SESSION TITLE: Rapid Session: Bedside Imaging SESSION TYPE: Original Investigation Slide PRESENTED ON: Wednesday, November 1, 2017 at 07:30 AM - 08:30 AM PURPOSE: Bedside ultrasonography (US) may facilitate rapid diagnosis of thoracic pathology in the Intensive Care Unit (ICU). The feasibility and accuracy of bedside thoracic US performed by acute care nurse practitioners (ACNP) in the ICU are unknown. We hypothesized that bedside thoracic US by ACNPs in the ICU would be feasible and have a high sensitivity and specificity for thoracic pathology visualized on CXR. METHODS: We conducted a prospective observational cohort study in a single medical ICU. Prior to study initiation, all 10 ACNPs staffing the ICU attended a focused, peer-led thoracic US training session, which included didactic instruction, hands-on practice of image acquisition, and image interpretation. During the study, each time a CXR was ordered by treating clinicians, ACNPs were asked to complete a bedside thoracic US prior to interpretation of the CXR. The ACNPs then documented in a secure online database the presence or absence of five key sonographic features in each of the four chest regions scanned. If an US was unable to be performed, the reason was documented. RESULTS: From July 2015- July 2016, 335 portable CXRs were ordered by treating clinicians. In 246 (73%) cases, a thoracic US was completed by the ACNP prior to interpretation of the CXR. In 89 (27%) cases, a thoracic US was not performed, most frequently due to inadequate ACNP time due to other urgent patient care needs. Among the 246 cases with US performed prior to CXR, the CXR identified parenchymal pathology in 141 (57.3%) cases, pleural effusion in 75 (30.6%) cases, and pneumothorax in 3 (1.2%) cases. For parenchymal abnormality on CXR, US demonstrated a sensitivity of 83.7% (95% CI 76.5-89.4%), specificity of 67.6% (95% CI 57.8-76.4%), positive predictive value of 77.6% (95% CI 72.3-82.2%), and negative predictive value of 75.5% (95% CI 67.5-82.1%) (kappa 0.52, P < 0.001). For pleural effusion on CXR, US demonstrated a sensitivity of 68.0% (95% CI 56.2-78.3%), specificity of 85.9% (95% CI 79.7-90.7%), positive predictive value of 68.0% (95% CI 58.7-76.1%), and negative predictive value of 85.9% (95% CI 81.3-89.5%) (kappa 0.54, P < 0.001). The negative predictive value of US for pneumothorax on CXR was 99.6%. The sensitivity and specificity of ultrasound for any pulmonary pathology was 87.7% (95% CI 82.2-92.0%) and 86.3% (95% CI 73.7-94.3%), respectively. The positive predictive value of ultrasound for pulmonary pathology was 96.1% (95% CI 92.5-98.0%). CONCLUSIONS: Bedside thoracic ultrasound demonstrated moderate sensitivity and specificity for pulmonary parenchymal abnormalities and pleural effusions compared to CXR. It demonstrated high sensitivity for pneumothorax and high positive predictive value for any pulmonary pathology. CLINICAL IMPLICATIONS: Bedside thoracic US performed by ACNPs can frequently be completed prior to CXR performance and shows moderate agreement with CXR findings. The primary limitation of bedside thoracic US by ACNPs compared to CXR may be the competing demands of other urgent patient care. DISCLOSURE: The following authors have nothing to disclose: Lisa Flemmons, Sarah Bloom, Todd Rice, Arthur Wheeler No Product/Research Disclosure Information