Table 2.
The effect of sequential replacement of parameters for Black women in the White model on the predicted breast cancer mortality rate for Black women age 25 years and older for the period 2004–2006
| White value replaced with Black value (in bold) | ||||||||
|---|---|---|---|---|---|---|---|---|
| observed (White) | none (White model) | demography and incidence | demography, incidence, and natural history | demography, incidence, natural history, and screening | demography, incidence, natural history, and treatment | all (Black model) | observed (Black) | |
| MISCAN-Fadia | ||||||||
| Mortality per 100,000 | 36.1 | 37.5 | 32.5 | 36.9 | 38.4 | 40.3 | 41.9 | 49.8 |
| Difference, (obs-pred) | 17.4 | 12.9 | 11.5 | 9.6 | 8.0 | |||
| % explained by replaced value1 | 26% | 8% | 19% | 54% | ||||
| SPECTRUM | ||||||||
| Mortality per 100,000 | 36.1 | 37.4 | 32.2 | 40.1 | 41.3 | 42.0 | 43.2 | 49.8 |
| Difference, (obs-pred) | 17.6 | 9.8 | 8.5 | 7.8 | 6.6 | |||
| % explained by replaced value1 | 44% | 7% | 11% | 62% | ||||
1
calculated as the ratio of reduction of the difference between observed and predicted mortality rate and the difference between observed and predicted mortality, taken into account the lower incidence among Black women. So, in MISCAN-Fadia substituting Black natural history parameters into the White model explains 26% of the Black-White differences based on a reduction in the difference from 17.4 to 12.9 per 100,000, or 4.5 of the 17.4 per 100,000, i.e. 26%.
obs = observed; pred = predicted