SESSION TITLE: Outcomes/Quality Control Posters
SESSION TYPE: Original Investigation Poster
PRESENTED ON: Wednesday, October 30, 2013 at 01:30 PM - 02:30 PM
PURPOSE: To determine the value of level of consciousness (LOC) in an early warning score (EWS) model.
METHODS: Two EWS models (EWS1 and EWS2) each containing systolic blood pressure, heart rate, respiratory rate and temperature were used. EWS1 also contained a LOC determination using the Glasgow Coma Score (GCS). EWS2 contained age and oxygen saturation without LOC. Both scores were calculated on medical/surgical patients of a 176 bed community hospital from 7/2011 to 1/2013. EWS1 without GCS and EWS2 with GCS were also calculated. A valid data set required vital signs to have been taken within a 6 hour window and a GCS within 24 hrs. Receiver Operating Characteristic (ROC) curves were determined for all 4 calculations.
RESULTS: 11,630 scores (patient mean age 57 years, standard deviation: 19, 62.5% female) were determined for all 4 variations. Including LOC did not significantly change EWS performance regardless of score used. The values for EWS1; EWS1 without GCS; EWS2; EWS2 with GCS respectively are: ROC: 0.77, 0.75, 0.75, 0.76. Sensitivity: 0.40, 0.31, 0.26, 0.34. Specificity: 0.96, 0.98, 0.98, 0.95. Negative Predictive Value (NPV): 0.97, 0.97, 0.96, 0.97. Positive Predictive Value (PPV): 0.32, 0.50, 0.35, 0.25.
CONCLUSIONS: Use of LOC in an EWS adds little benefit to its predictive capability. Similar to previously published EWS models for medical/surgical wards, where prevalence of deterioration is less than 10%, our scores have low sensitivities and PPV with acceptable specificities and NPV.
CLINICAL IMPLICATIONS: Published EWS models use differing variables, weightings and point allocations. With few exceptions these are based on clinical intuition. Virtually all EWS models contain a LOC component, however to date, data does not exist as to its contribution to the EWS. Comparative data on over 11,000 scores demonstrates the manual determination of LOC does not aid significantly in the score determination. As health systems move to automated vital sign determinations, removing LOC would ease hospital workflow allowing time to address other clinical activities and potentially save costs.
DISCLOSURE: Lin Yang: Employee: Employee of Philips Research Erkan Hassan: Employee: Employee of Philips Healthcare William Lord: Employee: Employee of Philips Research Larry Eshelman: Employee: Employee of Philips Research Julie Reisetter: Other: Principal Investigator from Banner Health, who sponsors Clinical Site researcher from Philips Research
No Product/Research Disclosure Information