Table 2 |
|||
|
Comparison of the results of the random-effects meta-analysis (relative proportion of patients with resolved metabolic syndrome (MetS) and difference in mean) |
|||
|
Total (95% CI) |
Model |
Point Estimate d |
95% CI (lower to upper) |
|
|
|||
|
The ratio of proportions patients with resolved MetS (7 interventions)a |
Random |
2.0 |
1.5 to 2.7 |
|
Bayesian |
2.4 |
1.5 to 5.5 |
|
|
Fixed |
1.9 |
1.7 to 2.2 |
|
|
|
|||
|
Mean difference (7 interventions) b |
|||
|
Fasting blood glucose, mg/dl |
Random c |
-11.5 |
-22.4 to -0.6 |
|
(5 interventions) |
Bayesian |
-5.1 |
-8.6 to -0.9 |
|
Fixed |
-18.7 |
-20.4 to -17.0 |
|
|
Waist circumference, cm (6 interventions) |
Random c |
-2.7 |
-4.6 to -0.9 |
|
Bayesian |
-2.6 |
-4.6 to -0.6 |
|
|
Fixed |
-2.0 |
-2.1 to -1.9 |
|
|
SBP, mmHg |
Random c |
-6.4 |
-9.7 to -3.2 |
|
Bayesian |
-7.9 |
-12.3 to -3.9 |
|
|
Fixed |
-3.1 |
-3.6 to -2.7 |
|
|
DBP, mmHg |
Random c |
-3.3 |
-5.2 to -1.4 |
|
Bayesian |
-3.5 |
-6.2 to -0.97 |
|
|
Fixed |
-2.0 |
-2.3 to -1.8 |
|
|
Triglyceride, mg/dl |
Random c |
-12.0 |
-22.2 to -1.7 |
|
Bayesian |
-4.8 |
-28.1 to 18.8 |
|
|
Fixed |
-18.7 |
-20.4 to -17.0 |
|
|
HDL, mg/dl |
Random c |
1.3 |
-0.6 to 3.1 |
|
Bayesian |
0.7 |
-2.4 to 3.8 |
|
|
Fixed |
2.9 |
2.5 to 3.3 |
|
|
|
|||
|
a8 interventions including estimated data denote 2.0 (1.5 to 2.5) for the random-effects, 2.7 (1.6 to 5.8) for the Bayesian, and 1.9 (1.7 to 2.2) for the fixed model, respectively. Number of subjects that could have resolved MetS = 1400 (intervention) and 1439 (control), 2839 in total. Proportions of resolved patients = 35.0% (intervention) and 17.6% (control). bNumber of subjects for mean difference = 379 (intervention) and 369 (control), 748 in total. c Random = DerSimonian-Laird method, 95% CI = 95% confidence interval. d The estimates by the sensitivity analyses using r = 0.3 and r = 0.7 were the same with r = 0.5 for fasting blood glucose and waist circumference, and it varied for SBP (-7.1, -6.4), DBP (-3.6, -2.9), triglycerides (-12.7, -11.5), and HDL (0.9, 1.6). The estimates by the subgroup analyses using 4 interventions [18,20,21] (random-effects model) denote -3.1 (-5.3 to -0.9) for fasting blood glucose, -3.1 (-5.2 to -1.0) for waist circumference, -5.1 (-8.8 to -1.9) for SBP, -2.3 (-3.3, -1.3) for DBP, -11.5 (-22.4 to -0.6) for triglyceride, and 1.5 (-0.6 to 3.47) for HDL. |
|||
|
Yamaoka and Tango BMC Medicine 2012 10:138 doi:10.1186/1741-7015-10-138 |
|||
