Abstract 234: The Variability of Visual Assessment of Traditional Capillary Refill Measurement

Abstract

Objective: Capillary refill time (CRT) measured at the bedside is widely promulgated as an acceptable method to identify patients in shock. However, inter-observer reliability of CRT visual assessments is questionable. To investigate the variability that occurs when healthcare providers (HCP) visually assess CRT, we conducted a study in the emergency department (ED) that simultaneously measured CRT while recording the change in the patients’ fingertip color by video. Methods: Three HCPs were selected to perform manual CRT assessments at the bedside to classified patients as having either normal (≤2 seconds) or abnormal (>2 seconds) CRT. An attending ED physician, blinded from the HCP classification, quantitatively measured CRT using a chronograph ( visual CRT ). A video camera was mounted on top of the hand tool in order to obtain a digital recording of the change in color, as the fingertip was compressed. The videos were thereafter used to calculate CRT via image software analysis ( image CRT ). Additionally, nine HCPs, including the ED physician, reviewed the videos in a separate setting to visually assess CRT while blinded from any patient information ( video assessment CRT ). Results: Thirty ER patients were enrolled in the ED. The ED physician identified 10 patients with abnormal CRT (> 2.0 seconds), while only two patients were identified by HCPs. Mean visual CRT was 2.0±0.9 (range: 1.2-4.4) seconds; mean image CRT, 2.4±2.1 (range: 0.5-8.0); mean video assessment CRT by the ED physician, 1.8±0.6 (range: 1.0-2.7). The correlation between visual CRT and image CRT was strong (r=0.648, p=0.002), while it was weak between visual and video assessment CRT (r=0.312, p=0.18). Inter-observer reliability of video assessment CRT among HCPs was low (intra correlation coefficient: 0.15, 95% CI 0.05-0.33). Conclusions: Inter-observer reliability of visual CRT assessment is low. The reliability of skilled physician in a real clinical situation increases and it infers that a reliable CRT test may only be capable if other patient information is available at the bedside, such as patient’s background, distressed appearance, cold peripheral temperature etc.