Table 3.
Evidence quality of the 29 SRs/meta-analyses
| Reviews | GRADE | Quality | ||||
|---|---|---|---|---|---|---|
| Limitation | Inconsistency | Indirectness | Imprecision | Publication bias | ||
| Ehlers [36] | − 11 | 0 | 0 | 0 | 0 | ⊕⊕⊕Ο Moderate |
| Shen [37] | 0 | − 12 | 0 | − 13 | − 14 | ⊕ΟΟΟ Very low |
| Lee [38] | 0 | 0 | 0 | 0 | − 15 | ⊕⊕⊕Ο Moderate |
| Singh [39] | 0 | − 12 | 0 | 0 | 0 | ⊕⊕⊕Ο Moderate |
| Lipsett [40] | − 11 | − 12 | 0 | 0 | − 14 | ⊕ΟΟΟ Very low |
| Juvet [41] | 0 | − 12 | 0 | 0 | 0 | ⊕⊕⊕Ο Moderate |
| Zhu [23] | 0 | − 12 | 0 | − 13, 6 | 0 | ⊕⊕ΟΟ Low |
| McNeely [42] | 0 | − 12, 6 | 0 | 0 | − 14 | ⊕⊕ΟΟ Low |
| Gu [43] | − 11, 8 | − 12 | 0 | − 13 | 0 | ⊕ΟΟΟ Very low |
| Vannorsdall [24] | − 18 | 0 | 0 | 0 | − 14 | ⊕⊕ΟΟ Low |
| van Vulpen [44] | − 18 | 0 | 0 | 0 | − 14 | ⊕⊕ΟΟ Low |
| Ramírez-Vélez [45] | 0 | − 12 | 0 | 0 | 0 | ⊕⊕⊕Ο Moderate |
| Liu [46] | 0 | − 22 | 0 | 0 | 0 | ⊕⊕ΟΟ Low |
| Lin [47] | 0 | − 22 | 0 | 0 | 0 | ⊕⊕ΟΟ Low |
| Duijts [48] | 0 | 0 | 0 | 0 | − 15 | ⊕⊕ΟΟ Low |
| Liu [49] | 0 | − 22 | 0 | − 16 | − 14, 5 | ⊕ΟΟΟ Very low |
| Zheng [50] | − 11 | − 12 | 0 | 0 | − 14 | ⊕ΟΟΟ Very low |
| Wu [51] | 0 | − 12 | 0 | − 16 | − 15 | ⊕ΟΟΟ Very low |
| Zhang [52] | − 11, 8 | − 22 | 0 | − 16 | 0 | ⊕ΟΟΟ Very low |
| O’Neill [53] | − 11 | − 18 | 0 | 0 | 0 | ⊕⊕ΟΟ Low |
| Dong [54] | 0 | − 12 | 0 | 0 | 0 | ⊕⊕ΟΟ Low |
| Hsueh [55] | − 11 | − 22 | 0 | 0 | 0 | ⊕ΟΟΟ Very low |
| Xu [56] | 0 | − 12 | 0 | 0 | 0 | ⊕⊕ΟΟ Low |
| Hu [57] | 0 | − 12 | 0 | 0 | − 14 | ⊕⊕ΟΟ Low |
| Yang [58] | − 17 | 0 | 0 | 0 | − 14 | ⊕⊕ΟΟ Low |
| Zhang [59] | 0 | − 12 | 0 | − 16 | − 14 | ⊕ΟΟΟ Very low |
| Zou [60] | − 12 | − 23 | 0 | − 16 | 0 | ⊕ΟΟΟ Very low |
| Luo[61] | − 11 | 0 | 0 | 0 | − 14 | ⊕⊕ΟΟ Low |
| Liu[62] | − 11 | 0 | 0 | − 16 | − 15 | ⊕ΟΟΟ Very low |
1The included reviews were biased in terms of randomization, allocation concealment, and blinding method; 2the confidence intervals of different studies overlapped greatly, and the combined result of heterogeneity was large (> 80%, decreased by two grades); 3significant benefits or harms were included in the confidence interval (RR < 0.75 or RR > 1.25 were the criteria); 4whether gray literature and manual retrieval were included was not stated in the review; 5the number of included reviews was small and all positive, so publication bias should be considered; 6the invalid value (RR = 1.0) was included in the confidence interval; 7only one study was included, so heterogeneity could not be measured; 8most of the included studies were of moderate methodological quality; 9incomplete reports and outcome events and selective outcome bias (including adverse reactions, negative results) were not presented or explained. The rating standard of 1 to 9 is referenced in Sects. 4 to 8 in the GRADE guidelines [63–67]