Development and Validation of an Algorithm to Identify Prenatal Care in Administrative Data: Predictive Validity for...

Area of Science:

• Health Services Research
• Maternal and Child Health
• Health Informatics

Background:

• Prenatal care (PNC) is crucial for positive birth outcomes.
• Accurate identification of PNC utilization and continuity of care from claims data is challenging.
• Existing methods may not fully capture the nuances of PNC provider relationships.

Purpose of the Study:

• To develop and validate a hierarchical algorithm for assigning PNC encounters using claims data.
• To ensure the algorithm accurately reflects continuity of care.
• To assess the predictive validity of the algorithm by linking provider continuity to birth outcomes.

Main Methods:

• Retrospective cohort study of South Carolina Medicaid beneficiaries (2016-2021).
• Developed a six-step hierarchical algorithm using specialty, diagnostic/procedure codes, and adjustments for inpatient/supplemental visits.
• Assessed predictive validity by examining associations between predominant provider status and adverse birth outcomes using generalized estimating equations and cross-validation.

Main Results:

• The algorithm successfully identified predominant providers for 98% of pregnancies with at least two PNC encounters.
• Predominant provider status was significantly associated with reduced risk of preterm birth (aRR: 0.68) and low birth weight (aRR: 0.68).
• The algorithm demonstrated good model fit and performance through 10-fold cross-validation.

Conclusions:

• A validated claims-based algorithm can effectively identify PNC utilization and continuity.
• Continuity of care, identified by predominant provider status, is associated with improved perinatal health outcomes.
• This algorithm provides a foundation for interventions aimed at optimizing PNC delivery and improving birth outcomes.