Does the Medicare Principal Inpatient Diagnostic Cost Group Model Adequately Adjust for Selection Bias?

by Hongjun Kan

Download eBook for Free

FormatFile SizeNotes
PDF file 0.6 MB

Use Adobe Acrobat Reader version 10 or higher for the best experience.

Objective: To determine whether the principal inpatient diagnostic cost group (PIP-DCG) model adequately adjusts for HMO (health maintenance organization) favorable selection, and, if not, to quantify the model's bias in predicting resource use of HMO enrollees that is due to unobserved selection. Data Sources/Study Setting: Data on hospital use of the elderly Medicare population in the state of California from 1994-1996. Personal characteristics and hospital discharge records of HMO and FFS beneficiaries were obtained from the Centers for Medicare and Medicaid Services and the California Office of Statewide Health Planning and Development. Study Design: A simultaneous equations model of HMO enrollment and subsequent hospital use was used to test whether the PIP-DCG risk factors adequately adjust for selection bias. Simulations were conducted to quantify the magnitude of bias in a naive PIP-DCG model developed on FFS (fee-for-service) data ignoring unobserved selection. Principal Findings: (1) The PIP-DCG model does not fully control for selection bias. Unobserved selection is favorable in the HMO population and adverse in the FFS population even after controlling for the PIP-DCG risk factors. (2) A naive PIP-DCG model developed on FFS data ignoring unobserved selection predicts 28 percent more hospital days for new HMO enrollees than if they had remained in FFS. Conclusions: The PIP-DCG model continues to overpay Medicare + Choice plans. To reduce excess payments, a more comprehensive model that better captures underlying health status is needed.

Table of Contents

  • Chapter One


  • Chapter Two

    Data, Sample Selection, and Descriptive Analyses

  • Chapter Three


  • Chapter Four

    Estimation Results

  • Chapter Five


  • Chapter Six

    Policy Implications

  • Chapter Seven

    Conclusions and Limitations

  • Appendix A

    STATA Code for Estimating the Simultaneous Equations Model

Research conducted by

This report is part of the RAND Corporation Dissertation series. Pardee RAND dissertations are produced by graduate fellows of the Pardee RAND Graduate School, the world's leading producer of Ph.D.'s in policy analysis. The dissertations are supervised, reviewed, and approved by a Pardee RAND faculty committee overseeing each dissertation.

This document and trademark(s) contained herein are protected by law. This representation of RAND intellectual property is provided for noncommercial use only. Unauthorized posting of this publication online is prohibited; linking directly to this product page is encouraged. Permission is required from RAND to reproduce, or reuse in another form, any of its research documents for commercial purposes. For information on reprint and reuse permissions, please visit

The RAND Corporation is a nonprofit institution that helps improve policy and decisionmaking through research and analysis. RAND's publications do not necessarily reflect the opinions of its research clients and sponsors.