Skip to main content
. Author manuscript; available in PMC: 2022 May 19.
Published in final edited form as: Environ Int. 2021 Dec 24;159:106983. doi: 10.1016/j.envint.2021.106983

Table 6.

Restricted maximum likelihood (REML) and DerSimonian and Laird 1-step random effects model fits to various subsets (as given by Supplementary Table S2).a

Endpoint Excess relative risk (ERR) / Gy (95% CI) p-value Residual heterogeneity p-value I2 (%)
Analysis using lower dose risk estimatesb
Postnatal exposure
Leukaemia 5.20 (− 1.13,11.53)c 0.108c 0.027c 44.13c
Lymphoma (including CLL) 1.21 (−11.96,14.38) 0.857 0.899 0.00
Brain/CNS 6.81 (0.58,13.04) 0.032 <0.001 78.57
Lung 2.23 (−5.55,10.02) 0.574 0.069 69.87
Thyroid (including nodules) 0.00 (−0.07,0.08)c 0.928c <0.001c 65.64c
All solid except brain/CNS, lung, thyroid 0.00 (−0.41,0.42) 0.983 0.189 0.00
All six endpoints onlyd 0.23 (0.07,0.39)c 0.006c <0.001c 72.61c
All endpointsd 0.26 (0.10,0.43)c 0.001c <0.001c 73.90c
In utero exposure
Leukaemia − 2.70 (−11.07,5.67) 0.527 0.966 0.00
Lymphoma (including CLL) 229.10 (−200.94,659.14) 0.296 1.000 0.00
Brain/CNS 70.00 (−229.00,369.00) 0.646 1.000 0.00
Lung −1.40 (−3.82,1.03) 0.258 1.000 0.00
Thyroid (including nodules) 2.55 (0.20,4.90) 0.033 0.937 0.00
All solid except brain/CNS, lung, thyroid −1.09 (−2.75,0.56) 0.196 0.754 9.23
All six endpoints onlyd −0.26 (−1.92,1.39) 0.757 0.478 29.53
All endpointsd 0.96 (− 1.12,3.04)c 0.365c 0.004c 58.84c
Analysis using higher dose risk estimates e
Postnatal exposure
Leukaemia 5.20 (− 1.13,11.53)c 0.108c 0.027c 44.13c
Lymphoma (including CLL) 1.21 (−11.96,14.38) 0.857 0.899 0.00
Brain/CNS 6.87 (1.02,12.72) 0.021 <0.001 85.15
Lung 0.41 (−0.23,1.06) 0.209 0.483 0.00
Thyroid (including nodules) 0.01 (−0.08,0.09)c 0.874c <0.001c 73.37c
All solid except brain/CNS, lung, thyroid 0.77 (−0.61,2.14) 0.274 0.007 83.57
All six endpoints onlyd 0.28 (0.12,0.44)c <0.001c <0.001c 74.85c
All endpointsd 0.41 (0.24,0.58)c <0.001c <0.001c 80.79c
In utero exposure
Leukaemia − 2.70 (−11.07,5.67) 0.527 0.966 0.00
Lymphoma (including CLL) 229.10 (−200.94,659.14) 0.296 1.000 0.00
Brain/CNS 70.00 (−229.00,369.00) 0.646 1.000 0.00
Lung −1.40 (−3.82,1.03) 0.258 1.000 0.00
Thyroid (including nodules) 1.54 (−0.15,3.22) 0.074 0.890 0.00
All solid except brain/CNS, lung, thyroid −1.09 (−2.75,0.56) 0.196 0.754 9.23
All six endpoints onlyd −0.38 (−1.87,1.10) 0.613 0.550 26.44
All endpointsd 0.70 (− 1.17,2.57)c 0.463c 0.006c 56.70c
a

using binomial odds model to refit thyroid nodule data < 0.799 Gy of Hatch et al (Hatch et al. 2019) and leukaemia data of Stevens et al (Stevens et al. 1990), and inverse-variance weighted linear model to refit thyroid cancer data < 0.284 Gy of Kopecky et al (Kopecky et al. 2006)

b

using refitted thyroid nodule data < 0.799 Gy of Hatch et al (Hatch et al. 2019), thyroid cancer data < 0.284 Gy of Kopecky et al (Kopecky et al. 2006), Cardis et al (Cardis et al. 2005) thyroid cancer data using a linear model restricted to < 1 Gy, Lubin et al (Lubin et al. 2017) data restricted to < 0.1 Gy, Preston et al (Preston et al. 2007) brain/CNS and breast cancer data restricted to < 0.1 Gy, Cahoon et al (Cahoon et al. 2017b) lung cancer data restricted to < 0.1 Gy.

c

indications of non-convergence for REML model so that 1-step random effects model of DerSimonian and Laird (DerSimonian and Laird 1986) was employed instead.

d

for “all endpoints” all endpoints were considered, whereas for “all six endpoints” the endpoint considered within a particular study had to lie within one of the six specified endpoints shown.

e

using full range thyroid nodule data of Hatch et al (Hatch et al. 2019), thyroid cancer data of Kopecky et al (Kopecky et al. 2006), Cardis et al (Cardis et al. 2005) thyroid cancer data using a linear model restricted to < 2 Gy, Lubin et al (Lubin et al. 2017) data restricted to < 0.2 Gy, Preston et al (Preston et al. 2007) brain/CNS and breast cancer data restricted to < 1 Gy, Cahoon et al (Cahoon et al. 2017b) lung cancer data restricted to < 1 Gy.