BACKGROUND AND PURPOSE:

• Over the past 20 years in the US, maternal morbidity has increased by approximately 200%
• Rossi et al. (Obstetrics & Gynecology, 2019) evaluated risk factors associated with maternal ICU admission and incorporated these features into a multivariable predictive model

METHODS:

• Population-based cohort study (2012 to 2016)
• Participants
  • All live births | Gestational age 20w0d to 44w6d
• Risk factors
  • Maternal demographics (e.g., age, race, prenatal care etc.)
  • Co-morbidities
  • Pregnancy specific conditions (e.g., GDM, preeclampsia etc.)
  • Other Ob characteristics (e.g., previous C/S, route of delivery etc.)
• Data analysis
  • Predictive model: Internal validation followed by external validation on a separate live birth cohort (2006–2011; n=856,255)
• Primary outcome
  • Prenatal factors associated with increased risk of maternal ICU admission

RESULTS:

• 18,745,615 live births | 0.15% (27,602) of mothers admitted to the NICU
• Most significant risk factors included
  • Scheduled cesarean | Gestational age (decreasing) | Preeclampsia | Induction of labor | Chronic hypertension | Maternal Age | Race
  • BMI was not associated with ICU admission as a continuous variable | BMI ≥50 was ‘moderately’ associated with ICU admission
• Fourteen variables were selected for inclusion in the predictive model for maternal ICU admission
  • Area under the curve (AUC) of 0.81 (95% CI 0.79 to 0.81)
  • Indicated women who were at 30-fold risk for ICU admission
• Predicted minimal and maximal risk for ICU admission ranged from 0–25%
• External validation matched up with internal results
  • AUC of 0.83 (95% CI, 0.82 to 0.84)
• Adjusting up to high cut point of ≥5.0% risk for ICU admission only achieved a positive predictive value (PPV) of 4.0%

CONCLUSION:

• The predictive model can help to better understand effect of risk factors on ICU admission risk
• However, PPV was low due to underlying low a priori risk and therefore the authors caution against the use of their model
• It is possible that the algorithm can be clinically useful in a high risk setting, which would improve PPVs

Learn More – Primary Sources:

Predictive Model of Factors Associated With Maternal Intensive Care Unit Admission

Related ObG Topics: