Table 1 Model updating methods used in the analysis
Model | Explanation | Reason for updating |
|---|---|---|
1 | No change — original model | -- |
2a — White cohort 2b — South Asian cohort | Adjustment of the baseline risk/hazard | Miscalibration in the large; overall O/E ratio is not close to 1 |
3a — White cohort 3b — South Asian cohort | Method 1 + adjusting the magnitude of the linear predictor | Calibration slope not close to 1 — the regression coefficients are over/under-fitted |
4 | Method 2 + addition of ethnicity as a predictor | A new factor is found to be important to the model |
5 | Re-estimation of all regression coefficients + addition of ethnicity as a predictor | The weighting of the coefficients in the original model needed to be adjusted after inclusion of a new predictor |
6 | Method 4 + accounting for competing risks (Fine and Gray model) | Other events (which prevent the primary event from occurring) |