Table 4.

Associations of eating or drinking prior to bedtime with weekdays sleep duration among participants of the American time use survey who did not report wake after sleep onset (WASO)

(Odds ratio and 95 % confidence intervals)

	Sleep duration†		Short sleep‡		Long sleep‡
	Beta	95 % CI	Odds ratio	95 % CI	Odds ratio	95 % CI
All respondents
Ate < 1 h before bedtime
Crude model	30	27, 33*	0·90	0·84, 0·97*	1·54	1·44, 1·64*
Adjusted model§	29	26, 32*	0·88	0·81, 0·95*	1·77	1·66, 1·90*
Women
Ate < 1 h before bedtime
Crude model	37	32, 42*	0·85	0·76, 0·95*	1·59	1·45, 1·74*
Adjusted model§	35	30, 39*	0·84	0·75, 0·94*	1·79	1·63, 1·97*
Men
Ate < 1 h before bedtime
Crude model	25	21, 30*	0·92	0·84, 1·02	1·54	1·41, 1·68*
Adjusted model§	24	20, 28*	0·91	0·82, 1·00	1·76	1·61, 1·94*

^*Statistically significant at P < 0·05; WASO, wake after sleep onset.

^†From a linear regression model with sleep duration in minutes as the outcome and eating or drinking < 1 h before bed as a dichotomous predictor.

^‡From a multinomial logistic model with short sleep duration, sufficient sleep duration and long sleep duration as categorical outcomes (sufficient sleep was the reference) and eating or drinking < 1 h before bed as a dichotomous predictor.

^§Adjusted for gender, age, race, education, work status, cohabitation, the presence of children < 18 years in the household, day of the week and survey year.