Screening for older inpatients at risk for long length of stay: which clinical tool to use?

BACKGROUND: Screening for inpatients at risk for long length of stay (LOS) is the first step of an effective hospital care plan for older inpatients. This study aims, in older adults admitted to a geriatric acute care ward, to examine and compare the 6-item brief geriatric assessment (BGA) and the "Programme de Recherche sur l'Intégration des Services pour le Maintien de l'Autonomie" (PRISMA-7) risk levels with long LOS, and to establish their performance criteria (i.e., sensitivity, specificity, positive predictive value, negative predictive value, likelihood ratios) for LOS. METHODS: Based on an observational, retrospective, cohort design, 166 inpatients aged ≥75 admitted to a geriatric acute care ward of a McGill University-affiliated hospital (Montreal, Quebec, Canada) were recruited. The risk levels of the 6-item BGA (low, moderate and high) and the PRISMA-7 (low versus high) were calculated from a baseline assessment. The LOS was subsequently calculated in number of days. RESULTS: Only the 6-item BGA high risk level was associated with a long LOS (Odds ratio = 1.1 with P = 0.028 and Hazard ratio = 2.1 with P = 0.004). Kaplan-Meier distributions showed that there was no significant difference in the delay of hospital discharge between the low and high-risk level reported by the PRISMA-7 (P = 0.381), whereas the 6-item BGA three risk levels differed significantly (P = 0.008), with individuals at high risk levels being discharged later when compared to those with low (P = 0.001) and moderate (P = 0.019) risk levels. Both tools' performance criteria were poor (i.e., < 0.70), except for PRISMA-7's sensitivity which was 100%. CONCLUSION: The 6-item BGA risk levels were associated with LOS, low risk-level being associated with short LOS and high-risk level with long LOS, but no association was reported with the PRISMA-7 risk levels. Both tools had poor performance criteria for long LOS, suggesting that they cannot be used as prognostic tools with current scientific knowledge.