Pannu et al developed models for predicting severe hypoxemia and mortality in patients on mechanical ventilation. The authors are from Ohio State University, Mayo Clinic, and Guang An Men Hospital (Beijing).
Patient selection: mechanical ventilation
Outcomes: severe hypoxemia, mortality, length of stay
Parameters of full model:
(1) cirrhosis
(2) hematologic malignancy
(3) aspiration
(4) estimated dead space fraction
(5) oxygenation index (OI)
(6) serum pH
(7) vasopressor use (dopamine, epinephrine, vasopressin, phenylephrine)
Parameter |
Finding |
Points |
cirrhosis |
absent |
0 |
|
present |
1 |
hematologic malignancy |
absent |
0 |
|
present |
1 |
aspiration |
absent |
0 |
|
present |
-1 |
estimated dead space fraction |
<= 0.6 |
0 |
|
> 0.6 |
1.5 |
oxygenation index |
< 18 |
0 |
|
18 to 30 |
1 |
|
> 30 |
2 |
serum pH |
> 7.35 |
0 |
|
7.20 to 7.35 |
1 |
|
< 7.2 |
3 |
vasopressor use |
no |
0 |
|
yes |
1 |
total score =
= SUM(points for all 7 parameters)
Interpretation:
• minimum score: -1
• maximum score: 9.5
• The higher the score the worse the outcome.
Score |
Mortality |
-1 |
5% |
0 |
10-15% |
1 |
25-35% |
2 |
50-55% |
3 |
65-75% |
4 |
80-90% |
5 |
90-95% |
6 |
95-100% |
7 |
100% |
Performance:
• The area under the ROC curve was 0.85 for the derivation cohort and 0.79 for the validation cohort.
In addition to the full model, a reduced model using parameters 4, 5 6 and 7 was developed. This was scored the same as the full model and ranged from 0 to 7.5.
Specialty: Critical Care