Model A | Model B | Model C | |
---|---|---|---|
Age at screening (years) | 0.029 (0.003, 0.056) 0.03 | 0.039 (0.009, 0.064) 0.01 | 0.039 (0.012, 0.068) 0.01 |
DAS28-ESR at treatment escalation | −0.426 (−0.819, −0.045) 0.03 | −0.348 (−0.764, 0.057) 0.09 | −0.229 (−0.673, 0.212) 0.31 |
DAS28-ESR 6 weeks later | −0.378 (−0.771, −0.0004) 0.05 | −0.423 (−0.756, −0.049) 0.04 | −0.432 (−0.853, −0.031) 0.04 |
DAS28-ESR 12 weeks later | −0.416 (−0.773, −0.074) 0.02 | −0.395 (−0.756, −0.049) 0.03 | −0.359 (−0.723, −0.009) 0.05 |
Intercept | 3.347 (1.264, 5.5593) 0.002 | 5.445 (2.961, 8.137) < 0.0001 | 4.881 (2.314, 7.643) 0.0003 |
Body mass index | NA | −0.090 (−0.148, −0.037) 0.001 | −0.085 (−0.143, −0.032) 0.002 |
ESR measured at treatment escalation, mm/h | NA | NA | −0.016 (−0.038, 0.004) 0.14 |
AIC used for model selection2 | 210.94 | 201.17 | 200.82 |
c statistic3 | 0.80 | 0.82 | 0.83 |
↵1 Data are shown as beta coefficients, 95% profile CI, p values.
↵2 Lower AIC values indicate a better-fitting model.
↵3 Higher c statistic indicates better model discrimination. DAS28-ESR: Disease Activity Score in 28 joints using the ESR; ESR: erythrocyte sedimentation rate; NA: not applicable; AIC: Akakie Information Criteria. Example calculation: model A, for 40-year-old with DAS at time of treatment escalation and 6 and 12 weeks later of 5.8, 4.7, and 4.2: Intercept + (40 × 0.029) + (5.8 × −0.426) + (4.7 × −0.378) + (4.2 × −0.416) = −1.474. Predicted response score = (1/(1+e1.474) = 0.19. Patient is a predicted nonresponder.