Screening for post 32-week preterm birth risk: how helpful is routine perinatal data collection?

Area of Science:

• Perinatal epidemiology
• Predictive modeling in healthcare
• Maternal-fetal medicine

Background:

• Preterm birth prediction remains challenging despite existing biomarkers.
• Current biomarkers are often reserved for high-risk pregnancies.
• The utility of routinely collected antenatal data for early preterm birth risk stratification in asymptomatic women is largely unexplored.

Purpose of the Study:

• To evaluate the predictive value of the Victorian Perinatal Data Collection (VPDC) dataset for preterm birth.
• To assess the potential of VPDC data for screening women for invasive tests.
• To determine if additional factors like previous preterm birth and partner smoking improve prediction accuracy.

Main Methods:

• Logistic regression with elastic-net regularization was applied to VPDC data (2009-2013).
• Models predicted 37-week preterm birth using antenatal variables from the VPDC.
• Model performance was assessed using Brier scores, Nagelkerke's R(2), and c-statistic, with and without additional variables.

Main Results:

• The VPDC model showed reasonable discrimination (c-statistic 0.646) and good calibration for predicting preterm birth.
• At a 0.2 decision threshold, the model achieved a Positive Predictive Value (PPV) of 0.333 and a Negative Predictive Value (NPV) of 0.941.
• Inclusion of previous preterm birth and partner smoking status did not significantly enhance prediction accuracy.

Conclusions:

• Routinely collected perinatal data offers predictive information for preterm birth risk in multiparous women, comparable to some specialized data.
• The VPDC dataset contains insufficient information on antenatal complications for accurate preterm birth prediction in nulliparous women.
• Further research is needed to enhance prediction models for nulliparous populations.