Table 1

Data and calculations
	time period 0			time period 1			yearly change in incident breast cancer deaths per 100,000

county	number eligible	incident breast cancer deaths	fraction screened	number eligible	incident breast cancer deaths	fraction screened	estimate	standard error

Varmland	39946	36	.03	40853	37	.73	.16	7.55
Sodermanland	49526	98	.007	53760	79	.77	-7.89	4.28
Uppsala	43340	110	.03	52426	112	.82	-5.73	4.43
Vastamanland	49827	112	.14	52329	85	.92	-8.88	3.93
Orebro	52936	133	.04	54366	108	.64	-9.74	5.36
Gavleborg	56794	311	.00	56145	219	.84	-13.40	3.64

Fraction screened is from Table 1 of [5]. Incident breast cancer deaths are from Table 3 of [5]. Number eligible is person years from Table 3 of [5] divided by years in time period from Table 2 of [5]. Calculations assume k = 0 for sensitivity analysis. Above yields Q = 3.72, σ²= 0, and overall estimate of -9 with 95% confidence (-14, -4) per 100,000.
Baker et al. BMC Medical Research Methodology 2004 4:12 doi:10.1186/1471-2288-4-12