Table 4.

Relationships between change in exercise behavior and incidence of MetS by Cox regression^a (n = 22,594).

Sex	Incidence of MetS
	Incidence of MetS	Model 1			Model 2			Model 3			Model 4
	(Per 1,000	ref = maintain regular exercise			ref = maintain irregular exercise			ref = no exercise			ref = increase
	person-years)	HR	95 % CI		HR	95 % CI		HR	95 % CI		HR	95 % CI
Menb	43.31c
Change in exercise behavior
Maintain regular exercise	48.44	–
Maintain irregular exercise	38.95	0.93	0.78	1.11	–
Maintain no exercise	40.07	0.79	0.68	0.92	0.85	0.71	1.02	–
Increase	42.46	0.85	0.73	0.98	0.91	0.77	1.09	1.07	0.93	1.24	–
Decrease	46.48	0.91	0.77	1.06	0.98	0.81	1.18	1.15	0.98	1.35	1.07	0.92	1.25
Womenb	43.70c
Change in exercise behavior
Maintain regular exercise	45.34	–
Maintain irregular exercise	36.77	0.95	0.84	1.09	–
Maintain no exercise	46.55	1	0.9	1.11	1.05	0.92	1.19	–
Increase	41.76	0.94	0.85	1.04	0.98	0.87	1.12	0.94	0.86	1.03	–
Decrease	45.59	1.01	0.91	1.13	1.06	0.93	1.22	1.02	0.91	1.13	1.08	0.98	1.2
a Adjusted for all covariates: age, education, degree of PA at work, living area, BMI, number of screenings attended, number of MetS components, personal disease history, health-related behaviors (smoking, chewing betel nuts, drinking alcohol), and dietary habits (vegetables, fruit, beverages, snacks).
b In general, there is no statistically significant sex difference (p = 0.29) in the incidence of MetS by multivariate Cox regression model (n = 22,594).
c Incidence of MetS = number of new cases of MetS/summed person-years of observation.