 | Related Articles |
Prediction of in-hospital mortality with the 6-item Brief Geriatric Assessment tool: An observational prospective cohort study.
Maturitas. 2018 Apr;110:57-61
Authors: Launay CP, Rivière H, Chabot J, Beauchet O
Abstract
BACKGROUND: The 6-item Brief Geriatric Assessment (BGA) is a screening tool to identify frail inpatients who are at risk of adverse health events. Its predictive value for in-hospital mortality has not been examined yet.
OBJECTIVE: This study examined whether the BGA is able to predict in-hospital mortality in older patients.
METHODS: A total of 1082 participants were included in this observational prospective cohort study. At their admission to the medical wards of Angers University Hospital (France), all inpatients aged ≥65 years were screened with the BGA. Its 6 items are: age ≥85 years, male gender, polypharmacy (i.e., ≥5 drugs per day), non-use of home-help services, history of falls in the previous 6 months, and temporal disorientation (i.e., inability to give the month and/or year). Three levels (low, intermediate and high) of risk of adverse health events had previously been identified, based on different combinations of BGA items. Patients were separated into 2 groups using the occurrence of in-hospital death. The length of stay was calculated as the number of days in hospital using the hospital registry. The use of psychoactive drugs and the reason for admission were used as covariates.
RESULTS: Older inpatients who died were more frequently admitted for an acute organ failure (P < 0.001). Cox regression models showed that a priori intermediate risk (HR = 1.89, P < .001) and high risk (HR = 2.34, P < .001) risk levels predicted in-hospital mortality. Kaplan-Meier survival curves confirmed that inpatients at high risk (P = .047) and those at intermediate risk (P = .013) died earlier than patients at low risk.
CONCLUSIONS: Combinations of items on the BGA successfully predicted the risk of in-hospital mortality in this sample of older inpatients.
PMID: 29563036 [PubMed - indexed for MEDLINE]