Real-time x-ray abnormality alerts for emergency departments using a radiographer comment model - a multisite pilot study

Abstract

The timely communication of clinically significant image appearances to Emergency Department (ED) referrers is necessary for optimum patient care. Australian reliance on verbal communication only is time-limited, open to misinterpretation and lacks transparency. A combined radiographer alert and comment model was designed to reliably communicate image abnormalities to ED referrers in real-time. A multidisciplinary steering group designed the model for all ED general imaging. Protocols were developed to document radiographer comments (critical, urgent and clinically significant) in patients' medical records. Critical findings were communicated directly to ED. Five NSW hospitals varying in size, complexity and population demographics piloted the model between three to twelve months during 2021-2022. Site auditors compared comments with the radiology report and designated each as True Positive (TP), False Positive (FP), indeterminate and clinically significant. Indeterminate cases were analysed by an external radiologist. Inter-observer consensus was obtained for all classifications via two independent auditors. The Positive Predictive Value (PPV), or precision of the comment, was calculated for each site. Radiographers (n = 69) provided comments for 1102 cases. The pooled average PPV for TP was 0.96; (0.947-0.971; 95% CI). The weighted mean error (FP comments) was 3.9%; (2.9% - 5.3%.; 95% CI). The Radiographer Comment model provided consistent levels of commenting precision and reproducibility across a range of sites with a pooled average PPV (0.96). The False Positive rate or weighted mean error (FP) of 3.9% (2.9% - 5.3%.; 95% CI) was low. A strategic, interprofessional approach in the implementation of an image alert combined with a Radiographer Comment can be adapted across a variety of hospital settings for ED and other departments.