Fazio et al reported a model for predicting 30-day mortality following surgery for a colorectal cancer based on records at the Cleveland Clinic. The authors are from the Cleveland Clinic and St. Mary's Hospital in London.
Patient selection: surgery for colorectal carcinoma
Outcome: 30-day mortality after surgery
Parameters:
(1) age in years
(2) ASA grade
(3) stage based on TNM
(4) mode of presentation (need for surgery)
(5) cancer resection
(6) hematocrit in percent
|
Parameter |
Finding |
Beta Coefficient |
|
age in years |
< 65 years |
0 |
|
|
65 to 74 years |
1.08 |
|
|
75 to 84 years |
1.394 |
|
|
>= 85 years |
1.979 |
|
ASA grade |
I |
0 |
|
|
II |
0.939 |
|
|
III |
1.47 |
|
|
IV or V |
1.924 |
|
Stage |
I, II or III |
0 |
|
|
IV |
0.966 |
|
presentation |
urgent |
0.739 |
|
|
not-urgent |
0 |
|
cancer resection |
resected |
0 |
|
|
not resected |
1.507 |
|
hematocrit |
< 31 percent |
0.612 |
|
|
31 to 36 percent |
0.473 |
|
|
> 36% |
0 |
where:
• The timing of the hematocrit is not specified. The assumption is that it is preoperative.
• Nonresection of cancer is not specified. The assumption is that it refers to palliative surgery for unresectable cancer. Also no distinction is made between partial and complete resection.
• The number of patients in ASA 5 was not specified.
X =
= SUM(beta-coefficients for all 6 parameters) - 6.386
probability of 30-day mortality =
= 1 / (1 + EXP((-1) * X))
Performance:
• The area under the ROC curve was 0.80.
Specialty: Anesthesiology
