Skip to main content
. 2016 Dec;80:68–76. doi: 10.1016/j.jclinepi.2016.07.003

Table 2.

Weight-loss programs: summary results (difference in mean weight loss [kg]) and GRADE assessment of the direct and indirect analyses and NMAs

Comparison (active vs. control) Mean difference (95% CI)
Quality of evidence (GRADE)
Direct Indirect Network Direct Indirect Network
LEARN vs. No diet 3.67 (−3.88, 11.21) 3.63 (0.36, 6.91) 5.16 (2.68, 7.63) Low Low Low
Moderate vs. No diet 4.84 (2.82, 6.86) 4.69 (1.73, 7.75) 5.70 (4.14, 7.35) Low Low Moderate
Low Carb vs. No diet 9.34 (7.31, 11.37) 5.16 (2.25, 8.18) 7.25 (5.33, 9.25) Low Moderate Moderate
Low fat vs. No diet 5.97 (2.01, 9.92) 6.15 (2.96, 9.40) 7.27 (5.26, 9.34) Moderate Moderate Moderate
Moderate vs. LEARN 0.21 (−4.64, 5.05) 0.94 (−1.74, 3.66) 0.55 (−1.71, 2.87) Low Low Low
Low Carb vs. LEARN 1.23 (−1.22, 3.67) 2.48 (−0.19, 5.19) 2.10 (−0.20, 4.47) Low Low Low
Low fat vs. LEARN 4.00 (−0.21, 8.21) 2.64 (−0.02, 5.33) 2.12 (−0.33, 4.59) Low Low Low
Low Carb vs. Moderate 1.07 (0.16, 1.97) 2.05 (−0.92, 4.96) 1.55 (0.13, 2.95) Moderate Low Moderate
Low fat vs. Moderate 1.84 (0.96, 2.72) 1.38 (−0.75, 3.51) 1.56 (−0.17, 3.30) Moderate Low Moderate
Low fat vs. Low Carb 0.33 (−0.86, 1.52) 0.39 (−1.92, 2.70) 0.02 (−1.78, 1.79) Low Moderate Moderate

Abbreviations: GRADE, Grading of Recommendations Assessment, Development and Evaluation; NMAs, network meta-analyses; CI, confidence interval.

Adapted from [14]. Direct estimates were reported by original authors as being based on Der Simonian and Laird [37]; network estimates were reported as based on hierarchical Bayesian network meta-regression [38] “accounting for exercise and behavioral support” [14].