Table.

Multiplicative age cohort model with Poisson regression used to estimate mortality from breast cancer, 1990-8

	Rate ratio (90% CI)
Age group (years):
40-44	0.42 (0.41 to 0.43)
45-49	0.71 (0.70 to 0.73)
50-54	1.00 (baseline)
55-59	1.23 (91.21 to 1.25)
60-64	1.45 (1.42 to 1.47)
65-69	1.67 (1.64 to 1.70)
70-74	1.94 (1.90 to 1.98)
75-79	2.33 (2.28 to 2.38)
Cohort*:
1896	0.84 (0.82 to 0.86)
1901	0.86 (0.84 to 0.88)
1906	0.91 (0.89 to 0.92)
1911	0.94 (0.93 to 0.95)
1916	0.95 (0.94 to 0.97)
1921	1.00 (baseline)
1926	1.02 (1.00 to 1.03)
1931	1.00 (0.98 to 1.02)
1936	0.96 (0.94 to 0.98)
1941	0.93 (0.90 to 0.96)
1946†	0.90

^*Middle year of five year period of birth. 

^†Model not used for rate predictions because of statistical instability (wide confidence limits). 1946 cohort rate ratio estimated by extrapolating the 1896 to 1941 cohort rate ratios with quadratic model (rate ratios increase to 1926 and decrease thereafter). 

Model predicts mortality of 75.1 per 100 000 for women aged 50-54 born in 1921 cohort (baseline, sometimes referred to as corner or grand mean). For women aged 75-79 born in 1906 cohort, model therefore predicts rate of (2.3×0.91×75.1) per 100 000=159.3 per 100 000, etc.