Table 2.

Multivariable regression and Mendelian randomisation analysis to assess associations between sleep traits among UK Biobank spouse-pairs.

		Multivariable regression						Mendelian randomisation (2SLS)
Sleep trait	N (pairs)	Risk/mean difference	SE	CIL	CIU	p-value	Risk/mean difference		SE	CIL	CIU	p-value	z-test for difference
Self-reported
Chronotype	47,235	–0.113	0.005	–0.122	–0.104	2.42E-135	–0.152		0.055	–0.26	-0.042	0.006	0.480
Ease of waking	47,325	0.015	0.005	0.005	0.023	2.00E-03	–0.039		0.079	–0.193	0.116	0.625	0.503
Sleep duration	47,050	0.111	0.005	0.102	0.120	7.97E-126	0.131		0.048	0.037	0.226	0.007	0.679
Insomnia	47,369	0.005	0.005	-0.004	0.014	0.304	–0.046		0.054	–0.151	0.058	0.386	0.347
Snoring	45,546	–0.008	0.004	-0.017	0.000	0.051	–0.154		0.061	–0.274	-0.035	0.011	0.017
Accelerometer-derived
L5-timing	3454	0.241	0.016	0.209	0.272	1.56E-50	0.486		0.232	0.032	0.939	0.036	0.292
Sleep duration	3454	0.108	0.016	0.077	0.140	1.38E-11	0.132		0.131	–0.124	0.387	0.312	0.856
Sleep episodes	3454	0.069	0.016	0.038	0.100	1.63E-05	–0.267		0.247	–0.751	0.217	0.28	0.175
Sleep efficiency	3454	0.075	0.016	0.043	0.106	3.84E-06	0.024		0.138	–0.246	0.295	0.859	0.714

Estimates represent the mean difference in the spouse’s sleep trait (in SD) per SD increase in an individual’s own sleep trait, with the exception of snoring for which estimates represent risk difference. Multivariable regression was adjusted for age at the assessment and assessment centre for both spouses. Mendelian randomisation was adjusted for age at assessment, assessment centre, genotyping chip and 10 genetic principal components (PCs) for both spouses. 2SLS = two-stage least squares.