Diagnostic yield of lung point-of-care ultrasound in children with suspected pneumonia in a resource limited setting: a prospective observational study.

Abstract

Pneumonia is a leading cause of mortality for children under five worldwide and is often diagnosed using WHO clinical criteria. In resource-limited settings, accurate diagnosis remains challenging. Lung ultrasound (LUS) is an effective, low-cost diagnostic tool, however, its performance in smaller regional hospitals remains underexplored. We conducted a prospective observational study to evaluate the diagnostic yield of LUS among children hospitalized with WHO-defined pneumonia in Eastern Uganda. All children received a systematic 12-zone LUS by a trained operator using a linear probe in the sagittal plane. LUS was remotely interpreted by two blinded reviewers. Chest X-ray (CXR) was performed when feasible. We enrolled 46 patients with a total of 523 lung zones reviewed. The average age was 22 months (± 26.8 months); 67% of patients demonstrated hypoxemia. LUS was adequate in 42/46 (91%) children. LUS identified pneumonia in 25/42 (59%), bronchiolitis in 2/42 (5%), and normal lung in 15/42 (36%). CXR was completed in 40/46 (87%) children and showed pneumonia in 18/40 (45%). For the 36 children with both LUS and CXR, four (11%) demonstrated pneumonia on CXR but not LUS, compared to 11 (30%) who had pneumonia on LUS but not CXR. Our study demonstrated that LUS is feasible and identified features of pneumonia is 59% of LUS studies. While the overall diagnostic yield of LUS is comparable to CXR, a substantial proportion of children had pulmonary findings detected by only one imaging modality. These findings suggest a complementary rather than redundant role for LUS and CXR.