Diagnostic Performance of Point of Care Ultrasound Compared to Chest X-Ray in Patients with Hypoxia at a Teaching Hospital Emergency Department in Uganda

Background: Hypoxia is a common presentation in the Emergency Department (ED) worldwide. It affects 9-12% of hospitalized adults in Sub-Saharan Africa. Timely diagnosis of the multiple causes such as pneumonia, heart failure among others is challenging. Chest X-Ray (CXR), one of the most utilized imaging modalities has many limitations, and the gold standard (Computed Tomography scan) is inaccessible. Point of care ultrasound (PoCUS) is more available and increasingly being used, however little is known of its performance in resource limited EDs. The study aimed to assess the diagnostic performance of PoCUS compared with CXR in identifying the causes of hypoxia in the medical ED.Methods: 49 adults presenting with hypoxia (SP02 <= 88%) in the medical ED were evaluated. Ultrasound of the lungs and heart (PoCUS) was done, then CXR obtained. Lung ultrasound (LUS) was compared with CXR (first reference standard). Chest X-Ray and PoCUS were each compared to the physician diagnosis (second reference standard) to determine agreement using an acceptable disagreement cut-off of 15%.Results: 31% more abnormalities were identified by LUS than CXR. Lung ultrasound findings agreed with CXR in 86% of the participants with moderate reliability ( kappa=0.75). There was no significant difference between the actual findings of the two tests (X 2 = 2, p 0.1). Using the second reference, 82% of the CXRs were similar with weak reliability ( kappa=0.5) compared to 98% of PoCUS findings with strong reliability ( kappa=0.9). Compared to PoCUS, CXRs significantly differed from the physician diagnosis (X 2 = 0.85, p 0.38 vs X 2 = 8.5, p 0.004 respectively).Conclusion: Overall, PoCUS was not inferior to CXR when compared to final physician diagnosis in identifying causes of hypoxia, and LUS and CXR had comparable performance. Significantly more abnormalities were iden-tified on PoCUS and it demonstrated better agreement and strong reliability with the physician diagnosis than CXR. We recommend PoCUS use in patients with hypoxia attending resource limited in-and pre-hospital settings.