Table 1. Summary of quality assessment criteria.
| Criteria | Yes (%) | No (%) | Unclear (%) | Not applicable |
| Indirect comparison method | ||||
| Is the method applied to undertake the indirect comparison adequate?(1) | 41 (95) | 2 (5) | 0 (0) | 0 |
| If an adequate method is used, is a treatment effect estimate and measure of precision reported? | 25 (61) | 16 (39) | 0 (0) | 2 |
| Similarity | ||||
| Is the assumption of similarity stated? | 11 (26) | 32 (74) | 0 (0) | 0 |
| Is a method described to assess the similarity assumption within the review methods section?(2) | 0 | 43 (100) | 0 (0) | 0 |
| Is a reasonable approach used to assess the assumption of similarity?(3) | 19 (44) | 22 (51) | 2 (5) | 0 |
| Are patient or trial characteristics reported for all trials in the indirect comparison? | 38 (88) | 5 (12) | 0 (0) | 0 |
| Are patient or trial characteristics compared across the two trial sets involved in the indirect comparison? | 11 (26) | 32 (74) | 0 (0) | 0 |
| Are patient or trial characteristics reported to be comparable for the two trial sets involved in the indirect comparison? | 4 (9) | 5 (12) | 34 (79) (2 unclear if comparable; 32 not reported) | 0 |
| Homogeneity across trials within each of the two trial sets involved in the indirect comparison | ||||
| Is the method used to determine the presence of statistical heterogeneity adequate?(4) | 24 (60) | 12 (30) | 4 (10) | 3 |
| Is the homogeneity assumption satisfied or is statistical heterogeneity accounted for if present?(5) | 19 (48) (8 homogeneous; 11 accounted) | 3 (8) | 18 (45) (17 unclear if homogeneous; 1 unclear if accounted) | 3 |
| If the homogeneity assumption is not satisfied, is clinical or methodological homogeneity across trials in each trial set involved in the indirect comparison investigated by an adequate method?(6) | 12 (38) | 19 (59) | 1 (3) | 11 |
| Consistency | ||||
| Is consistency of effects assessed?(7) | 6 (35) (1used statistical method) | 11 (65) | 0 (0) | 26 |
| If the direct and indirect evidence is reported to be consistent, is the evidence combined and the result presented?(8) | 1 (25) | 3 (75) | 0 (0) | 39 |
| If inconsistency is reported, is this accounted for by not combining the direct and indirect evidence?(9) | 2 (100) | 0 (0) | 0 (0) | 41 |
| Are patient or trial characteristics compared between direct and indirect evidence trials?(10) | 5 (29) | 12 (71) | 0 (0) | 26 |
| Are patient or trial characteristics for direct and indirect evidence trials reported to be comparable?(11) | 2 (12) | 1 (6) | 14 (82) (2 unclear if comparable; 12 not reported) | 26 |
| Are any included 3-arm trials correctly analysed?(12) | 3 (25) | 9 (75) | 0 (0) | 31 |
| Is justification given for using indirect evidence and direct evidence?(13) | 8 (47) | 9 (53) | 0 (0) | 26 |
| Does the review present results from all trials providing direct evidence ?(14) | (65) | 6 (35) | 0 (0) | 26 |
| Interpretation | ||||
| Is a distinction made between direct comparisons and indirect comparisons? | 25 (58) | 18 (42) | 0 (0) | 0 |
| Does the review state that more trials providing direct evidence are needed? | 24 (56) | 19 (44) | 0 (0) | 0 |
| Reporting | ||||
| Does the review present both of the meta-analysis results from each of the two trial sets involved in the indirect comparison? | 37 (86) | 6 (14) | 0 (0) | 0 |
| Was it highlighted which results were from indirect evidence?(15) | 24 (56) | 19 (44) | 0 (0) | 0 |
| Are the individual trials' treatment effect estimates reported? | 23 (53) | 20 (47) | 0 (0) | 0 |
Yes: method preserves randomization. No: method does not preserve randomization.
Yes: reported a method that is stated will assess the assumption of similarity. No: do not report a method.
Yes: sensitivity analysis, subgroup analysis, or meta-regression used to assess consistency of the indirect comparison across different trial or patient characteristics. No: no method, no analysis that includes all the trials in the indirect comparison. Unclear: unclear if the trials used in the analysis are the same trial sets involved in the indirect comparison.
Yes: Chi-square test, I-squared statistic, estimating the between trial variance from a random effects models. No: no method applied, or not applied to the two trial sets contributing to the indirect comparison. Unclear: unclear if heterogeneity was assessed for the two trial sets contributing to the indirect comparison. Not applicable: only one trial in each trial set.
Yes: no heterogeneity present (reported by authors or determined from the results), or accounted for heterogeneity using the random effects model. No: heterogeneity not accounted for using the random effects model. Unclear: unclear if heterogeneity present, or unclear if heterogeneity taken into account using the random effects model. Not applicable: only one trial in each trial set.
Yes: sensitivity analysis, subgroup analysis, or meta-regression used to assess homogeneity across different trial or patient characteristics within each of the two trials sets involved in the indirect comparison. No: no method, no analysis that includes the trials in each of the two trial sets involved in the indirect comparison. Unclear: unclear if the trials used in the sensitivity analysis, subgroup analysis, or meta-regression are the same set of trials as those in each of the two trial sets involved in the indirect comparison. Not applicable: only one trial in each trial set; or homogeneity assumption satisfied.
Not applicable: both indirect and direct evidence are not presented for the same comparison.
Not applicable: reported to be inconsistent, or unclear if consistent based on text or results.
Not applicable: reported to be consistent, or unclear if inconsistent based on text or results.
Yes: three-arm trials are correctly analysed i.e. indirect evidence (AC, BC) is not included and direct evidence (AB) is analysed, and data from a three-arm trial is not combined as though it is from two different studies. No: three-arm trials are incorrectly analysed i.e. indirect evidence (AC, BC) is included and direct evidence (AB) is not analysed, or data from a three-arm trial is combined as though it is from two different studies. Na: no three-arm trials are included in the review.
Yes: the term indirect comparison is stated when referring to the result or the result is presented under a heading that states the result is based on an indirect comparison. No: result is presented without noting it is an indirect comparison.