Abstract / Summary
Capillary refill time (CRT) is a simple marker of peripheral perfusion but suffers from limited reproducibility when assessed visually. We evaluated the performance of the YesReCan software platform for video-based CRT measurement. In this single-center observational study, 126 standardized CRT videos from 34 critically ill patients were analyzed by trained clinicians and using a semi-automated software approach. Reproducibility was assessed using intraclass correlation coefficients (ICC). Agreement, measurement precision, and diagnostic performance for detecting prolonged CRT (> 3 s) were also evaluated. Semi-automated CRT demonstrated excellent reproducibility, with intra- and inter-observer ICCs of 0.945 (95% CI, 0.837–0.995) and 0.992 (95% CI, 0.985–0.997), respectively, compared with 0.640 (95% CI, 0.474–0.765) and 0.603 (95% CI, 0.428–0.736) for clinician assessment. Measurement precision was substantially improved, with a relative least significant change of 40.7% (95% CI, 11.7%–66.9%) versus 74.7% (95% CI, 56.1%–89.1%) for single measurements. Agreement with clinician assessment was moderate (CCC 0.689, 95% CI, 0.533–0.772), and discrimination for prolonged CRT was excellent (AUC 0.885, 95% CI, 0.785–0.962). Video-based CRT analysis substantially improves reproducibility and measurement precision compared with clinician assessment while maintaining excellent diagnostic performance. These findings support the potential of automated CRT measurement to standardize peripheral perfusion assessment in clinical practice and research, although external validation in larger multicenter cohorts is required before widespread implementation. Not applicable.