The following information was tabulated by one reviewer and confirmed by a second reviewer: publication data; characteristics of exposed and unexposed groups; matching criteria for exposed and unexposed subjects; oral anticoagulant; and bone density.
Only cross-sectional studies were found. Six studies were from Europe, two from the United States and one from Japan. The majority of studies were conducted in Caucasian populations. Indications for anticoagulant treatment and matching criteria for controls differed among studies.
Validity scores ranged from 7 to 15 (maximum possible was 24).
Standardized mean difference (SMD) and 95% CI were as follows:
Ultra distal radius: bone mineral density was significantly decreased in the exposed group (P = 0.007). SMD = - 0.39 (95% CI: - 0.67, -0.10). Homogeneity P = 0.8231.
Distal radius: non significant trend towards decreased bone density in exposed group (P = 0.07). SMD = - 0.47 (95% CI: - 0.97, 0.04). Homogeneity P = 0.0096. After removal of outlier study the data were more homogeneous (P = 0.543) with pooled SMD = -0.21 (95% CI: - 0.48, 0.07).
Lumbar spine: non significant trend towards decreased bone density in exposed group (P = 0.092). SMD = - 0.27 (95% CI: - 0.59, 0.05). Homogeneity P = 0.0761. No single outlier was observed.
Femoral neck: no significant difference in bone density between exposed and unexposed groups (P = 0.803). SMD = 0.03 (95% CI: - 0.22, 0.29). Homogeneity P = 0.4991.
Femoral trochanter: no significant difference in bone density between exposed and unexposed groups (P = 0.219). SMD = - 0.18 (95% CI: - 0.48, 0.11). Homogeneity P = 0.3929.