Table 2.

Effects on risk of breast cancer of reduced-fat dairy milks compared with full-fat milk

Dairy exposure	RR	Extreme valuesa	(95% CI)	P for trend	RR	(95% CI)	P for trend
	Model 1b				Model 2c
Total subjects
Full-fat dairy milk kcal/d)d	1.45	3.2/152	1.12–1.88	0.005	1.51	1.16–1.98	0.002
Reduced-fat dairy milk (kcal/dd	1.42	3.2/152	1.17–1.72	0.0004	1.53	1.24–1.88	<0.0001
Post-menopausal
Full-fat dairy milk kcal/d	1.47	2.7/153	1.10–1.97	0.009	1.53	1.14–2.07	0.005
Reduced-fat dairy milk kcal/d	1.43	2.7/153	1.15–1.78	0.0012	1.55	1.23–1.94	0.0002
Premenopausal
Full-fat dairy milk kcal/d	1.39	4.4/146	0.66–2.90	0.39	1.49	0.69–3.19	0.31
Reduced-fat dairy milk kcal/d	1.46	4.4/146	0.84–2.56	0.18	1.59	0.88–2.88	0.13

^aMedians of extreme intake quintiles of total milk are used for both full-fat and reduced-fat milk, to ensure comparable contrasts.

^b,cModel 1: same variables as indicated in footnote to Table 1. Model 2: Covariates as in model 1, plus dairy calories, total calcium, unprocessed and processed red meats, poultry, fish, seeds and nuts.

^dThese HRs use results from the model containing variables for total milk and reduced-fat milk as described in Statistical Analyses section. The full-fat milk HRs are derived from the coefficient for total milk alone, as this implies that reduced-fat milk is zero. The HR for reduced-fat milk alone comes from the given contrast being applied to the sum of both coefficients, as this implies the same milk quantities for both variables, hence all milk is reduced-fat. In this last case, the confidence intervals and P-values employ the necessary covariances between the beta coefficients.