Table 4.
Comparison of psychometric properties and diagnostic accuracy of the GFI-C with previously published results
| Results | Previous study resultsa | |
|---|---|---|
| Reliability | ||
| Internal consistency | Cronbach's α = 0.87 | Cronbach's α = 0.68 |
| Stability | ICC = 0.87, p < 0.001 (95% CI = 0.78–0.92) | r = 0.939 (p < 0.001)b |
| Validity | ||
| Face validity | 100% acceptable | 84% of older persons had no difficulty completing the GFI |
| Content validity | CVI = 0.98 | I-CVI = 0.83–1.0; S-CVI = 0.98 (S-CVI/UA = 0.66)c |
| Criterion-related validity | r = 0.76, p < 0.001 | – |
| Construct validity | ||
| 1. Known-groups method | t = 8.71, p < 0.001 (95% CI = 2.95–4.52) | Statistically significant |
| 2. Hypothesis testing | ||
| Correlation with GFI-C and AMT score | r = −0.77, p < 0.001 | The correlations for the convergent (0.45–0.61) and discriminant validity (0.08–0.50) were also as hypothesised. |
| Correlation with GFI-C and SBI score | r = − 0.67, p < 0.001 | |
| 3. Factor analysis |
χ2/df = 2.87, TLI = 0.92, CFI = 0.93, GFI = 0.92, RMR = 0.014, RMSEA = 0.073 |
χ2 = 235.02, df = 84, (p < 0.00001), GFI = 0.98, RMR = 0.0063, RMSEA = 0.074b |
| Diagnostic accuracy | ||
| Sensitivity for frailty |
Cut-off value ≥3 Sensitivity = 88.2% (95% CI: 81.8–93.0%) |
Cutoff value ≥4, Sensitivity = 66% (95% CI: 56–75%)d |
| Sensitivity for pre-frailty |
Cut-off value ≥2 Sensitivity = 71.5% (95% CI: 65.9–76.6%) |
Nil |
| Specificity for frailty |
Cut-off value ≥3 Specificity = 79.6% (95% CI: 73.5–84.9%) |
Cutoff value ≥4, Specificity = 87% (95% CI: 76–94%)d |
| Specificity for pre-frailty |
Cut-off value ≥2 Sensitivity = 84.7% (95% CI: 73.0–92.8%) |
Nil |
CI Confidence Interval, ICC Intraclass Correlation Coefficient, I-CVI, Item-level Content Validity Index, S-CVI Scale-level Content Validity Index, TLI Tucker–Lewis Index, CFI Comparative Fit Index, GFI Goodness-of-fit Index, RMR Root Mean Square, RMSEA Root Mean Square Error of Approximation, df Degree of Freedom
aPrevious study was based on Peters et al. (2012)
bThis result was based on the previous study of Luh, Yu & Yang (2018)
cThis result was based on the previous study of Xiang et al. (2019)
dThis result was based on the previous study of Baitai et al. (2013)