Health plan enrollment and mortality in the Medicare program

Area of Science:

• Health Services Research
• Biostatistics
• Health Economics

Background:

• Previous studies indicated lower mortality for Medicare Health Maintenance Organization (HMO) enrollees compared to traditional Fee-For-Service (FFS) Medicare beneficiaries.
• However, these earlier analyses did not adequately account for potential confounding factors.

Purpose of the Study:

• To compare 2-year predicted mortality between Medicare HMO and FFS enrollees.
• To control for both observed and unobserved confounders using advanced statistical modeling.

Main Methods:

• Utilized Medicare Current Beneficiary Survey (MCBS) data.
• Employed a sample selection model to address selection bias and unobserved heterogeneity.
• Compared mortality probabilities before and after adjusting for observed and unobserved characteristics.

Main Results:

• Raw mortality was 0.5% lower in HMOs.
• After adjusting for observed factors, HMO enrollees had 0.6% higher mortality.
• After adjusting for unobserved confounders, HMO enrollees showed a statistically significant 4.2% higher mortality, suggesting adverse selection.

Conclusions:

• Unobserved confounders, potentially linked to adverse selection, significantly impact mortality estimates in the Medicare HMO sector.
• The findings challenge previous assumptions and highlight the need to identify specific confounders influencing HMO enrollee outcomes.
• The study demonstrates a robust methodological approach for comparative effectiveness research between healthcare delivery systems.