8%, 27.8% 2.5%, P<0.01). The proportion of patients with a Pediatric Logistic Organ Dysfunction-2 score ≥2 was 10 times higher in the AVP group than in the MPP group (17.4% 1.7%, P<0.01). Bilateral pneumonia was present in 90.2% of the AVP group. Biomarkers, such as interleukin (IL)-2 receptor, IL-10 and lactic dehydrogenase (LDH), were considerably higher in the AVP group than in the MPP group (P<0.01). The predictive model included eight variables, namely: age, severe pneumonia, bilateral pneumonia, ground-glass attenuation, consolidation, atelectasis, C-reactive protein, and LDH. The AUC was 86.6%.

Conclusions: Compared with MPP, AVP affects younger children, presents a more severe respiratory tract involvement, results in a larger range of lung lesions, and is associated with higher inflammatory biomarkers. Our predictive model includes a combination of clinical features, imaging findings, and biomarkers. It may help pediatricians in the early differentiation of AVP from MPP.