Table 2. Overall and subgroup analyses of the association between use of ADs and incidence of diabetes.

Stratification	Groups	No. of studies	Fixed effect model	Random effect model	I2 (%)
Overall studies		20	1.31 (1.28–1.33)	1.27 (1.19–1.35)	71
Type of antidepressant
	SSRI	7	1.23 (1.15–1.31)	1.21 (1.07–1.37)	70
	Non-SSRI	7	1.40 (1.32–1.48)	1.31 (1.16–1.47)	65
Country
	USA	10	1.28 (1.23–1.34)	1.28 (1.16–1.43)	78
	Non-USA	10	1.31 (1.29–1.34)	1.25 (1.13–1.39)	62
Type of study
	Cohort	16	1.31 (1.28–1.33)	1.26 (1.17–1.35)	74
	Nested Case Control	4	1.29 (1.18–1.42)	1.33 (1.12–1.58)	67
Source of information about AD treatment
	Self-report	12	1.23 (1.17–1.29)	1.25 (1.14–1.37)	59
	Prescriptions	8	1.32 (1.29–1.34)	1.28 (1.16–1.43)	78
Source of information about the diagnosis of diabetes
	Self-report	9	1.23 (1.17–1.29)	1.23 (1.15–1.32)	34
	Antidiabetics prescriptions or clinical diagnosis	11	1.32 (1.29–1.34)	1.29 (1.16–1.44)	79
Study Quality	NOS <8	14	1.30 (1.28–1.33)	1.24 (1.15–1.33)	78
	NOS 8–9	6	1.36 (1.24–1.49)	1.40 (1.24–1.57)	19
Adjustment for specific risk factors
	BMI	11	1.22 (1.16–1.28)	1.21 (1.12–1.31)	54
	Depression	13	1.30 (1.28–1.33)	1.25 (1.16–1.34)	68
	BMI and depression	10	1.21 (1.16–1.27)	1.20 (1.10–1.30)	57