Reanalysis of cancer mortality in Japanese A-bomb survivors exposed to low doses of radiation: bootstrap and simulation methods

Table 7 Optimal (integral) latency, LRT and ERR, including "city", by dose range

Dose range	Site	Φ_m	LRT_2p-con at Φ_m	LRT_2p-lin at Φ_m	ERR₁ at Φ_m	Latency Range	ERR_0.025 at Φ_m	Latency Range
0-20	Stom	11	13.02 p = 0.007	12.27 p = 0.003	0.38 (0.11, 0.70) (0.16, 0.65)*	5-18 5-21* 9-21^nl	0.34 (0.11, 0.63) (0.14, 0.59)*	5-21 5-21*
5-500	Stom	8	9.03 p = 0.025	5.74 p = 0.07	0.29 (0.06, 0.60) (0.09, 0.55)*	7-9 5-11*
0-20	Liver	37	28.41 p < 10^-7	18.96 p < 10^-4	1.26 (0.61, 2.14) (0.69, 1.97)*	29-43 29-43* 29-41^nl	0.68 (0.18, 1.25) (0.24, 1.20)*	29-44 29-44*
5-500	Liver	37	14.47 p = 0.003	10.55 p = 0.005	0.74 (0.25, 1.43) (0.32, 1.30)*	34-41 34-42* 34-41^nl
0-20	Lung	16	10.43 p = 0.03	7.13 p = 0.05	1.00 (0.36, 1.84) (0.42, 1.72)*	10-21 10-21* 12-16^nl	0.80 (0.21, 1.66) (0.28, 1.53)*	10-16 10-21*
5-500	Lung	16	21.69 p < 10^-4	19.63 p < 10^-5	1.33 (0.67, 2.22) (0.73, 2.10)*	7-21 7-21* 12-21^nl
0-20	Panc	10	14.89 p = 0.002	13.7 p < 0.001	11.32 (5.35, 16.46) (5.44, 15.60)*	5-11 5-11* 7-11^nl	10.77 (4.21, 16.05) (4.70, 15.31)*	5-11 5-11*
5-500	Panc	42	1.97 p = 0.64	1.96 p = 0.38	0.44 (-, -)	-
0-20	Leuk	25	21.66 p < 10^-4	21.66 p < 10^-5	1.39 (0.09, 4.31) (0.26, 3.79)*	24-25 24-26* 17-36^nl	0.89 (0.11, 2.22) (0.16, 2.16)*	19-33 17-33*
5-500	Leuk	28	14.76 p < 0.001	8.60 p = 0.013	1.83 (0.45, 3.92) (0.59, 3.66)*	24-31 24-32* 24-28^nl

Dose range: data on which the two-phase model is fitted, including "city" as a control covariate
Φ_m: latency, restricted to integers 5, 6,... 44, which maximises LRT_2p-con subject to ERR₁ ≥ 0
LRT_2p-con: Likelihood Ratio Test comparison of two-phase and control models at Φ_m. p values derived from Table 8
LRT_2p-lin: Likelihood Ratio Test comparison of two-phase and linear models at Φ_m. p values derived from Table 8
ERR₁: Excess Relative Risk at 10 mSv, from the two-phase model fitted across the dose range with latency Φ_m, and its BC_a 95%CI (5000 replications)
Latency range: integral latency range around Φ_m for which BC_a 95%CI is strictly positive
*: BC_a 90%CIs (5000 replications)
^nl: integral latency range around Φ_m for which LRT_2p-lin > 6
ERR_0.025: as above, for 0.25 mSv (computed only when Dose range = 0-20 mSv)

ISSN: 1476-069X