Authors; Year; Country | Setting | Population aged group | Data Source | Study Design | Statistical Model | Derivation Cohort (% of total cohort) | Internal Validation Cohort (% of total cohort) External Validation Cohort | Falls Outcome Prediction | Falls Rate (%) | Risk Score/ Category | Number of models | Discrimination (AUC), (95% CI) | Implemented in practice |
---|---|---|---|---|---|---|---|---|---|---|---|---|---|
Volrathongchai, et al.; 2005; US [31] | Residential Care | 65–100 years | MDS | Retrospective | LBP | 9,980 (100%) | NR | Fall within 3-months | NR | No | 1 | NR | No |
Marier, et.al.; 2016; US [30] | Residential Care | NR | EMR and MDS | Retrospective | Repeated events survival model | 2,527 (49.3%) | 2,602 (50.7%) | NR | 2.3–32.3% across the deciles in validation cohort depending on the model used | Yes | 4 1:MDS Assessments 2: MDS Assessments & EMR only 3: MDS assessments & EMR duplicates 4: MDS assessments & EMR Only & EMR Duplicates | AIC 1: 6733 2: 6749 3: 6614 4:6626 | No |
Kuspinar, et al.; 2019; Canada [29] | Home care | 77 ± 14 years with no previous fall in the last 90 days | RAI-HC | Prospective | Decision tree | 88,690 (70%) | Internal: 38,013 (30%) External: 2,738 1,226 9,566 | NR | 5–35% across risk categories in derivative cohort | Yes | 1 | NR | No |
Lo, et al.; 2019; US [32] | Home care | 65 + years | OASIS and EHR | Retrospective | Random Forest Algorithm | 29,514 (50%) | 29,514 (50%) | NR | 5.14% (for emergency care or hospitalisation) | No | 3 – Validated against the MAHC-10 1: Combined 2: OASIS 3: MACH model | 1: 0.67 2: 0.67 (0.66, 0.68) 3: 0.6 (0.59,0.62) | No |