Predictors of hospital length of stay and nursing home placement in an elderly medical population

OBJECTIVE: To identify predictors of adverse outcomes (nursing home placement and hospital length of stay) in a cohort of older hospitalized patients. DESIGN: Prospective cohort study. SETTING: A suburban medical center in the New York City metropolitan area. PARTICIPANTS: Patients aged 65 years and older admitted from the emergency department to medical services who could be seen within 24 hours. Patients with terminal illness, planned stays < 2 days, and admission to other than medicine were excluded. MEASUREMENTS: Sociodemographic data, mobility, Activities of Daily Living (ADL), Instrumental Activities of Daily Living (IADL), mental status, length of stay (LOS), discharge diagnosis, and disposition (nursing home vs. other). RESULTS: Of 681 patients screened, 322 were eligible and assessed at admission,206 were reassessed at discharge. Using multiple regression, greater dependency in IADL, more household help, and impaired mental status at admission predicted hospital LOS significantly (P = 0.0001). Compared with those discharged home, the nursing home (NH) group had lower scores on mean mobility and mean ADLs and IADLs and was more likely to exhibit a decline in at least one ADL. Stepwise discriminant analysis was performed using admission, discharge, and combined variables. In our final model of the predictors of discharge disposition, the use of admission functional variables, age, and sex correctly classified 100% of the NH group and 91% of the other group, with IADL, ADL, and mobility defining the function that discriminated the groups. CONCLUSIONS: Our data indicate the value of cognitive and physical function at admission as predictors of adverse outcomes and support early assessment of the elderly for discharge planning for long-term care.