Table 2.

Logistic regression analysis of internet use and internet use frequency among middle-aged and older adults.

Characteristics and variables			Internet usea			Internet use frequencyb
		ORc (95% CI)		P value		OR (95% CI)	P value
Gender (men vs women)			1.15 (1.00-1.31)		.04	1.18 (1.04-1.35)		.005
Education (vs no education)
	Primary school or lower	4.76 (3.47-6.70)		<.001		4.96 (3.62-6.99)	<.001
	Middle school	11.95 (8.71-16.86)		<.001		12.76 (9.31-17.99)	<.001
	High school or higher	30.06 (21.80-42.62)		<.001		32.01 (23.24-45.34)	<.001
Nationality (Han vs ethnic minorities)			—		—	—		—
Work (working vs not working/retired)			1.16 (1.03-1.32)		.02	—		—
Age (vs 45-54 years)
	55-64	0.37 (0.33-0.42)		<.001		0.38 (0.34-0.42)	<.001
	65-74	0.21 (0.18-0.25)		<.001		0.21 (0.18-0.25)	<.001
	≥75	0.10 (0.07-0.14)		<.001		0.11 (0.08-0.14)	<.001
Area (urban vs rural)			0.44 (0.40-0.50)		<.001	0.42 (0.38-0.47)		<.001
Marital status (married vs unmarried)			—		—	—		—
Health insurance (insured vs uninsured)			—		—	—		—
Smoking status (smoking vs not smoking)			1.15 (1.01-1.31)		.03	1.14 (1.01-1.30)		.02
Drinking status (drinking vs not drinking)			0.57 (0.51-0.63)		<.001	0.57 (0.51-0.64)		<.001
Restless sleep times (vs rare)
	Some or a little	1.06 (0.93-1.21)		.37		1.03 (0.90-1.18)	.33
	Occasionally	0.90 (0.77-1.04)		.15		0.85 (0.73-0.99)	.02
	Mostly or all the time	0.86 (0.75-0.99)		.04		0.82 (0.72-0.94)	.003
ADLd (unimpaired vs impaired)			0.55 (0.45-0.67)		<.001	—		—

^aBased on univariate analysis and stepwise regression screening, all sociodemographic factors except nationality, marital status, and health insurance were included in the binary logistic regression analysis for internet use.

^bAfter stepwise regression screening and eliminating variables that caused collinearity, all sociodemographic factors except nationality, work, marital status, health insurance, and ADLs were included in the ordinal logistic regression analysis for internet use frequency.

^cOR: odds ratio.

^dADL: activity of daily living.