Evaluation of the threshold value for the Early Warning Score on general wards.

Autor: van Rooijen CR; Department of Internal Medicine, Medical Centre Alkmaar, Alkmaar, the Netherlands. c.r.van.rooijen@mca.nl, de Ruijter W, van Dam B
Jazyk: angličtina
Zdroj: The Netherlands journal of medicine [Neth J Med] 2013 Jan; Vol. 71 (1), pp. 38-43.
Abstrakt: Introduction: The Early Warning Score (EWS) is used for early detection of deteriorating vital parameters and has been correlated with adverse outcomes. Unfortunately, neither its value on general wards nor the optimal cut-off value have been investigated. We aimed to find the optimal cut-off value for EWS on general wards, and investigated the possibility to raise this value from EWS ≥ 3 without compromising sensitivity too much.
Methods: From May 2010 until May 2011, EWS was calculated from vital parameters in all patients in medical and surgical wards in the Medical Centre Alkmaar. Cut-off value was defined as EWS ≥ 3, unless otherwise specified. Six responses were defined and categorised as interventions (infusion prescription, medication changes, ICU consultation) and other actions (no action, change EWS cut-off value, oxygen supplementation), and it was registered whenever the threshold was exceeded.
Results: 71,911 EWS values were obtained, 31,728 (44%) on medical wards and 40,183 (56%) on surgical wards. On medical wards, the cut-off value was exceeded 3734 times, and response was registered in 29% of the cases with 141 (12%) interventions. On surgical wards, the cut-off value was exceeded 3279 times, and response was registered in 19% of the cases with 633 (36%) interventions. Sensitivity and specificity for EWS ≥ 3 could not be calculated. For a calculated cut-off at EWS ≥ 4, sensitivity decreased to 74%.
Conclusion: Raising the EWS threshold to 4 on general wards in the hospital would lead to an unacceptable decrease in sensitivity. Therefore, we recommend that the pre-defined cut-off should remain 3, with the possibility to personalise the threshold.
Databáze: MEDLINE