Validation and inter-rater reliability of a three item falls risk screening tool

Catherine Maree Said, Leonid Churilov, Kathryn Shaw

Falls screening tools are routinely used in hospital settings and the psychometric properties of tools should be examined in the setting in which they are used. The aim of this study was to explore the concurrent and predictive validity of the Austin Health Falls Risk Screening Tool (AHFRST), compared with The Northern Hospital Modified St Thomas's Risk Assessment Tool (TNH-STRATIFY), and the inter-rater reliability of the AHFRST. A research physiotherapist used the AHFRST and TNH-STRATIFY to classify 130 participants admitted to Austin Health (five acute wards, n = 115 two subacute wards n = 15; median length of stay 6 days IQR 3-12) as 'High' or 'Low' falls risk. The AHFRST was also completed by nursing staff on patient admission. Falls data was collected from the hospital incident reporting system. Six falls occurred during the study period (fall rate of 4.6 falls per 1000 bed days). There was substantial agreement between the AHFRST and the TNH-STRATIFY (Kappa = 0.68, 95% CI 0.52-0.78). Both tools had poor predictive validity, with low specificity (AHFRST 46.0%, 95% CI 37.0-55.1; TNH-STRATIFY 34.7%, 95% CI 26.4-43.7) and positive predictive values (AHFRST 5.6%, 95% CI 1.6-13.8; TNH-STRATIFY 6.9%, 95% CI 2.6-14.4). The AHFRST showed moderate inter-rater reliability (Kappa = 0.54, 95% CI = 0.36-0.67, p < 0.001) although 18 patients did not have the AHFRST completed by nursing staff. There was an acceptable level of agreement between the 3 item AHFRST classification of falls risk and the longer, 9 item TNH-STRATIFY classification. However, both tools demonstrated limited predictive validity in the Austin Health population. The results highlight the importance of evaluating the validity of falls screening tools, and the clinical utility of these tools should be reconsidered.