Table 2 Multivariable modeling investigating factors affecting probability of prenatal consult for pregnant people in the analysis cohorts.

From: Inequities in prenatal neonatology consultation in high-mortality neonatal populations

Model

Effect

Contrast

Odds ratio

p value

Periviable cohort (N = 172)a

 1

Race of pregnant person

White vs. non-White

–

0.0092

Time from admission to delivery (h)

 

–

0.0096

Time from admission to delivery by race of pregnant person interaction

White vs. Non-White | Time from admission to deliveryb = 1 h

6.88 (1.38, 34.20)

0.0403

White vs. Non-White | Time from admission to deliveryb = 3 h

2.44 (0.51, 11.76)

Insurance

Commercial vs. Medicaid/Self-pay

4.31 (0.91, 20.36)

0.0653

Congenital anomaly cohort (N = 197)c

 1

Race of pregnant person

White vs. Non-White

1.96 (0.90, 4.28)

0.0911

  1. aRegular logistic regression was used for the periviable cohort model because the variance estimation for random site effect was zero. Language was not considered due to very small sample size of pregnant people who spoke a language other than English.
  2. bTime from admission to delivery was used as a continuous variable in the model. The two values, 1 h and 3 h, were chosen to represent the early and later time in the 4-h period when drastic changes were seen in the probability of receiving consult between White and non-White pregnant people (Fig. 3).
  3. cMixed-effect logistic regression was used for the congenital cohort model. Time from admission to delivery was not considered to be clinically relevant in this cohort. Language was not able be tested as effect modifier since there was not enough data when stratified by the outcome and race. Prenatal care and insurance were dropped from the model due to their insignificance.
Back to article page