Table 5. Relative risks (RR) of melanoma by daily intakes of dietary vitamin C and total retinol, stratified by risk based on nondietary factors.

	Low risk			Middle risk			High risk
	Cases	P-Yc	Multivariateb RR(95% CId)	Cases	P-Yc	Multivariateb RR(95% CId)	Cases	P-Yc	Multivariateb RR(95% CId)	Interactione
Dietary
Vitamin C (mg)										P=0.12
<90	22	167	1.00	12	75	1.00	19	40	1.00
90–114	24	158	1.15 (0.64–2.07)	25	74	2.10 (1.01–4.37)	16	39	0.82 (0.41–1.64)
115–139	17	163	0.78 (0.41–1.49)	27	81	2.34 (1.15–4.76)	21	41	1.12 (0.60–2.12)
140–174	32	179	1.29 (0.74–2.25)	34	90	2.25 (1.10–4.62)	37	44	1.76 (0.99–3.13)
⩾175	29	172	1.16 (0.65–2.06)	25	85	1.55 (0.41–5.82)	35	42	1.68 (0.93–3.02)
P for linear trend			0.68			0.18			0.01
Total
Retinol (μg)										P=0.001
<400	38	162	1.00	16	74	1.00	19	38	1.00
400–699	22	186	0.51 (0.30–0.86)	20	86	0.99 (0.50–1.94)	26	44	1.07 (0.58–1.95)
700–1099	24	162	0.61 (0.28–1.35)	28	79	1.47 (0.78–2.77)	21	41	0.89 (0.47–1.69)
1100–1799	23	168	0.50 (0.29–0.85)	30	83	1.56 (0.84–2.89)	33	43	1.39 (0.78–2.48)
⩾1800	17	160	0.39 (0.22–0.71)	29	82	1.50 (0.81–2.81)	29	40	1.36 (0.76–2.44)
P for linear trend			0.01			0.16			0.46

^aResults from NHS (1984–1998) and NHS II (1991–1999) were pooled to obtain risk estimates for the combined cohorts. All P-values were >0.05 in tests for heterogeneity. See text for description of risk score.

^bRelative risks were adjusted for the covariates listed in Table 2 footnote c.

^cPerson-years of follow-up (thousands).

^d95% confidence interval.

^eP-value for interaction between risk score and nutrient intake.