OBJECTIVES: In-hospital falls in older patients are frequent, but the identification of patients at risk of falling is challenging. Aim of this study was to improve the identification of high-risk patients. Therefore, a simplified screening-tool was developed, validated, and compared to the STRATIFY predictive accuracy. DESIGN: Retrospective analysis of 4,735 patients; evaluation of predictive accuracy of STRATIFY and its single risk factors, as well as age, gender and psychotropic medication; splitting the dataset into a learning and a validation sample for modelling fall-risk screening and independent, temporal validation. SETTING: Geriatric clinic at an academic teaching hospital in Hamburg, Germany. PARTICIPANTS: 4,735 hospitalised patients ≥65 years. MEASUREMENTS: Sensitivity, specificity, positive and negative predictive value, Odds Ratios, Youden-Index and the rates of falls and fallers were calculated. RESULTS: There were 10.7% fallers, and the fall rate was 7.9/1,000 hospital days. In the learning sample, mental alteration (OR 2.9), fall history (OR 2.1), and insecure mobility (Barthel-Index items 'transfer' + 'walking' score = 5, 10 or 15) (OR 2.3) had the most strongest association to falls. The LUCAS Fall-Risk Screening uses these risk factors, and patients with ≥2 risk factors contributed to the high-risk group (30.9%). In the validation sample, STRATIFY SENS was 56.8, SPEC 59.6, PPV 13.5 and NPV 92.6 vs. LUCAS Fall-Risk Screening was SENS 46.0, SPEC 71.1, PPV 14.9 and NPV 92.3. CONCLUSIONS: Both the STRATIFY and the LUCAS Fall-Risk Screening showed comparable results in defining a high-risk group. Impaired mobility and cognitive status were closely associated to falls. The results do underscore the importance of functional status as essential fall-risk factor in older hospitalised patients.
OBJECTIVES: In-hospital falls in older patients are frequent, but the identification of patients at risk of falling is challenging. Aim of this study was to improve the identification of high-risk patients. Therefore, a simplified screening-tool was developed, validated, and compared to the STRATIFY predictive accuracy. DESIGN: Retrospective analysis of 4,735 patients; evaluation of predictive accuracy of STRATIFY and its single risk factors, as well as age, gender and psychotropic medication; splitting the dataset into a learning and a validation sample for modelling fall-risk screening and independent, temporal validation. SETTING: Geriatric clinic at an academic teaching hospital in Hamburg, Germany. PARTICIPANTS: 4,735 hospitalised patients ≥65 years. MEASUREMENTS: Sensitivity, specificity, positive and negative predictive value, Odds Ratios, Youden-Index and the rates of falls and fallers were calculated. RESULTS: There were 10.7% fallers, and the fall rate was 7.9/1,000 hospital days. In the learning sample, mental alteration (OR 2.9), fall history (OR 2.1), and insecure mobility (Barthel-Index items 'transfer' + 'walking' score = 5, 10 or 15) (OR 2.3) had the most strongest association to falls. The LUCAS Fall-Risk Screening uses these risk factors, and patients with ≥2 risk factors contributed to the high-risk group (30.9%). In the validation sample, STRATIFY SENS was 56.8, SPEC 59.6, PPV 13.5 and NPV 92.6 vs. LUCAS Fall-Risk Screening was SENS 46.0, SPEC 71.1, PPV 14.9 and NPV 92.3. CONCLUSIONS: Both the STRATIFY and the LUCAS Fall-Risk Screening showed comparable results in defining a high-risk group. Impaired mobility and cognitive status were closely associated to falls. The results do underscore the importance of functional status as essential fall-risk factor in older hospitalised patients.
Authors: K M Tan; B Austin; M Shaughnassy; C Higgins; M McDonald; E C Mulkerrin; S T O'Keeffe Journal: Ir J Med Sci Date: 2005 Jul-Sep Impact factor: 1.568
Authors: W von Renteln-Kruse; U Dapp; J Anders; F Pröfener; S Schmidt; C Deneke; R Fertmann; J Hasford; C Minder Journal: Z Gerontol Geriatr Date: 2011-08 Impact factor: 1.281
Authors: Wolfgang von Renteln-Kruse; Lilli Neumann; Björn Klugmann; Andreas Liebetrau; Stefan Golgert; Ulrike Dapp; Birgit Frilling Journal: Dtsch Arztebl Int Date: 2015-02-13 Impact factor: 5.594