Skip to main content
. 2012 Mar 27;27(5):319–332. doi: 10.1007/s10654-012-9674-x

Table 5.

Pooled mean differences of BMD comparing diabetes with non-diabetes

Site Groups Number of studies Sample size (case/control) Mean difference of BMD (g/cm2) P value Heterogeneity
I2 (%) Q test P value
Femoral neck Total 12 2,720/12,707 0.04 [0.02, 0.05] <0.00001 83 <0.0001
Female 10 1,234/5,752 0.04 [0.03, 0.06] <0.00001 71 0.0002
Male 7 1,486/6,955 0.03 [0.00, 0.05] 0.09 87 <0.0001
Hip Total 3 1,143/7,282 0.06 [0.04, 0.08] <0.00001 78 0.0002
Female 3 777/6,610 0.07 [0.04, 0.11] <0.00001 82 0.001
Male 2 366/672 0.04 [0.01, 0.08] 0.007 63 0.07
Spine Total 12 2,833/17,677 0.06 [0.04, 0.07] <0.00001 66 <0.0001
Female 11 1,583/11,354 0.07 [0.05, 0.09] <0.00001 62 0.003
Male 6 1,250/6,323 0.05 [0.01, 0.07] 0.008 74 0.002
Forearm Total 6 918/1,013 −0.003 [−0.02, 0.02] 0.90 88 <0.0001
Female 5 601/652 0.01 [−0.02, 0.03] 0.68 93 <0.0001
Male 4 317/361 −0.01 [−0.04, 0.02] 0.44 79 0.003

The weighted mean difference estimates of BMD were calculated as DerSimonian and Laird estimators using random effects models

Tests for heterogeneity were performed by applying the Cochran Q test