Classification of isolated versus multiple birth defects: An automated process for population-based registries

Area of Science:

• Epidemiology
• Medical Genetics
• Public Health

Background:

• Epidemiologic studies often distinguish between isolated and multiple birth defects.
• Classifying defects as isolated versus multiple can be challenging, especially when considering developmentally related co-occurring conditions.

Purpose of the Study:

• To develop and apply automated classification procedures for differentiating isolated from multiple birth defects, accounting for developmentally related defects.
• To improve the accuracy of birth defect analyses in large registries.

Main Methods:

• Developed automated classification algorithms to identify isolated versus multiple birth defects.
• Applied these procedures to a dataset of 235,544 nonsyndromic cases from the Texas Birth Defects Registry (1999-2018).
• Analyzed the proportion of isolated defects and identified developmentally related co-occurring defects for specific conditions.

Main Results:

• 89% of nonsyndromic cases were classified as having isolated defects.
• Significant proportions of cases with spina bifida (44%), lower limb reduction defects (44%), and holoprosencephaly (32%) had developmentally related defects.
• Proportions of isolated defects varied widely (25%-92%) across 43 specific defects.

Conclusions:

• Automated classification accounting for developmental relationships is crucial for accurate birth defect analysis.
• Findings highlight the need to consider isolated versus multiple defects and developmental relationships in risk factor association studies.
• Implications for interpreting previous birth defect research and registry methodologies.