Predicting a long hospital stay after admission to a geriatric assessment unit: Results from an observational retrospective cohort study

Objective: Morbidities and related disabilities often lead to older inpatients having a long hospital stay. The aim of this study was to examine whether the 6-item brief geriatric assessment (BGA), developed and validated in France to determine a priori levels of risk of a long hospital stay (i.e.; low, moderate, high), could be successfully used with patients admitted to a geriatric assessment unit (GAU) in Quebec. Study design: Observational retrospective cohort design. Setting: A GAU of a McGill University affiliated hospital (Montreal, Quebec, Canada). Participants: 499 inpatients (84.7 +/- 7.2 years; 73.3% female) recruited upon their admission. Main outcome measures: The BGA comprises 6 items: age >= 85 years, male gender, 5 drugs per day, use of home-help support, history of falls and temporal disorientation. It was administered at baseline and a priori levels of risk of a long hospital stay (i.e., low, moderate, high) were determined. Length of hospital stay (LHS, in days) was calculated using the hospital registry. The association between a priori levels of risk from the BGA and LSH was examined using regression models and Kaplan-Meier curves. Results: The LHS increased with the 6-item BGA a priori level of risk (P = 0.010). High-risk (Hazard ratio (HR) = 1.68 with P < 0.001) and moderate-risk (HR = 1.24 with P = 0.039) of a long hospital stay successfully predicted a long stay. Kaplan-Meier distributions of time to discharge showed that inpatients classified as having high and moderate risk levels for a long hospital stay were discharged later than those with a low risk level (P < 0.001 and P = 0.013). Conclusion: The 6-item BGA a priori levels of risk for a long hospital stay successfully predicted a long stay among patients admitted to a GAU in Quebec.