Table 5.

Logistic regression analysis for effects of leisure activities on persistent loneliness at 2011.

	Model A		Model B
	Odds Ratio	95% CI	Odds Ratio	95% CI
2005
Cards/Mahjong (ref: Never)
Frequent	0.44**	(0.28, 0.69)	0.43*	(0.26, 0.72)
Occasional	0.68	(0.41, 1.13)	0.79	(0.47, 1.36)
TV/Radio (ref: Never)
Frequent	0.65*	(0.48, 0.90)	1.07	(0.75, 1.54)
Occasional	0.60+	(0.36, 1.01)	0.76	(0.44, 1.31)
Social Activities (ref: Never)
Frequent	0.75	(0.44, 1.29)	1.05	(0.60, 1.86)
Occasional	0.63*	(0.41, 0.97)	0.85	(0.54, 1.34)
2008
Cards/Mahjong (ref: Never)
Frequent	0.81	(0.55, 1.19)	1.41	(0.91, 2.20)
Occasional	0.67	(0.38, 1.20)	1.04	(0.57, 1.89)
TV/Radio (ref: Never)
Frequent	0.51**	(0.37, 0.68)	0.68*	(0.48, 0.95)
Occasional	0.53*	(0.33, 0.87)	0.61+	(0.37, 1.00)
Social Activities (ref: Never)
Frequent	0.51	(0.25, 1.04)	0.72	(0.34, 1.52)
Occasional	0.84	(0.54, 1.30)	1.10	(0.69, 1.74)

Note: **p < 0.001; *p < 0.05; ⁺p < 0.06.

Model A: Crude model, no adjustment.

Model B: Adjusted for gender, age, ethnicity, education, residence, marital status, living arrangements, and self-rated health at baseline. Omnibus Test: ${\chi }^2\left(22\right)=75.968,p<0.001;$ Hosmer-Lemeshow: ${\chi }^2\left(8\right)=12.276,p=0.139$.