The Predictive Value of Clinical Signs to Identify Shock in Critically Ill Patients

Background/Objectives: Current guidelines recommend the use of clinical signs to diagnose shock and cellular hypoperfusion in critically ill patients. However, these recommendations are based on limited scientific evidence. The objective was to determine the predictive value of clinical signs to identify shock. Methods: Retrospective cohort study including adult (≥18 years) patients admitted to the critical care resuscitation unit of a tertiary hospital. The primary goal was to determine the predictive value of tachycardia, prolonged capillary refill time (CRT), skin mottling, weak radial pulse, inadequate peripheral perfusion, shock index >0.8, altered mental state, and diaphoresis to identify shock. Two-by-two contingency tables were used for statistical analysis. Results: Three-hundred-seventeen patients (no shock, n = 231; shock, n = 86) were included. As a single clinical sign, skin mottling [sensitivity, 0.38; specificity, 0.92; negative likelihood ratio (LR−), 0.68; positive likelihood ratio (LR+), 4.62], prolonged CRT (sensitivity, 0.44; specificity, 0.89; LR−, 0.62; LR+, 4.17), shock index >0.8 [sensitivity, 0.77; specificity, 0.64; LR−, 0.36; LR+, 2.15], a weak radial pulse [sensitivity, 0.62; specificity, 0.79; LR−, 0.49; LR+, 2.88], and inadequate peripheral perfusion [sensitivity, 0.68; specificity, 0.73; LR−, 0.44; LR+, 2.52] predicted shock. Prolonged CRT, skin mottling, inadequate peripheral perfusion, a weak radial pulse, and a shock index >0.8 predicted shock states with low cardiac output. A shock index >0.8, tachycardia, and a weak radial pulse were predictive of distributive/vasodilatory shock. The accuracy to identify shock were higher if ≥2 clinical signs were present compared to only one. Conclusions: Skin mottling, prolonged CRT, shock index >0.8, weak radial pulse, and inadequate peripheral perfusion can identify patients with shock, particularly shock states with low cardiac output, with high specificity and LR+.