Validation of the risk prediction models: calibration and discrimination
| Model | Discrimination: AUC (95% CI) | Calibration: Hosmer–Lemeshow p Value | Variance explained (RS-I) (%) | ||
|---|---|---|---|---|---|
| Internal (RS-I) | External (RS-II) | External (Chingford) | |||
| Gender, age and BMI (1) | 0.66 (0.64 to 0.69) | 0.60 (0.52 to 0.67) | 0.61 (0.56 to 0.66) | 0.19 | 6 |
| 1+ ‘questionnaire’ variables (2) | 0.66 (0.64 to 0.69) | 0.60 (0.52 to 0.67) | 0.63 (0.59 to 0.68) | 0.46 | 8 |
| 1+2+ genetic risk score (unweighted) (3) | 0.67 (0.64 to 0.70) | 0.62 (0.54 to 0.69) | 0.64 (0.59 to 0.69) | 0.79 | 8 |
| Full model (4) | 0.79 (0.77 to 0.82) | 0.86 (0.82 to 0.89) | 0.75 (0.71 to 0.80) | 0.73 | 34 |
| Minimal model (5) | 0.79 (0.77 to 0.81) | 0.86 (0.82 to 0.90) | 0.76 (0.72 to 0.80) | 0.90 | 34 |
AUC, area under the curve; BMI, body mass index; Model 1, gender, age and BMI; Model 2, gender, age, BMI and questionnaire variables; Model 3, gender, age, BMI, questionnaire variables and genetic risk score; Model 4, gender, age, BMI, questionnaire variables, genetic risk score and radiograph variables; Model 5, gender, age, BMI, knee pain and KL score 1; KL, Kellgren & Lawrence; RS-I, Rotterdam Study I; RS-II, Rotterdam Study II.