Our literature search identified 75 articles. After the 75 abstracts had been retrieved and reviewed, 37 articles were deemed not relevant to the scope of our analysis. Therefore, 38 full manuscripts were retrieved and reviewed, and 14 were included in the meta-analysis (Figure 1). Ten cross-sectional studies were analyzed for the association between BAC and CAD.[3,5,6,8,9,10,11,12,18,19] Six cross-sectional studies were reviewed for the association between BAC and stroke.[13,18,19,20,21,22] Two of 16 studies assessed both CAD and stoke as outcomes; therefore, a total of 14 studies were included in this meta-analysis. Only one prospective study and one retrospective study were identified by our literature search; thus, they were not included in the meta-analysis. The quality characteristics of all included studies are summarized in Table 1 . The results from these studies can be found in Table 2 .
Graphical outline of the selection strategy of our literature review for the meta-analysis.
The statistical testing for heterogeneity was not significant for 10 CAD studies (Q = 4.4, P > 0.10, I 2 = 0%); however, six stroke studies were moderately heterogeneous (Q = 10.9, P < 0.10, I 2 = 54%). A fixed-effects model (Mantel-Haenszel) was used for OR calculations. No evidence of severe publication bias was suggested by the normal quantile plot for either CAD meta-analysis (Figure 2) or stroke meta-analysis (Figure 3).
We analyzed 10 cross-sectional studies with CAD diagnosed by coronary angiogram as the primary outcome (n = 3,952). The pooled OR (95% CI) for CAD in those with BAC versus those without BAC is 3.86 (3.25-4.59) (P < 0.0001) (Figure 4), similar to the prospective study results (OR, 3.54; 95% CI, 2.28-5.50; P < 0.001). For stroke, six cross-sectional studies were analyzed (n = 18,888). The OR (95% CI) for stroke in those with BAC versus those without BAC is 1.54 (1.25-1.90) (P < 0.0001) (Figure 5). OR for stroke was not available in the prospective study.
Quantile plot depicting the effect sizes of 10 cross-sectional studies and analyzing angiographically proven coronary artery disease (CAD) as a primary study outcome. The plot shows no evidence of severe publication bias.
Quantile plot depicting the effect sizes of six cross-sectional studies and analyzing stroke as a primary study outcome. The plot shows no evidence of severe publication bias.
Forest plot depicting the odds ratio (OR) of angiographically proven coronary artery disease (CAD). CL, confidence limit; LCL, lower CL; UCL, upper CL.
Menopause. 2015;22(2):136-143. © 2015 The North American Menopause Society