Sievert et al developed a model evaluating women undergoing medically indicated early preterm induction of labor. This can help to predict how the woman will deliver. The authors are from the University of Alabama at Birmingham.
Patient selection: medically indicated early preterm induction of labor
Outcome: vaginal or cesarean delivery
Parameters:
(1) gestational age at delivery in weeks (23 to 34 weeks)
(2) modified Bishop score of Laughon et al (from 0 to 9)
(3) fetal weight as percentile (as indicator for intrauterine growth retardation, IUGR)
(4) chronic hypertension
(5) body mass index
Parameter
|
Finding
|
Points
|
fetal weight
|
>= 10th percentile
|
0
|
|
< 10th percentile (IUGR)
|
1
|
chronic hypertension
|
absent
|
0
|
|
present
|
1
|
X =
= (-0.2451882 * (gestational age in weeks)) - (0.3510045 * (modified Bishop score)) + (1.204289 * (points for fetal weight)) + (0.7868689 * (points for hypertension)) + (0.442397 * (body mass index)) - 6.757939
where:
• In Figure 2 the intercept is given as 6.757939. Usually in a logistic regression model the value is negative. Various permutations were tried, with the formula above the only one giving reasonable results.
probability of cesarean section =
= 1 / (1 + EXP((-1) * X)
Interpretation:
• A probability < 50% was associated with vaginal delivery.
• A probability >= 50% was associated with a cesarean section.
Performance:
• The area under the ROC curve was 0.77 in the validation set.