Prediction of unplanned hospital admissions in older community dwellers using the 6-item Brief Geriatric Assessment: Results from REPERAGE, an observational prospective population-based cohort study.

Background: The 6-item Brief Geriatric Assessment (BGA) provides a priori risk stratification of incident hospital health adverse events, but it has not been used yet to assess the risk of unplanned hospital admission for older patients in primary care. This study aims to examine the association between the a priori risk stratification levels of the 6-item BGA performed by general practitioners (GPs) and incident unplanned hospital admissions in older community patients.

Methods: Based on an observational prospective cohort design, 668 participants (mean age 84.7 ± 3.9 years; 64.7% female) were recruited by their GPs during an index primary care visit. The 6-item BGA was completed at baseline and provided an a priori risk stratification in three levels (low, moderate, high). Incident unplanned hospital admissions were recorded during a 6-month follow-up.

Results: The incidence of unplanned hospital admissions increased with the risk level of the 6-item BGA stratification, the highest prevalence (35.3%) being reported with the high-risk level (P = 0.001). The risk of unplanned hospital admission at the high-risk level was significant (crude odds ratio (OR) = 5.48, P = 0.001 and fully adjusted OR = 3.71, P = 0.032, crude hazard ratio (HR) = 4.20; P = 0.002 and fully adjusted HR = 2.81; P = 0.035). The Kaplan-Meier's distributions of incident unplanned hospital admissions differed significantly between the three risk levels (P-value = 0.002). Participants with a high-risk level were more frequently admitted to hospital than those at a low-risk level (P = 0.001). Criteria performances of all risk levels were poor, except the specificity of the high-risk level, which was 98.2%.

Conclusions: The a priori 6-item BGA risk stratification was significantly associated with incident unplanned hospital admissions in primary care older patients. However, except for the specificity of the high-risk level, its criteria performances were poor, suggesting that this tool is unsuitable for screening older patients in primary care settings at risk of unplanned hospital admission.