Christensen et al developed a prognostic index for predicting survival in patients with cirrhosis who have a first episode of gastrointestinal bleeding and/or coma. This can help separate those patients with a relatively good prognosis from those who may require more aggressive therapy such as liver transplant. The authors are from the University of Copenhagen in Denmark.
Parameters for Cox multiple regression model:
(1) type of first episode
(2) ascites
(3) prothrombin index
(4) serum creatinine in µmol/L
|
Parameter
|
Finding
|
Points
|
|
type of first episode
|
bleeding
|
0
|
|
|
coma
|
1
|
|
|
both bleeding and coma
|
2
|
|
ascites
|
absent
|
0
|
|
|
slight
|
1
|
|
|
marked
|
2
|
|
serum creatinine
|
< 150 µmol/L
|
0
|
|
|
> 150 µmol/L
|
1
|
|
prothrombin index
|
|
|
after Table VII page 1003
where:
• The point of serum creatinine = 150 µmol/L was unassigned. In the implementation I have assigned it 0 points.
• According to Thomson, the prothrombin index in percent = (normal PT in seconds) / (patient's PT in seconds) * 100.
• The prothrombin index can be > 1.00 if the patient's PT is shorter than the control.
prognostic index (PI) =
= (0.59 * (points for first episode)) + (0.38 * (points for ascites)) - (1.53 * (prothrombin index as decimal fraction)) + (0.58 * (points for serum creatinine))
Interpretation:
• minimum index: -2.06
• maximum index: > 2.3
• The higher the PI, the worse the prognosis.
|
Prognostic Index
|
2 Year Survival Rate
|
5 Year Survival Rate
|
|
-2
|
90%
|
82%
|
|
-1
|
72%
|
60%
|
|
0
|
42%
|
23%
|
|
1
|
10%
|
2%
|
|
2
|
0%
|
0%
|
from Figure 2, page 1004.
If this data is plugged into JMP, the following line estimates can be derived.
estimated probability of 2 year survival =
= (-27 * (PI)) + 40
estimated probability of 5 year survival =
= (-27.7 * (PI)) + 27.9
