Date: October 14 2021
Summary: A note that provides an overview to intensive care unit scoring systems
Keywords: ##overview #intensive #care #unit #rating #scoring #prediction #quality #care #archive
A. Rapsang and D. C. Shyam, "Scoring systems in the intensive care unit: A compendium," Indian Journal of Critical Care Medicine, vol. 18, no. 4, pp. 220–228, Apr. 2014, doi: 10.4103/0972-5229.130573.
This paper was useful as it provided a good overview of intensive care unit scoring systems. It enumerated their characteristics and also talked about their advantages. The paper also went over a variety of scoring systems but as I was concerned with learning what scoring systems in general were, I stopped there.
What is one approach clinical staff in an intensive care unit utilize to predict patient outcomes and to improve clinical decisions?
In short, intensive care unit scoring systems!
What is an intensive care unit scoring system?
These scoring systems (also referred to as severity scales) are calculated alongside traditional diagnosis methods. They are used to:
Predict patient outcomes
Compare quality of care
Create breakdowns for clinical trials
They are handy for also analyzing unexpected patient outcomes and what precipitated the outcome.
What is a scoring system comprised of?
A scoring system generally has two components:
Score - a number assigned to a given disease based on severity.
Probability Model - determines the probability of patient deaths.
How are scoring system probability models used?
In practice, a model gives clinical decision makers additional insight into comparing patient populations for:
Treatment
Triage
Comparative analysis
Understanding treatment effectiveness
Optimizing resources
What are the three characteristics of an ideal model for scoring?
An ideal model must be:
Valid
Calibrated
Discriminated
What does it mean for a model to have "validity"?
"Validity" means how well a model performs on test datasets during the model's creation.
What does a "calibrated" model mean?
The accuracy between estimated mortality probabilities and the actual mortalities experienced by patients. What is key to note is that it is also able to be statistically robust/rigorous. [1] Essentially, is the model well suited to the actual patients being evaluated?
What does a model that is "discriminatory"?
How well the model can determine what patients live and what patients die in practice. Examples of measures used to understand discriminatory patterns in a model are:
Sensitivity
Specificity
False positive rate
False negative rate
Positive predictive power
Misclassification rate
Area under the receiver operating characteristic curve and
Concordance. [2]
Zelko, Jacob. Scoring systems in the intensive care unit: A compendium. https://jacobzelko.com/10152021023937-icu-rating-systems. October 14 2021.
[1] J.-R. Le Gall, “The use of severity scores in the intensive care unit,” Intensive Care Med., vol. 31, no. 12, pp. 1618–1623, 2005.
[2] H. Champion, “Trauma scoring,” Scand. J. Surg., vol. 91, no. 1, pp. 12–22, 2002.