Table 2.

Multivariable and mendelian randomisation Cox regression analysis for risk of breast cancer associated with sleep traits

Sleep trait	Basic model*				Fully adjusted model†				Mendelian randomisation analysis‡
	No (incident cases)	Hazard ratio (95% CI)	P value		No (incident cases)	Hazard ratio (95% CI)	P value		No (incident cases)	Hazard ratio (95% CI)	P value
Chronotype (per category increase§)	151 421 (2732)	0.94 (0.92 to 0.97)	<0.001		138 529 (2500)	0.95 (0.93 to 0.98)	0.002		151 421 (2732)	0.85 (0.70 to 1.03)	0.10
Sleep duration (per hour increase)	150 845 (2723)	1.01 (0.98 to 1.05)	0.55		138 228 (2495)	1.00 (0.96 to 1.04)	0.98		150 845 (2723)	1.06 (0.70 to 1.59)	0.78
Insomnia symptoms (per category increase¶)	149 005 (2740)	1.02 (0.97 to 1.08)	0.44		138 771 (2505)	1.02 (0.97 to 1.08)	0.44		149 005 (2740)	1.37 (0.59 to 3.20)	0.47

^*Adjusted for age, assessment centre, and top 40 genetic principal components.

^†Adjusted for age, assessment centre, top 40 genetic principal components, degree status, body mass index, alcohol intake, smoking, strenuous physical activity, family history of breast cancer, parity, age at menarche, menopause status, use of oral contraceptives, and hormone replacement therapy.

^‡Adjusted for age, assessment centre, top 40 genetic principal components, and genotyping chip.

^§From definite evening, intermediate evening, don’t know, intermediate morning, and definite morning.

^¶From none, some, and frequent.