Table 5. The network effects of different label types on the perceived disease risk of consuming products.
| Outcome | Comparison | Number of comparisons | Direct estimate (RMD and 95% CI) | Indirect estimate (RMD and 95% CI) | Network estimate (RMD and 95% CI) | Proportion of direct evidence |
|---|---|---|---|---|---|---|
| Less healthful products | HW vs. control* | 10 | 0.124 (0.053, 0.196) | 0.6 (0.153, 1.048) | 0.136 (0.065, 0.207) | 1.00 |
| HW vs. NW* | 2 | 0.063 (−0.09, 0.215) | −0.562 (−0.766, −0.357) | −0.16 (−0.283, −0.038) | 0.67 | |
| HW vs. NW + HW* | 1 | 0.03 (−0.181, 0.241) | −0.653 (−1.006, −0.3) | −0.15 (−0.331, 0.031) | 0.93 | |
| NW vs. control* | 3 | 0.43 (0.299, 0.561) | −0.304 (−0.582, −0.027) | 0.296 (0.178, 0.415) | 0.66 | |
| NW vs. NW + HW* | 1 | −0.11 (−0.324, 0.104) | 0.463 (0.048, 0.878) | 0.01 (−0.18, 0.2) | 0.61 | |
| NW + HW vs. control | 1 | 0.36 (0.146, 0.574) | 0.102 (−0.236, 0.439) | 0.286 (0.105, 0.467) | 0.23 | |
| More healthful products | HW vs. control | 2 | 0.024 (0.008, 0.04) | . | 0.024 (0.008, 0.04) | 1.00 |
RMD refers to the percentage of change comparing intervention with control group (RMD = (x2 − x1) / x1; x1: mean of continuous outcome in the intervention group or after implementation of intervention, x2: mean of continuous outcome in the control group or before implementation of intervention).
*The direct and indirect effects were observed significantly inconsistent (p < 0.05), and the network estimate may violate the assumption of consistency and transitivity of NMA, thus only direct evidence was used for interpretation.
CI, confidence interval; HW, health warning; NMA, network meta-analysis; NS, Nutri-Score; NW, nutrient warning; RMD, relative mean difference; TLS, traffic light labelling system.