|| Checking for direct PDF access through Ovid
To compare and improve the diagnostic ability of standard fall risk scores in a healthcare facility with diverse patient background. Medical records of 200 adult inpatients who fell (case) and the other 200 randomly selected inpatients who were admitted on the same day (control) at Bumrungrad International Hospital (BIH) during 2014–2016 were reviewed. All data required for calculating the Hendrich II Fall Risk Model (H, 7 items), STRATIFY Risk Assessment Tool (S, 5 items), Morse Fall Scale (M, 6 items), and Johns Hopkins Hospital Fall Risk Assessment Tool (J, 7 items) were extracted. Eight non-clinical determinants proposed by the Fall Risk Committee were also analyzed (B1-B8). The diagnostic ability of the standard scores were assessed using Area under the Receiver Operating Characteristic (AUC) analysis.The overall mean age was 54.22 years, female 45.25%, Asian 47.00%. The cases were older (58.88 vs 49.57; p<0.001) and male (61.64% vs 35.91%; p<0.001). More Middle Eastern patients (67.48%) fell than other ethnic origins (Caucasian 54.43%, Asian 36.17%; p<0.001).Five B determinants were significantly associated with fall event: admission on the arrival day vs within 6 days after arrival (B1: 100% vs 39.24%; p<0.001), anticipated surgery with sedation (B2: 28.66% vs 63.79%; p<0.001), first admission (B3: 37.13% vs 68.71%; p<0.001), personal caregiver dependency (B4: 48.40% vs 75.00%; p=0.012), and interpreter need (B5: 65.57% vs 43.17%; p<0.001). The AUC of the scores were: H 84.97%, M 77.91%, J 57.29%, and S 49.26% whereas the AUC of the B sub-scores were: B1 75.78%, B2 33.25%, B3 34.75%, B4 47.00% and B5 59.50%. Adding B1 and B5 to H improved the AUC from 84.97% to 91.47% and 85.29%, respectively.Diagnostic ability of fall risk scores is context-specific and could be improved by adding contextual determinants.