Table 3.

Logistic regression models for acute erythema (0/1).

	Non-SNP model	+ rs2087947 C/T (PER3) (fixed effects)	+ rs2087947 C/T (PER3) (random effects)
BMI	1.04 (1.02 – 1.07) P = 0.0005	1.04 (1.02 – 1.07) P = 0.0006	1.05 (1.03 – 1.08) P < 0.0001
Biological equivalent dose to breast (BED) (Gy)	1.02 (1.01 – 1.03) P = 0.0004	1.02 (1.01 – 1.03) P = 0.0005	1.02 (1.01 – 1.04) P = 0.003
Mean treatment time (hours from solar midnight)	1.05 (0.999 – 1.1) P = 0.05	1.05 (1 – 1.1) P = 0.043	1.02 (0.966 – 1.07) P = 0.49
Standard deviation of treatment times (solar time)	0.64 (0.491 – 0.832) P = 0.0009	0.645 (0.495 – 0.839) P = 0.001	0.812 (0.545 – 1.08) P = 0.2
rs2087947 (dosage, T allele)		1.27 (1.03 – 1.56) P = 0.024	1.27 (1 – 1.54) P = 0.027
(Random Effect) Treatment centre stdev:__(Intercept)			0.59
N	1198	1198	1198
ROC AUC (95% CI) a	0.61 (0.55–0.67)	0.62 (0.56–0.68)
AIC	1,244	1,241	1,203

Logistic regression. Statistics shown: odds ratio (95% CI).

^asee Supplementary Figure 1.