Skip to main content
. 2022 Apr 1;22:271. doi: 10.1186/s12877-022-02973-0

Table 2.

The performance of baseline PH-FRAT against the actual fall occurrence within the first six months of PH-FRAT assessment

Risk score cut-off 14 Risk score cut-off 10
PH-FRAT Predicted Falls Observed Falls No. of
falls
Resident
days
Observed Falls No. of
falls
Resident
days
Fallers Non-fallers Fallers Non-fallers
Fallers 627 388 2512 147,454 1626 1559 5570 446,080
Non-fallers 1782 3091 4975 673,298 783 1920 1917 374,672
Total 2409 3479 7487 820,752 2409 3479 7487 820,752
Standard method
 AUROC (95% CI) 0.57 (0.56–0.59) 0.61 (0.60–0.63)
 Sensitivity (95% CI) 26.0 (24.3–27.8) 67.5 (65.6–69.4)
 Specificity (95% CI) 88.8 (87.8–89.9) 55.2 (53.5–56.9)
 PPV (95% CI) 61.8 (58.7–64.8) 51.1 (49.3–52.8)
 NPV (95% CI) 63.4 (62.1–64.8) 71.0 (69.3–72.7)
 Youden’s index 0.148 0.227
Event rate method
 SensitivityER (95% CI) 33.6 (30.3–36.8) 74.4 (71.9–76.9)
 SpecificityER (95% CI) 82.0 (80.9–83.1) 45.6 (44.2–47.1)
 Youden’s indexER 0.156 0.201

AUROC Area Under Receiver Operating Characteristics curve, PPV Positive Predictive Value, NPV Negative Predictive Value. AUROC interpretation: poor (0.5–0.6), sufficient (0.6–0.7), good (0.7–0.8), very good (0.8–0.9) and excellent discrimination (0.9–1.0). Youden’s index ranges from 0 (no discrimination) to 1 (perfect discrimination)