Table 3.
Presentation and incorporation of risk of bias assessment into the analysis in individual systematic reviews
| Overall (n=200) | Cochrane (n=100) | Non-Cochrane (n=100) | Difference between proportions (95% CI) | |
|---|---|---|---|---|
| Presentation of risk of bias assessment | ||||
| Description or table only | 39 (19.5%) | 0 | 39 (39%) | −39.0 (−48.6 to −29.4) |
| Description and table | 56 (28%) | 38 (38%) | 18 (18%) | 20.0 (7.8 to 32.0) |
| Description, table and figure/graph | 66 (33%) | 62 (62%) | 4 (4%) | 58.0 (47.8 to 68.3) |
| Not reported | 39 (19.5%) | 0 | 39 (39%) | −39.0 (−48.6 to −29.4) |
| Proportion of trials at risk of bias per systematic review * | ||||
| ≥1 Trial at high risk of bias | 116/154 (75%) | 81/100 (81%) | 35/54 (65%) | |
| Median proportion per review (IQR) | 50% (31% to 89%) | 57% (33% to 89%) | 50% (25% to 100%) | |
| ≥1 Trial at unclear risk of bias | 119/154 (77%) | 99/100 (99%) | 20/54 (37%) | |
| Median proportion per review (IQR) | 85% (57% to 100%) | 91.5% (69.5% to 100%) | 63.5% (41% to 100%) | |
| Not reported | 46 | 0 | 46 | |
| ≥1 Trial at high risk of bias and incorporated into interpretation of results | ||||
| Abstract | 65/116 (56%) | 51/81 (63%) | 14/35 (40%) | |
| Plain language summary | – | 34/81 (42%) | – | |
| Discussion | 101/116 (87%) | 78/81 (96%) | 23/35 (66%) | |
| Conclusion | 47/116 (41%) | 40/81 (49%) | 7/35 (20%) | |
| Assessment incorporated into GRADE | ||||
| Yes | 51 (25.5%) | 45 (45%) | 6 (6%) | 39.0 (28.2 to 49.8) |
| Not applicable (GRADE not used) | 149 (74.5%) | 55 (55%) | 94 (94%) | |
| How assessment was incorporated into the results | ||||
| Descriptive only | 174 (87%) | 89 (89%) | 85 (85%) | 4.0 (−5.3 to 13.3) |
| Meta-analysis only | 1 (0.5%) | 0 | 1 (1%) | −1.0 (−2.9 to 0.9) |
| Both | 18 (9%) | 11 (11%) | 7 (7%) | 4.0 (−3.9 to 11.9) |
| Not performed | 7 (3.5%) | 0 | 7 (7%) | −7.0 (−12.0 to −1.9) |
*Proportion of trials at risk of bias per systematic review: based on approach or scoring system used by authors of systematic review and where it was possible to evaluate (eg, Cochrane ≥ one key domain not adequate (high risk of bias) or not reported (unclear risk of bias); Jadad ≥ three high quality (low risk of bias) ≤ two low quality (high risk of bias); Pedro≥six high quality (low risk of bias)≤five low quality (high risk of bias).
GRADE, Grading of Recommendations Assessments, Development and Evaluation.