Diagnostic accuracy of three types of fall risk methods for predicting falls in nursing homes
Journal article, Peer reviewed
MetadataShow full item record
Original versionBentzen, H., Bergland, A. & Forsen, L. (2011). Diagnostic accuracy of three types of fall risk methods for predicting falls in nursing homes. Aging Clinical and Experimental Research, 23 (3), 187-195. http://www.ncbi.nlm.nih.gov/pubmed/21993165
AIMS: To determine the diagnostic accuracy of three different methods for identifying individuals at high risk of falling. The St- Thomas Risk Assessment tool (STRATIFY- modified for nursing homes), staff judgment of fall risk, and previous falls remembered by the staff were evaluated. We also examined whether a combination of two of the methods would increase accuracy. MATERIALS AND METHODS: A prospective observational cohort study was carried out for 18 months. One thousand one hundred and forty-eight participants were included and assessed for fall risk. Falls among these residents were recorded from the date of inclusion to the date of death, transfer, or end of observation time. Diagnostic accuracy was evaluated in terms of sensitivity, specificity, predictive values and likelihood ratios, as well as Kaplan-Meier estimates and the Cox proportional hazard model, with time to the first fall as the dependent variable. Sensitivity, specificity, predictive value and likelihood ratios were calculated for falls within 30, 90 and 180 days of assessment for fall risk. RESULTS: Five hundred and seventy (49.6%) of the 1148 residents had one or more falls during the observation period. One thousand one hundred had more than 30 days of observation, 987 more than 90 days, and 867 more than 180 days. For falls within 30 days of assessment for fall risk, sensitivity varied from 65% to 72%, specificity from 69% to 75%, positive predictive value from 31% to 35% and negative predictive value from 91% to 92%. Sensitivity and negative predictive value decreased for falls within 90 days and decreased further for falls within 180 days, whereas specificity and positive predictive value increased for all three assessment methods. Staff judgment of fall risk was the single method having the highest sensitivity but the lowest specificity. A combination of either two of them increased sensitivity to more than 80%, but decreased specificity. The positive Likelihood ratio varied from 2.24 to 2.70 and the negative Likelihood ratio from 0.41 to 0.49 for falls within 30 days. The relative risk of sustaining a fall was 2.4, 2.9 and 3.0 times higher for those assessed to be at high risk of falls compared with those assessed to be at low risk, according to STRATIFY, staff judgment of fall risk and previous falls remembered by the staff, respectively. CONCLUSIONS: The diagnostic accuracy of the three methods did not differ markedly. However, staff judgment had the highest sensitivity and the lowest specificity after 30, 90 and 180 days. A combination of either two of the methods showed the highest sensitivity but the lowest specificity.