. 2020 Sep 29;20:101213. doi: 10.1016/j.pmedr.2020.101213

Table 3.

Associations between 24-hour movement behaviors and work engagement in workers (n = 1086).

		Model 1^a					Model 2^b
		Odds ratio^d	95% CI			p-value^e	Odds ratio^d	95% CI			p-value^e
		Odds ratio^d	Lower	–	Upper	p-value^e	Odds ratio^d	Lower	–	Upper	p-value^e
Behaviors in workday^c
	Sleep	0.33	0.17	–	0.62	<0.001	0.41	0.20	–	0.81	0.011
	SB	2.26	1.37	–	3.75	0.001	2.75	1.55	–	4.91	<0.001
	LPA	1.13	0.80	–	1.60	0.478	0.86	0.54	–	1.37	0.524
	MVPA	1.19	0.84	–	1.69	0.332	1.03	0.70	–	1.53	0.864
	Overall					0.002					0.005

Behaviors in non-workday^c
	Sleep	0.87	0.50	–	1.51	0.611	0.68	0.38	–	1.22	0.197
	SB	1.35	0.91	–	2.03	0.142	1.36	0.88	–	2.11	0.165
	LPA	1.05	0.73	–	1.53	0.784	1.31	0.87	–	1.98	0.194
	MVPA	0.81	0.66	–	0.99	0.043	0.82	0.66	–	1.03	0.088
	Overall					0.063					0.128

The results in bold are significant (p < 0.05).

Model 1 included composition of 24-hour movement behaviors.

Model 2 additionally included age, sex, body mass index, education years, self-rated economic status, marital status, health behaviors (alcohol consumption and smoking), job-type, hiring status, and average weekly overtime hours.

Time-use composition was expressed as isometric log ratio (ilr) coordinates, and each result was from the initial ilr coordinates.

The odds ratio corresponds to one unit increase in ilr coordinates. High work engagement was the reference category.

p-values for individual ilr coordinates are from Wald tests, p-value for the overall composition is from the likelihood ratio test.