Thinking outside the curve, part II: modeling fetal-infant mortality

Area of Science:

• Epidemiology
• Biostatistics
• Public Health

Background:

• Understanding fetal-infant mortality and its link to birthweight is crucial for public health.
• Existing frameworks for analyzing birthweight distributions and mortality have limitations.
• A realistic and tractable analytical framework is needed.

Purpose of the Study:

• To introduce a novel framework for analyzing birthweight distributions and estimating birthweight-specific mortality.
• To address methodological challenges in analyzing complex birthweight data.
• To reveal previously undetectable heterogeneity in fetal-infant mortality.

Main Methods:

• Utilizing a normal mixture model to represent birthweight distributions, with the number of components determined by the data.
• Estimating birthweight-specific mortality within each mixture component.
• Developing methods for constructing confidence intervals for mortality risk and odds ratios between components and populations.

Main Results:

• Demonstrated the framework's ability to identify distinct mortality patterns within different birthweight components.
• Illustrated with an example showing significantly higher mortality odds in one component compared to another for white singleton infants.
• Proposed an extension to incorporate covariates, linking them to mixture components for subpopulation identification.

Conclusions:

• The proposed framework accommodates heterogeneous mortality curves across different birthweight groups, unlike simpler models.
• It can detect mortality variations that are missed by contaminated normal or 2-component normal mixture models.
• This approach offers a more nuanced understanding of fetal-infant mortality linked to birthweight.