Table 2. Screening-specific excess hazard ratios at five years after diagnosis for each model in turn, comparing the most deprived to the least deprived group (reference).
| Model | Variable | Form of variable in the model | Excess Hazard Ratios (95% CIs) | |||
|---|---|---|---|---|---|---|
| Non-screen-detected | Screen-detected | Non-screen-detected (n=8,962) | Screen-detected (n=8,541.5)a | |||
| 1 | = Baseline (3 df for baseline hazard) | Age (continuous) | Non-linear (2 df); time- varying effect (1df) | Linear | 1.64 (1.41-1.87) | 2.12 (1.48-2.76) |
| Year of diagnosis (continuous) | Linear | Linear | ||||
| Deprivation (categorical) | Linear | Linear | ||||
| 2 | = 1 + stage of disease at diagnosis | Extent of disease (categorical) | Linear | Linear | 1.47 (1.26-1.68) | 1.66 (1.18-2.13) |
| Tumor size (continuous) | Linear | Linear | ||||
| 3 | = 2 + tumor characteristics | Histology (categorical) | Linear | Linear | 1.45 (1.24-1.65) | 1.65 (1.18-2.12) |
| 4 | = 3 + treatment | Surgery (categorical) | Linear | Linear | 1.44 (1.25-1.64) | 1.66 (1.19-2.13) |
| Time to surgery (continuous) | Linear | Not included | ||||
| Final model | = 4 + Comorbidity | Charlson score (continuous) | Linear | Linear | 1.39 (1.20-1.59) | 1.54 (1.10-1.98) |
a
The number of women in the screen-detected group is the mean of the 10 imputed datasets and excludes women considered to be overdiagnosed.