Table 3

Meta-regression of effect modifiers

	Estimate (LOR)	SE	t value	P value	95% CI
Crude model 51 studies (k=104)
Main effects
Intercept	−0.35	0.05	−6.8	<0.0001	−0.46 to −0.25
Female (%)	0.06	0.18	0.34	0.7335	−0.30 to 0.43
Age	−0.06	0.07	−0.89	0.3733	−0.19 to 0.07
Design	−0.17	0.06	−2.82	0.0059	−0.28 to −0.05
Follow-up	0.004	0.008	0.53	0.5965	−0.012 to 0.02
Exposure dose	−0.019	0.002	−8.6	<0.0001	−0.02 to −0.14
Depression measure	0.07	0.12	0.57	0.5707	−0.17 to 0.31
Physical activity measure	−0.18	0.04	−4.2	<0.0001	−0.27 to −0.10
Physical activity domain	−0.09	0.13	−0.70	0.4876	−0.34 to 0.16
Study quality	0.006	0.07	0.82	0.9352	−0.14 to 0.15
Adjusted model 91 studies (k=179)
Intercept	−0.26	0.022	−11.9	<0.0001	−0.31 to −0.22
Female (%)	−0.011	0.04	−0.26	0.7986	−0.10 to 0.08
Age	0.009	0.026	0.36	0.7212	−0.04 to 0.06
Design	−0.17	0.05	−3.8	0.0002	−0.26 to −0.08
Follow-up	−0.002	0.004	−0.60	0.5552	−0.01 to 0.005
Exposure dose	−0.08	0.007	−11.2	<0.0001	−0.09 to −0.06
Depression measure	−0.05	0.05	−0.99	0.3250	−0.14 to 0.05
Physical activity measure	0.017	0.05	0.35	0.7291	−0.08 to 0.11
Physical activity domain	−0.09	0.06	−1.57	0.1192	−0.20 to 0.024
Study quality	−0.032	0.036	−0.88	0.3805	−0.10 to 0.040
Adjustments	−0.012	0.032	−0.38	0.7062	−0.07 to 0.05

K, number of effects; LOR, log odds ratio.