Pregnancy Cohort Algorithm for MarketScan Claims Data

This repository contains SAS code to algorithmically identify a pregnancy cohort from MarketScan commercial claims data. The process involves extracting inpatient, outpatient, and medication fill claims using predefined code lists for prenatal care, pregnancy outcomes, and gestational age. The final output is a derived pregnancy cohort that may require cleaning for data anomalies.

Use Cases

• Implement the provided SAS algorithm to extract pregnancy episodes from MarketScan claims data using inpatient admission and outpatient service claims.
• Clean the derived pregnancy cohort by applying logic to handle pregnancies with unreasonable indexing prenatal claims or overlapping gestation periods.
• Utilize the Variable Identification.xlsx code lists for prenatal claims and pregnancy outcome codes to customize cohort identification criteria.
• Analyze medication fill claims data extracted by the algorithm to study prescription patterns during identified pregnancy episodes.

Strengths

• Algorithm is documented in a detailed user guide (Algorithm Documentation - User Guide.docx) explaining program files, inputs, and outputs.
• SAS program files are extensively commented to facilitate user understanding and implementation.
• Includes a specific cleaning guide (Cleaning Derived Pregnancies.docx) with logic and references for refining the pregnancy cohort.

Limitations

• The dataset size, number of rows, and specific columns are unknown from the provided input.
• The algorithm's performance and cohort accuracy depend on the underlying MarketScan data quality and completeness.
• Implementation requires a SAS environment and familiarity with connecting to remote servers, as indicated in the code.

Provenance

Source

MarketScan commercial claims data.

Collection Method

Algorithmic extraction using SAS code on a remote server, pulling inpatient, outpatient, and medication fill claims.

Time Range

null

Freshness

null

Geography

null