Table 3

Unadjusted and multivariable adjusted HRs of death in cohort 1 and cohort 2 according to May-adjusted 25OHD tertile

Model	n	Cohort 1
		Tertile 1 <18.1 nmol/L	Tertile 2 18.1–33.1 nmol/L			Tertile 3 >33.1 nmol/L			P_trend
		HR	HR	95% CI	P value	HR	95% CI	P value	P_trend
CRC death
Model 1*	1685	Ref	0.84	0.67 to 1.06	0.15	0.74	0.58 to 0.93	0.01	0.05
Model 2†	1486	Ref	0.85	0.67 to 1.09	0.20	0.71	0.55 to 0.92	0.01	0.03
Model 3‡	1058	Ref	0.77	0.59 to 1.04	0.09	0.66	0.49 to 0.89	0.006	0.007
All death
Model 1*	1685	Ref	0.77	0.65 to 0.92	0.004	0.70	0.58 to 0.84	0.0001	0.0006
Model 2†	1486	Ref	0.77	0.64 to 0.92	0.005	0.69	0.56 to 0.84	0.0002	0.0004
Model 3‡	1058	Ref	0.72	0.58 to 0.91	0.006	0.65	0.51 to 0.81	0.0002	0.0002

	n	Cohort 2
	n	Tertile 1 <38.0 nmol/L	Tertile 2 38.0–57.9 nmol/L				Tertile 3 >57.9 nmol/L		P_trend
CRC death		HR	HR	95% CI	P value	HR	95% CI	P value	P_trend
Model 1*	1844	Ref	0.70	0.47 to 1.04	0.08	0.56	0.37 to 0.86	0.008	0.10
Model 2†	1792	Ref	0.77	0.52 to 1.13	0.19	0.62	0.40 to 0.95	0.03	0.13
All death
Model1*	1844	Ref	0.76	0.55 to 1.04	0.09	0.58	0.41 to 0.82	0.002	0.04
Model 2†	1792	Ref	0.84	0.61 to 1.15	0.27	0.63	0.44 to 0.89	0.01	0.08

Final model is shown in bold.
*Model adjusted for age, sex and American Joint Committee on Cancer (AJCC) stage.
†Multivariable model additionally adjusted for body mass index, tumour site (colon/rectum), time between definitive treatment and sampling. Where data were missing, participants were excluded from adjusted model reflected in decreasing numbers of patients included in each sequential model.
‡Multivariable model additionally adjusted for C-reactive protein (CRP) which is not displayed for cohort 2 as only a small subset had CRP assayed. Trend—25OHD as a continuous variable. Adjustment of model 3 for dietary intake of vitamin D or vitamin D supplementation did not substantially alter observed HRs.
25OHD, 25-hydroxyvitamin D; CRC, colorectal cancer.