Using Pointwise Mutual Information for Breast Cancer Health Disparities Research With SEER-Medicare Claims

Area of Science:

• Health Services Research
• Biostatistics
• Medical Informatics

Background:

• Identifying specific medical procedures in claims data using standard coding systems (ICD, HCPCS) is complex.
• Existing methods for analyzing racial disparities in breast cancer outcomes may not fully capture treatment variations.

Purpose of the Study:

• To demonstrate the utility of Pointwise Mutual Information (PMI) for identifying associated procedure codes in medical claims data.
• To investigate racial differences in breast cancer survival using enhanced treatment definitions derived from PMI.

Main Methods:

• Utilized Surveillance Epidemiology and End Results (SEER) data linked with Medicare claims.
• Employed two approaches for treatment identification: established definitions and PMI-augmented definitions.
• Applied PMI to empirically identify relevant procedure codes for treatment classification.

Main Results:

• Initial analysis using standard definitions showed presentation differences between Black and White women narrowed the survival gap.
• Analysis using PMI-augmented definitions revealed complete elimination of survival disparities between racial groups.
• The enhanced definitions allowed for control of a broader spectrum of treatment patterns influencing survival.

Conclusions:

• PMI is a valuable tool for refining procedure code identification in medical claims research.
• Accounting for a wider range of treatment variations using PMI-driven definitions is crucial for accurately assessing racial disparities in cancer outcomes.
• This methodology offers a more comprehensive approach to understanding and potentially mitigating survival differences in breast cancer care.