|
|
|
Global Registry of Acute Coronary Events (GRACE) risk score
Methodology
Granger Model for In-hospital Death
Model estimates from multiple logistic regression model
| Intercept | -7.7035 |
| AGE (per 1 yr) | 0.053 |
| PULSE (per 1 BPM) | 0.0087 |
| SYSTOLIC BLOOD PRESSURE (per 1 mmHG) | -0.0168 |
| INITIAL SERUM CREATININE | 0.1823 |
| KILLIP CLASS (1,2,3, or 4) | 0.6931 |
| ¹CARDIAC ARREST at presentation | 1.4586 |
| ¹INITIAL CARDIAC ENZYME Positive | 0.4700 |
| ¹ST SEGMENT DEVIATION | 0.8755 |
¹Enter a value of 1 if factor is present, 0 otherwise.
To obtain estimated risk of death from above estimates. Compute XB, where X=individual patient's value for each factor (eg, age=57, pulse=70…), and B=estimates above, including the intercept.
XB is then the summed product of the patient characteristics times the estimates, with the intercept added for every patient.
For example, if a patient is age 57, pulse 70, SBP 110, creatinine 1.2, Killip class III, had cardiac arrest and ST deviation but not initial positive enzymes, XB is:
XB = -7.7035 + 57 x .0531 + 70 x .0087 - 110 x .0168 + 1.2 x .1823 + 3 x .6931 + 1 x 1.4586 + 0 x .47 + 1 x .8755 = -1.28364
The probability of in-hospital death is then
P = ( Exp ^ XB ) / ( 1 + Exp ^ (XB) ) = .21693
Where exp is 2.71828…and ^ means raised to that power (XB power).
© 1998 - 2010, Center for Outcomes Research, University of Massachusetts Medical School
|
|