Description

In addition to the simple score described in the previous section, Qureshi et al developed a model based on multivariate logistic regression to identify patients at risk for in-hospital mortality following a percutaneous coronary intervention (PCI). This can help identify patients who require closer monitoring and more aggressive therapy. The authors are from William Beaumont Hospital in Royal Oak, Michigan.


 

NOTE: The documentation and spreadsheet are based on the materials submitted by Dr Ana Paula Rodel from the Hospital Universitario de Santa Maria in Brazil.

 

Parameters:

(1) serum creatinine

(2) age

(3) history of AMI

(4) extent of coronary artery disease

(5) gender

(6) serum cholesterol

(7) peripheral vascular disease

(8) diabetes mellitus

Parameter

Finding

Points

serum creatinine

<= 1.5 mg/dL (normal)

0

 

> 1.5 mg/dL (elevated)

1

age of the patient in years

<= 65 years

0

 

> 65 years

1

history of AMI

none

0

 

< 14 days

1

 

>= 14 days

0

extent of coronary artery disease

single vessel

0

 

multiple vessels

1

gender

male

0

 

female

1

serum cholesterol

normal

0

 

elevated

1

peripheral vascular disease

absent

0

 

present

1

diabetes mellitus

absent

0

 

present

1

 

where:

• Multi-vessel disease was defined as >= 2 arteries with > 50% stenosis.

 

X =

= (0.7211 * (points for serum creatinine)) - (0.6694 * (points for age)) + (0.7621 * (points for AMI)) + (0.7866 * (points for extent of coronary artery disease)) + (1.2734 * (points for gender)) + (1.1891 * (points for serum cholesterol)) + (1.1675 * (points for peripheral vascular disease)) + (0.4338 * (points for diabetes)) - 7.1912

 

probability of in-hospital mortality =

= 1 / (1 + EXP((-1) * X))

 

where:

• An age > 65 lowers the mortality rate in this model as published. In the simple model age > 65 increased risk. It is possible that age <= 65 was intended. However, if the value is reversed the mortality rates become high relative to those reported in the paper.

 


To read more or access our algorithms and calculators, please log in or register.