An acute deterioration of a patients' general condition is often preceded by changes in individual vital parameters. An early warning system (EWS) shall be developed with a reduced number of physiological and individual parameters, compared to conventional early warning systems; and an algorithm will be generated that is able to predict clinical deterioration. Its predictive power and accuracy shall be investigated. In a second exploratory phase, different model variants will be analyzed and the applicability of the model variants in the context of continuous EWS on wearables will be examined.
An acute deterioration of a patients' general condition is often preceded by changes in individual vital parameters and may lead to adverse events, such as admission to the intensive care unit, heart attack or death. Some of them are potentially avoidable if appropriate measures are taken in a timely manner. Therefore early warning systems (Early Warning Scores= EWS) have been developed from a set of several physiological measurements, signs and symptoms. Individual parameters are weighted to sum up a score. Based on this score, the deterioration of a patients' general condition may be indicated and a predetermined reaction from the professional staff be triggered (so-called track-and-trigger system). It is important to determine all parameters since missing values influence the informative value of an EWS. This requires a higher effort by the staff and is one of the reasons why early warning systems are not yet used systematically in Switzerland. A reduction in the number of parameters to be measured could lower the hurdle for the use of these tools and enable a broader applicability. Therefore an early warning system shall be developed with a reduced number of physiological and individual parameters, compared to conventional early warning systems; and an algorithm will be generated that is able to predict clinical deterioration. Its predictive power and accuracy shall be investigated, based on various clinical outcomes such as mortality, cardiac arrest, transfer to the intensive care unit or sepsis. Retrospective, encrypted patient data (from 2016 until 2022) will be used to develop a statistical prediction model. In a second exploratory phase, different model variants will be analyzed and the applicability of the model variants in the context of continuous EWS on wearables will be examined.
Study Type
OBSERVATIONAL
Enrollment
210
Data collection of patient parameters (heart rate, respiratory rate, clinical outcomes (death, transfer to intensive care unit, adverse events like sepsis, infection, heart attack))
University Hospital Basel, Division of Internal Medicine
Basel, Switzerland
Accuracy of Early Warning Score
The accuracy of the early warning system in terms of its predictive power is measured by using the respective medical outcomes. Data models are used to create a relationship between the patient parameters (the predictors) and the clinical outcome (Condition stable vs. condition unstable: death, transfer to intensive care unit, heart attack, infection/sepsis, etc.) as a binary response variable (whether a deterioration occurs or not).
Time frame: up to 72 hours after hospital admission
This platform is for informational purposes only and does not constitute medical advice. Always consult a qualified healthcare professional.