A structural equation modeling approach to examining the predictive power of determinants of individuals' health expenditures

Jin Yuan Chern, Thomas T.H. Wan, James W. Begun

Research output: Contribution to journalArticlepeer-review

15 Scopus citations

Abstract

Understanding the determinants of health expenditures is essential for a fair and effective utilization profiling, particularly in the setting of capitation rates in risk-adjustment models. The objective of the study was to examine the relative importance of determinants in predicting future health expenditures, using structural equation modeling. Based on Andersen 's behavioral system model, individual determinants along with prior utilization and measures of health status from 1994 are evaluated in a longitudinal design for their predictive power for health expenditures in 1995. A total of 4255 policy-holders enrolled in three health plans at Trigon BlueCross/BlueShield of Virginia who responded to a mail survey were included for analysis. Person-level annual charges for health services utilization were used as the dependent variable. Five health scales were excerpted from Health Survey SF-36 to represent an individual's health status. Excluding prior utilization in 1994, health status (γ = -0.19, p < 0.001) and having diabetes (γ = 0.08, p < 0.001) are two statistically significant predictors of health expenditures in 1995. Including prior utilization, both health status (γ = -0.15, p < 0.001) and prior utilization (γ = 0.15, p < 0.001) are the most important predictors, followed by having diabetes (γ = 0.08, p < 0.001). Health status is a powerful predictor of future health expenditures, even when prior utilization is controlled.

Original languageEnglish (US)
Pages (from-to)323-336
Number of pages14
JournalJournal of Medical Systems
Volume26
Issue number4
DOIs
StatePublished - Aug 2002

Keywords

  • Health expenditures
  • Health status
  • Panel study
  • SF-36
  • Structural equation modeling

Fingerprint

Dive into the research topics of 'A structural equation modeling approach to examining the predictive power of determinants of individuals' health expenditures'. Together they form a unique fingerprint.

Cite this