Table 3.

Effect of one additional biennial screening invitation on cumulative mortality—sensitivity and validity checks¹

Check	Included confounders	Effect of program access in 1997 versus 1995 on deaths per 100,000 (95% CI)	p value
Effect on breast cancer mortality, eligible women aged 49–63 in January 1995
Control for province fixed effects (N = 69,254)2
No	Age January 1995	114 (−40 to 268)	0.14
Yes	Age January 1995	139 (24 to 254)	0.02
Exclude women moving out of/from municipality in 1995–1997 (N = 143,424)	Age January 1995	126 (13 to 238)	0.03
“Placebo effect” on breast cancer mortality, ineligible women aged 72–77 in January 1995 (N = 33,081)	Age January 1995	−101 (−544 to 342)	0.65
Effect on all‐cause mortality, eligible women aged 49–63 in January 1995 (N = 148,920)3	Age January 1995; female population size (1989–1994)	392 (−621 to 1,406)	0.44

¹Estimates are scaled coefficients on binary indicator of program entry in 1997 rather than in 1995 from linear probability models with the dependent variable a binary indicator of having died between 1995 and 2011. All regressions include age in January 1995 in years (49 [reference category], dummies for 50, 51, to 63, except in placebo test to 72 [reference category], 73, 74, 75, 76, 77). Statistical inference accounts for clustering among women in a municipality at its entry to the program.

²Municipalities in the three provinces where screening was already fully implemented by the beginning of 1997 are dropped.

³Female population size is the number of females in a municipality averaged over the 1989–1994 period.