Table 4:
Predictive models of Hip Fracture using difference approach for variable selection. Result for each variable is Coefficient (SE).
| BIC | Backward | BMA | |
|---|---|---|---|
| Intercept | −5.054 (0.536) | −5.143 (0.701) | −5.571 (0.485) |
| Age Enrollment, per year | 0.034 (0.008) | 0.041 (0.009) | 0.040 (0.008) |
| Male Gender | -- | -- | -- |
| Caucasian | -- | −0.350 (0.169) | −0.615 (0.158) |
| BMI, per Kg/m2 | -- | -- | -- |
| Current Smoker | 0.323 (0.126) | 0.338 (0.158) | -- |
| Oral steroid | -- | -- | -- |
| Osteoporosis | 0.489 (0.151) | 0.667 (0.180) | 0.646 (0.178) |
| Prevalent Hip Fracture | 1.559 (0.220) | 1.483 (0.278) | 1.563 (0.274) |
| Prevalent Compression Fracture | -- | -- | -- |
| Rheumatoid arthritis | -- | 0.489 (0.206) | -- |
| Diabetes | -- | 0.393 (0.178) | -- |
| FEV1 % | -- | -- | -- |
| GOLD | -- | -- | -- |
| Emphysema ≥ 5% | -- | -- | -- |
| Inhaled steroids | -- | -- | -- |
| BODE | -- | −0.093 (0.044) | -- |
| 6-MW distance (m) | −0.002 (0.0005) | −0.002 (0.001) | -- |
| MMRC | -- | -- | -- |
| Total exacerbations | 0.107 (0.015) | 0.114 (0.020) | 0.113 (0.019) |
| Bootstrap AUC (95%CI) (higher better) | 0.67 (0.63–0.69) | 0.68 (0.65–0.71) | 0.67 (0.63–0.70) |
| P value of H-L test for goodness of fit (higher better) | 0.70 | 0.03 | 0.49 |