Management of abnormal observations in the emergency department: A review

Objective: To assess utility and accuracy of general observation modified early warning score charts; and compare sensitivity and specificity of single- and multiple-parameter-based trigger scores on patient outcomes in the ED. Methods: Retrospective cohort clinical audit of all adult Modified Early Warning Score charts in the ED of a mixed tertiary hospital over 4 weeks. Data extracted included recorded parameters required to calculate Modified Early Warning Score and evidence of response. Results: Of 5901 ED presentations, medical records system identified 2482 Modified Early Warning Score; 347 were missing or blank. Of 2135 Modified Early Warning Score charts, 19.5% contained a calculation error, 51.9% had one or more missing parameters and 36.6% did not have usual/target systolic blood pressure recorded; with 25.1% (95% confidence interval [CI] 23.3–27.0) charts correctly completed. Four hundred and forty-three had a single-abnormal parameter of which chart review showed 96.6% (94.5–97.3) were identified as abnormal by nurses with 25.7% (21.9–30.0); only 5.6% (3.9–8.2) had evidence of recognition by medical staff. Modified Early Warning Score sensitivity and specificity for ward admission was 14.7% and 96.1%, respectively. Modelling using the dataset of a single-abnormal parameter suggested sensitivity and specificity of 31.6% and 85.8%, respectively. Conclusions: This study highlights serious deficiencies in documentation of abnormal parameters and emergency response. It has also shown poor accuracy of both single- and multiple-parameter-based trigger scores in predicting patient outcomes within the ED. However, single-parameter-based trigger scores are twice as sensitive as total Modified Early Warning Score for admission and reduces documentation error by 23%.