Abstract: Objective: To systematically review the falls risk tools for hospitalized elderly patients. Method: The English electronic databases Embase, PsycINFO, CINAHL Complete, ScienceDirect, Wiley Online Library, Web of science, and the Chinese databases CNKI and WANFANG were searched. Two reviewers independently screened the literatures, extracted data, and assessed methodological quality of all included studies. MetaDiSc 1.4 was used for Meta-analysis. Results: Fourteen studies were identified, eight studies were about STRATIFY (St Thomas's Risk Assessment Tool in Falling Elderly Inpatients), and three studies were combined. The pooled sensitivity was 0.81 (95% CI : 0.77-0.85), heterogeneity was 93.5% (χ2=30.8, P<0.001), pooled specificity was 0.65 (95% CI : 0.63-0.67), and heterogeneity was 98.0% (χ2=98.73, P<0.001). There were 5 studies about HIIFRM (The Hendrich Fall Risk Model II), and three of them were combined. The pooled sensitivity was 0.84 (95% CI?: 0.77-0.89), heterogeneity was 70.3% (χ2=10.10, P=0.018), pooled specificity was 0.44 (95% CI?: 0.42-0.47), and heterogeneity was 95.2% (χ2=62.68, P<0.001). There was one study about CS (Conley Scale), the sensitivity was 0.77, and specificity was 0.49. There was one study about VIC-FRAT (Victoria Falls Risk Assessment), the sensitivity was 0.76, and specificity was 0.60. There was one study about MFS (Morse Fall Scale), the sensitivity was 0.32, and specificity was 0.82. Conclusions: The specificity of STRATIFY is better than HIIFRM, and the sensitivity of HIIFRM is better than STRATIFY in assessing the fall risk of hospitalized elderly patients, and STRATIFY is suitable for more clinical departments.

Key words: aged; inpatients; accidental fall; risk assessment; systematic review