The Utility of Early Warning Score (EWS) in Predicting Serious Adverse Events Among Patients Presenting to Emergency Departments of Muhimbili National Hospital and MUHAS Academic Medical Center

Background: Early Warning Scores (EWS) have been widely studied and applied in high income countries (HIC) for more than a decade, especially in Europe. However, the EWS in low-income countries(LIC’s) such as Tanzania has not been assessed. Given the high acuity and wide range of challenges in management of critically ill patients in the emergency departments (ED) in LIC’s, it is possible that having an EWS would improve patient management and outcomes.

Methods: A prospective study of adult non-trauma, non-pregnant patients from October 2018 to January 2019 at Muhimbili National Hospital (MNH) and Muhimbili University of Health and Allied Sciences’ (MUHAS) Academic Medical Center (MAMC) where blood pressure (BP), pulse rate, respiratory rate, oxygen saturation, temperature, Alert, Verbal, Pain and Unresponsive (AVPU) were used to calculate an EWS obtained at 15 minutes (EWS1) and after 2 hours (EWS2). The primary outcomes were need for mechanical ventilation, cardiac arrest and death (regarded as serious adverse events (SAEs)) in the ED. Descriptive non parametric data were reported with median and interquartile range while a proportion was used to describe incidence of categorical descriptive variables.

The Area under the receiver operating characteristic curve (AUROC) curve was drawn to calculate for optimum cut-off of the EWS, sensitivity and specificity. The odds ratio with 95% confidence intervals was used to determine predictors of serious adverse events, P value of <0.05 was considered statistically significant

Results: We enrolled 527 patients with the median age of 52.0 (interquartile range (IQR) 38.5-66.0) years and 52.0% were males. A total of 522 (99%) out of the 527 received an initial EWS assessment (EWS1) and 507 (96.2%) received a second EWS (EWS2). The median EWS1 was 5.0 (IQR 2-5) and ranged from 0 to 14, whereas the median for EWS2 was 3.0 (IQR1-5) and ranged from 0 to 15. A total of 62 (11.9%) patients developed serious adverse events during the initial assessment (EWS1) with the median EWS in this group was 6.4 (IQR4-9) while patients who did not develop SAEs 460 (88.1%) had a median EWS1 of 4.0 (IQR 2-4). A total of 54 out of 507 who were recorded in the second assessment developed SAE’s with the median of EWS2 of 5.4 (4-8) whereas the remained 453 patients who did not develop serious adverse events had the median early warning score (EWS2) of 3.0 (IQR1-5), p <0.001. The AUROC-1 curve for EWS1 for overall SAEs was 0.76 (p-value 0.0001, 95% CI 0.69-0.83), with the optimum cut-off point (EWS) score of 5.5, which had a sensitivity of 72.5% and specificity of 70 %. AUROC -2 curve for EWS2 for overall SAEs was 0.75 (p-value 0.0001, 95% CI 0.68-0.82) with the optimum cut-off (EWS2) score of 4.5 and sensitivity of 70% and specificity of 72 %.

Conclusion & Recommendations: Use of EWS could provide a useful objective longitudinal monitoring system in the resuscitation units to identify early those patients with greatest risk of undergoing serious adverse events.

Further studies should focus on validation using heterogeneous population, sequential scoring of EWS, in which providers would be aware to see if EWS prevents serious events.