Table 4.

Results From Piecewise Regression Analyses of the Association Between Retirement and Categorical Behavioral Risk Factors in the China Health and Nutrition Survey, 1991–2011

Consumption Levela and Slope	Alcohol Drinking (n = 970)			Smoking (n = 1,084)
	ORb	95% CI	P Valuec	ORd	95% CI	P Valuec
Light/moderate use versus nonuse
Preretirement	0.912	0.882, 0.942	<0.001	0.950	0.927, 0.974	<0.001
Postretirement change	1.005	0.957, 1.057	0.83	1.003	0.965, 1.042	0.89
Heavy use versus nonuse
Preretirement	0.955	0.918, 0.995	0.03	0.960	0.935, 0.985	<0.01
Postretirement change	0.919	0.862, 0.982	0.01	0.963	0.923, 1.005	0.08

Abbreviations: CI, confidence interval; OR, odds ratio.

^a Light/moderate smoking was defined as <20 cigarettes/day; heavy smoking was defined as ≥20 cigarettes/day. Light/moderate drinking was defined as <60 g/day for men and <40 g/day for women; heavy drinking was defined as ≥60 g/day for men and ≥40 g/day for women.

^b Results were fully adjusted (see Table 2, footnote “a”) except for use of antihypertensive medication and preretirement alcohol consumption.

^cP value for piecewise logistic regression; tests of statistical significance were 2-sided.

^d Results were fully adjusted (see Table 2, footnote “a”) except for use of antihypertensive medication and preretirement smoking status.