Identifying elderly patients for early discharge after hospitalization for hip fracture

TitleIdentifying elderly patients for early discharge after hospitalization for hip fracture
Publication TypeJournal Article
Year of Publication1993
AuthorsEnsberg M.D, Paletta M.J, Galecki A.T, Dacko C.L, Fries B.E
JournalJournal of Gerontology
Volume48
Issue5
PaginationM187-95
Date PublishedSep
Accession Number8366261
Keywords*Algorithms, *Geriatric Assessment/cl [Classification], *Hip Fractures/di [Diagnosis], *Length of Stay/sn [Statistics & Numerical Data], Aged, Analysis of Variance, Hip Fractures/rh [Rehabilitation], Hospitals, University/ut [Utilization], Human, Medical Records, Michigan, Middle Aged, Patient Discharge/sn [Statistics & Numerical Data], Predictive Value of Tests, Prospective Studies, Regression Analysis, Retrospective Studies, Severity of Illness Index, Support, Non-U.S. Gov't, Support, U.S. Gov't, P.H.S.
Abstract

BACKGROUND. Some elderly patients can be successfully treated in hospitals with lengths of stay (LOS) shorter than the norms developed by the diagnosis-related groups. This study was designed to test the hypothesis that elderly patients with short LOS after hip fracture have characteristics that can be identified shortly after hospital admission. METHODS. A retrospective chart review was performed of 216 patients over age 55 discharged alive from a university hospital after hip fracture. Demographic, medical, and functional data available within 48 hours of admission were used to develop an algorithm to identify patients eligible for early discharge. A prospective study of an additional 33 patients was undertaken to test this algorithm and to examine the predictive value of additional functional and psychosocial information not routinely recorded in the chart. RESULTS. Retrospective chart review identified 4 predictors of short LOS in multivariate analysis: age less than 75, admission from a nursing home, normality of admission laboratory results, and "no surgery or surgery by day three." These variables explain 25% of the total variation of LOS. In our prospective study the variable "day of surgery" had the greatest variance explanation (30.5%) in multivariate analysis. A model including day of surgery and the presence of dementia explained 42.5% of the variance of LOS. CONCLUSION. Short LOS can be predicted within 48 hours of admission utilizing data that measure severity of illness, functional status, and available support. The development of algorithms to identify patients eligible for early discharge would be beneficial to care managers.

Short TitleJ GerontolJ Gerontol
Alternate JournalJ Gerontol