Supplementary Table 1Annual percentage of women screened by Federal State of Germany

Federal State / Start of MSP / Women aged 50-69 years / Percentage of women screened in year
In 2007 / 2005 / 2006 / 2007 / 2008 / 2009
Baden-Wurttemberg / Dec 06 / 1,295,778.0 / 0.0 / 0.0 / 5.2 / 13.5 / 21.3
Bavaria / Nov 05 / 1,533,188.0 / 0.0 / 0.5 / 9.2 / 18.0 / 18.0
Berlin / Jan 06 / 439,123.0 / 0.0 / 2.6 / 14.6 / 25.5 / 21.3
Brandenburg / Jan 08 / 355,160.0 / 0.0 / 0.0 / 0.0 / 15.0 / 25.5
Bremen / Jun 05 / 86,144.5 / 11.7 / 23.6 / 37.5 / 38.1 / 37.7
Hesse / May 06 / 763,478.5 / 0.0 / 3.6 / 11.4 / 18.8 / 25.0
Hamburg / Apr 08 / 207,556.5 / 0.0 / 0.0 / 0.0 / 12.4 / 22.7
Mecklenburg West-Pomerania / Jun 06 / 234,069.5 / 0.0 / 2.3 / 14.4 / 26.8 / 24.0
Lower Saxony / Apr 05 / 994,621.0 / 0.7 / 9.2 / 23.9 / 25.2 / 26.6
North Rhine-Westphalia / Oct 05 / 2,249,234.5 / 0.2 / 4.8 / 15.2 / 22.9 / 26.7
Rhineland-Palatinate / May 07 / 503,753.0 / 0.0 / 0.0 / 5.0 / 18.2 / 23.6
Saxony / Jul 07 / 604,931.0 / 0.0 / 0.0 / 2.3 / 21.0 / 31.6
Saxony-Anhalt / Oct 07 / 354,572.5 / 0.0 / 0.0 / 1.8 / 24.4 / 30.2
Saarland / Jan 07 / 138,881.0 / 0.0 / 0.0 / 5.0 / 18.9 / 22.1
Schleswig-Holstein / May 07 / 369,080.0 / 0.0 / 0.0 / 5.0 / 19.6 / 23.0
Thuringia / May 07 / 328,971.0 / 0.0 / 0.0 / 3.9 / 15.3 / 21.9
Germany / 10,458,542.0 / 0.2 / 2.6 / 10.4 / 20.1 / 24.2

Legend Supplementary Table 1: Midyear population in 2007; MSP: mammography screening program; data extracted from reports of the mammography screening program (26)(25)

Supplementary Table 2Surgery rates (per 100,000) for invasive plus in-situ breast cancer and estimated absolute rate changes (2009-2005)

Nationwide analysis / Model-based differences of rate change
2005 / 2009 / Rate changes
Rate / SE / Rate / SE / 95%CI / Comparison groups / Rate differences / 95%CI
Overall
0-49 years / 59.8 / 0.5 / 68.7 / 0.5 / 8.9 / 7.5; 10.3
50-69 vs. <50 years (Ref.) / n.a.
50-69 years / 337.1 / 1.8 / 468.3 / 2.1 / 131.2 / 125.7; 136.6
50-69 vs. 70+ years(Ref.) / 106.6 / 97.7; 115.6
70+ years / 318.3 / 2.2 / 342.9 / 2.2 / 24.5 / 18.4; 30.7
BCS
0-49 years / 42.6 / 0.4 / 49.7 / 0.5 / 7.1 / 5.9; 8.3
50-69 vs. <50 years (Ref.) / n.a.
50-69 years / 237.0 / 1.5 / 366.5 / 1.9 / 129.5 / 124.8; 134.2
50-69 vs. 70+ years (Ref.) / 95.9 / 88.8; 102.9
70+ years / 158.6 / 1.6 / 192.3 / 1.6 / 33.7 / 29.2; 38.1
MET
0-49 years / 17.3 / 0.3 / 19.1 / 0.3 / 1.8 / 1.1; 2.6
50-69 vs. <50 years (Ref.) / n.a.
50-69 years / 100.1 / 1.0 / 101.8 / 1.0 / 1.7 / -1.1; 4.4
50-69 vs. 70+ years (Ref.) / 10.8 / 5.3; 16.2
70+ years / 159.7 / 1.6 / 150.6 / 1.5 / -9.1 / -13.3; -4.9

Legend Supplementary Table 2

95%CI: 95% confidence interval; the estimated differences in absolute rate changes are estimated from an additive Poisson regression model with calendar year, age group, and the interaction of calendar year and age group as covariates; n.a.: not available

Supplementary Figure 1

On the necessity of an interaction term of age group and calendar year in our negative binomial models

For simplicity we only consider models with two age groups (<50, 50-69) and two calendar years.

In a negative binomial model without an interaction term of age group and calendar year, the time course of the number of hospitalizations between age groups across calendar years is restricted to be parallel (left panel, A). That is, the change of the number of hospitalizations over time is not allowed to vary between age groups.

However, our main question of interest is whether the time course of the number of hospitalizations across calendar years is different between the age groups 50-59 (the screened group) and <50 (the unscreened group). As such, we are explicitly interested whether there are non-parallel time courses (right panel, B) and this necessarily requires a model with an interaction term of age group and calendar year. As compared to the model in A, the change of the number of hospitalizations over time is now allowed to vary between age groups.