Table 3.

Time windows of excess relative risks (ERR/Gy) for occupational ionising radiation dose for risk of basal cell carcinoma in a cohort of U.S. radiologic technologists ^a, 1983-2005. All analyses have multivariate adjustment for various risk factors ^b, and are adjusted for attained age in 12 groups ^c.

Age During Exposure d
	ERR/Gy	95% CI
Age <30 e	0.59	−0.11	1.42
Age 30-39	−0.53	−2.09	1.03
Age ≥40	−1.33	−3.21	0.72
Comparison to null model using the likelihood ratio test f			P value	0.038
Years Since Exposure g
	ERR/Gy	95% CI
5-9	−0.43	−7.48	6.62
10-14	1.85	−3.32	7.02
≥15	0.14	−0.63	0.36
Comparison to null model using the likelihood ratio test f			P value	0.081
Calendar Year of Exposure h
	ERR/Gy	95% CI
Pre 1950	0.63	−0.45	1.70
1950-59	2.92	1.39	4.45
1960-69	−1.80	−2.90	−0.69
≥1970	−0.88	−1.88	0.12
Comparison to null model using the likelihood ratio test f			P value	<0.001

^aAbsorbed ionising radiation dose to the skin of the head, neck and arms Caucasians only.

^bPoisson regression with ionising radiation dose on the linear scale. Time begins at the return of the first questionnaire (1983) and ends at the end of the last follow-up (2005). Adjustments are made by five calendar periods: 1983-84, 1985-89, 1990-94, 1995-99, 2000-05. Additional adjustments by sex (stratified), education, income, cigarette smoking, alcohol consumption, body mass index, exercise, eye color, complexion, sun burn history, tanning sensitivity, dental x-rays and solar ultra violet exposure score. Confidence intervals are Wald-based (unless otherwise stated).

^c12 groups: <30, 30-35….80+ (5 year intervals)

^dBased on equation 2 in the main text.

^eConfidence intervals are profile-likelihood based.

^fComparison of the fit of the time windows model with the null model (without time windows) by the likelihood ratio test

^gBased on equation 3 in the main text.

^hBased on equation 4 in the main text.