Duan et al developed a score for predicting outcome following endovascular therapy of aneurysmal subarchnoid hemorrhage. The authors are from Changhai Hospital of the Second Military Medical University in Shanghai, China.
Patient selection: endovascular therapy (coiling) of subarachnoid hemorrhage AND age >= 60 years
Exclusions: traumatic aneurysms, pseudoaneurysms, iatrogenic aneurysms
Outcome: poor outcome at 1-year (modified Rankin scale >= 3)
Parameters:
(1) age in years
(2) hypertension
(3) Hunt-Hess score
(4) Fisher scale
(5) location relative to the Circle of Willis
(6) periprocedural complication (stroke, pneumonia, heart failure, acute renal dysfunction)
Parameter |
Finding |
beta-coefficient |
Points |
age in years |
< 75 years |
0 |
0 |
|
>= 75 years |
0.511 |
2 |
hypertension |
no |
0 |
0 |
|
yes |
0.249 |
1 |
Hunt-Hess score |
1 to 3 |
0 |
0 |
|
4 or 5 |
0.564 |
2 |
Fisher scale |
1 or 2 |
0 |
0 |
|
3 or 4 |
1.172 |
5 |
location relative to Circle of Willis |
proximal |
0 |
1 |
|
at or distant to |
0.347 |
0 |
periprocedural complication |
no |
0 |
0 |
|
yes |
0.846 |
3 |
where:
• The point assignment for the location relative to the Circle of Willis is inconsistent with the beta-coefficient.
X for multivariate logistic regression model =
= SUM(beta coefficients) - 0.408
probability of poor outcome =
= 1 / (1 + EXP((-1) * X))
total score =
= SUM(points for all 7 parameters)
Interpretation:
• minimum score: 0
• maximum score 14 (the paper says 16)
• The higher the score the worse the outcome.
• Values derived from the logistic regression model do not correlate with the score. The constant may not be correct.
Score |
Severity Class |
Risk |
Percent Poor Outcome |
0 to 3 |
I |
low |
7% |
4 to 7 |
II |
moderate |
46% |
>= 8 |
III |
high |
78% |
Performance:
• The sensitivity was 82% and specificity 78%.
• The area under the ROC curve was 0.86 (which seems high for the reported S and E).
Specialty: Neurology