Classification and Regression Tree (CART) Analysis of DeVita et al for Predicting Death Within 60 Minutes After Discontinuing Life-Sustaining Therapy in an Organ Donor in the ICU
DeVita et al developed a classification and regression tree (CART) model for predicting death within 60 minutes after discontinuing life support of an organ donor in the intensive care unit (ICU). The authors are from the University of Pittsburgh, UPMC Presbyterian Shadyside Hospital, Children’s Hospital of Pittsburgh, Case Western Reserve University and the University of Cincinnati.
Patient selection: donor in the ICU on life support
Parameters:
(1) Glasgow Coma Score (GCS)
(2) SaO2 to FIO2 ratio (both as decimal fractions or percents)
(3) peak inspiratory pressure (PIP) in cm H2O
Glasgow Coma Score
SaO2 to FIO2 ratio
PIP in cm H2O
Death <= 60 Minutes
3
NA
NA
63%
>= 4
< 2.3
< 35
32%
>= 4
< 2.3
>= 35
67%
>= 4
>= 2.3
NA
15%
Additional findings favoring early death:
(1) use of vasopressors at > 0.2 micrograms per kg per minute
(2) respiratory rate < 8 breaths per minute off mechanical ventilation
To read more or access our algorithms and calculators, please log in or register.
