The Sepulveda GEU Study revisited: long-term outcomes, use of services, and costs.

Published

Journal Article

The randomized controlled trial of the Geriatric Evaluation Unit (GEU) at the Sepulveda Veterans Hospital was the first to document the clinical and cost-effectiveness of hospital-based comprehensive geriatric assessment (CGA). Frail elderly inpatients were assigned randomly to the GEU for CGA, therapy, rehabilitation, and placement (N = 63), or to standard hospital care (N = 60). At one year, GEU patients had much lower mortality (24% vs 48%) and were less likely to have been discharged to a nursing home (NH) (13% vs 30%), or to have spent any time in NHs (27% vs 47%). GEU patients were more likely to improve in personal self-maintenance and morale. Further, controls had substantially more acute-care hospital days, NH days, and hospital readmissions, resulting in higher direct institutional care costs, especially after survival adjustment. Here, we report the results of long-term follow-up. There was a significant survival effect through two years. Despite prolongation of life, there was no indication that quality of life was worse for survivors in the GEU group. In fact, the proportion of persons independent in > or = 2 ADLs at two years was somewhat higher for GEU patients (0.44) than controls (0.33) (z = 1.27; p = 0.056). By three years, 43% of GEU subjects and 38% of controls were still alive. Over the entire 3-year period, the per capita direct cost difference was not significant, either before or after survival adjustment (unadjusted: $37,091 GEU vs $34,205 control; survival-adjusted: $54,315 GEU vs $63,362 control; p = 0.17).(ABSTRACT TRUNCATED AT 250 WORDS)

Full Text

Cited Authors

  • Rubenstein, LZ; Josephson, KR; Harker, JO; Miller, DK; Wieland, D

Published Date

  • June 1995

Published In

Volume / Issue

  • 7 / 3

Start / End Page

  • 212 - 217

PubMed ID

  • 8547380

Pubmed Central ID

  • 8547380

International Standard Serial Number (ISSN)

  • 0394-9532

Digital Object Identifier (DOI)

  • 10.1007/bf03324318

Language

  • eng