Development of a method to identify seniors at high risk for high hospital utilization

Med Care. 2002 Sep;40(9):782-93. doi: 10.1097/00005650-200209000-00008.


Background: A small percentage of older persons account for most Medicare costs. If persons at high risk for high health care utilization can be identified, resources can be directed to improve their health care and reduce utilization.

Objective: To develop an efficient and economical approach to identifying older persons at risk for high future health care utilization.

Design: Validation cohort.

Setting: Three communities.

Subjects: Five thousand one hundred thirty-eight community-dwelling persons aged 71 years or older.

Main outcome measures: High utilization (defined as >or=11 hospital days during 3 years) and overall Part A Medicare hospital costs during 3 years.

Results: Predictive multivariable models were created that relied on prior hospitalization only, self-report only, and combined self-report and physical examination/lab data. Ten self-report items (hospitalizations in prior year and year before that, male gender, fair/poor health, not working, infrequent religious participation, needing help bathing, unable to walk 1/2 mile, diabetes, and taking loop diuretics) and two lab tests (low serum albumin and iron) remained as independent predictors of high utilization. Based upon these variables, approximately 1/4 of the population was identified as being at high risk (>or=0.28 probability) for high health care utilization and those identified accounted for approximately half of all Medicare Part A costs for the entire population. Finally, a two-phase strategy was developed in which tests are only administered to individuals whose risk cannot be adequately determined by self-report variables (approximately 1/4 of subjects).

Conclusions: Simple questions and laboratory tests can accurately and efficiently identify seniors at high risk for high health care utilization.

Publication types

  • Research Support, U.S. Gov't, P.H.S.

MeSH terms

  • Aged
  • Cohort Studies
  • Female
  • Geriatric Assessment*
  • Hospitalization / statistics & numerical data*
  • Humans
  • Logistic Models
  • Male
  • Risk Assessment*
  • United States