Prognostic Value of Traditional and Optimised Vital Sign Thresholds for Risk of Serious Adverse Events in Continuously Monitored Hospitalised Patients

Technological advances enable continuous vital sign monitoring with wireless biosensors in general wards. However, traditional thresholds used in routine monitoring do not effectively identify serious adverse events (SAEs). This study aimed to assess the prognostic ability of combinations of threshold severity and duration in continuous vital sign monitoring for SAEs in acute medical and surgical patients. Data on heart rate (HR), respiratory rate (RR), and peripheral oxygen saturation (SpO2) were collected from four prospective observational studies and two randomised controlled trials. The primary outcome was an SAE within 30 days. Traditional thresholds were based on the National Early Warning Score, and optimised thresholds were derived by identifying threshold values and durations with the highest area under the receiver operating characteristic curve (AUROC) for any SAE and specific SAE types. Of 1492 patients, 1077 were included in HR and RR analyses and 761 in SpO2 analyses; 30% experienced an SAE. For any SAE, traditional thresholds had AUROC values of 0.47–0.50 (95% CI 0.43–0.54). Optimised thresholds also showed limited ability (AUROC 0.51, 95% CI 0.44–0.56). In contrast, 18 thresholds for specific SAE types had AUROC > 0.70; the highest AUROC was observed for HR > 104 beats/min for ≥36 min for upcoming respiratory SAEs (AUROC 0.82, 95% CI 0.74–0.90). Optimised thresholds showed low prognostic ability for any SAE but higher AUROC than traditional thresholds. Several thresholds demonstrated moderate prognostic ability for specific SAE groups and warrant validation.