Table 3.

Stratified analyses of pooled ORs (95% CI) of hypertension, diabetes and obesity comparing the bottom versus the top of the subjective social status ladder (number of pooled studies in parenthesis after effect estimate)

	Hypertension			Diabetes			Obesity
	OR (95% CI)	Heterogeneity p value	Meta-regression p value	OR (95% CI)	Heterogeneity p value	Meta-regression p value	OR (95% CI)	Heterogeneity p value	Meta-regression p value
Ethnicity
Predominantly (≥75% of sample) Caucasian	2.61 (1.42 to 4.79) (4)	0.002	0.238	2.35 (1.35 to 4.10) (4)	0.153	0.175	3.01 (1.24 to 7.34) (2)	0.082	0.065
Not predominantly (<75% of sample) Caucasian	1.24 (0.84 to 1.83) (2)	0.535		1.37 (0.87 to 2.15) (2)	0.838		1.06 (0.76 to 1.48) (3)	0.772
Sex
Male	1.57 (1.03 to 2.38) (4)	0.011	0.635	1.99 (1.40 to 2.84) (5)	0.246	0.819	1.02 (0.76 to 1.37) (3)	0.405	0.245
Female	1.77 (1.27 to 2.49) (4)	0.133		2.14 (1.34 to 3.42) (5)	0.069		1.66 (0.88 to 3.13) (3)	0.026
Outcome measure*
Objective				1.75 (0.69 to 4.47) (3)	0.369	0.910
Subjective				1.93 (1.29 to 2.88) (3)	0.031
Adjustment of non-SES confounders
Adequate (adjustment for more than age alone)	2.81 (1.25 to 6.28) (2)	0.108	0.322	2.35 (1.35 to 4.10) (4)	0.153	0.175	3.01 (1.24 to 7.34) (2)	0.082	0.065
Inadequate (no adjustment or for age alone)	1.51 (1.00 to 2.28) (3)	0.107		1.37 (0.87 to 2.15) (2)	0.838		1.06 (0.76 to 1.48) (3)	0.772
Selection bias†
Lower selection bias (not convenience sample and <20% exclusion rate)				2.21 (0.93 to 5.29) (3)	0.074	0.403	2.18 (1.08 to 4.41) (3)	0.030	0.221
Higher selection bias (convenience sample or ≥20% exclusion rate)				1.62 (1.12 to 2.35) (3)	0.412		0.99 (0.66 to 1.49) (2)	0.676

*Stratified analysis not performed for hypertension (5 of 6 studies had objective outcomes), or obesity (4 of 5 studies had objective outcomes).

†Stratified analysis not performed for hypertension (5 of 6 studies had lower selection bias).

SES, socioeconomic status.