Early markers of prolonged hospital stays in older people: a prospective, multicenter study of 908 inpatients in French acute hospitals.

Objectives: To identify early markers of prolonged hospital stays in older people in acute hospitals.
Design: A prospective, multicenter study.
Setting: Nine hospitals in France.
Participants: One thousand three hundred six patients aged 75 and older were hospitalized through an emergency department (Sujet Agé Fragile: Evaluation et suivi (SAFEs)--Frail Elderly Subjects: Evaluation and follow-up).
Measurements: Data used in a logistic regression were obtained through a gerontological evaluation of inpatients, conducted in the first week of hospitalization. The center effect was considered in two models as a random and fixed effect. Two limits were used to define a prolonged hospital stay. The first was fixed at 30 days. The second was adjusted for Diagnosis Related Groups according to the French classification (f-DRG).
Results: Nine hundred eight of the 1,306 hospital stays that made up the cohort were analyzed. Two centers (n=298) were excluded because of a large volume of missing f-DRGs. Two-thirds of subjects in the cohort analyzed were women (64%), with a mean age of 84. One hundred thirty-eight stays (15%) lasted more than 30 days; 46 (5%) were prolonged beyond the f-DRG-adjusted limit. No sociodemographic variables seemed to influence the length of stay, regardless of the limit used. For the 30-day limit, only cognitive impairment (odds ratio (OR)=2.2, 95% confidence interval (CI)=1.2-4.0) was identified as a marker for prolongation. f-DRG adjustment revealed other clinical markers. Walking difficulties (OR=2.6, 95% CI=1.2-16.7), fall risk (OR=2.5, 95% CI=1.7-5.3), cognitive impairment (OR=7.1, 95% CI=2.3-49.9), and malnutrition risk (OR=2.5, 95% CI=1.7-19.6) were found to be early markers for prolonged stays, although dependence level and its evolution, estimated using the Katz activity of daily living (ADL) index, were not identified as risk factors.
Conclusion: When the generally recognized parameters of frailty are taken into account, a set of simple items (walking difficulties, risk of fall, risk of malnutrition, and cognitive impairment) enables a predictive approach to the length of stay of elderly patients hospitalized under emergency circumstances. Katz ADLs were not among the early markers identified.