Table 3.
Assessment of risk of bias
| Adequate sequence generation | Allocation concealment | Masking | Incomplete outcome data addressed | Free of selective reporting | |
|---|---|---|---|---|---|
| TAX 350127 | Randomisation with stratification factors reported | Randomised | NA | All randomised patients included in the analyses | Yes, although survival not reported, data not mature |
| CHAARTED7 | Randomisation with stratification factors reported | Centrally randomised | NA | All randomised patients included in the analyses | Yes, all outcomes of interest are reported |
| GETUG-1225, 26 | Randomisation with stratification factors reported | Centrally randomised | NA | All randomised patients included in the analyses | Yes, outcomes of interest are reported, although survival data reported are not yet mature |
| STAMPEDE8 | Used a method of minimisation over a number of clinically important stratification factors with an additional random element | Central telephone randomisation | NA | All randomised patients included in the analyses | Yes, outcomes of interest are reported |
| RTOG 052128 | Randomisation with stratification factors reported | Centrally randomised | NA | 45 ineligible patients (3% of the total) were excluded from analyses; not clear if balanced by treatment group | Yes, outcomes of interest are reported |
| GETUG-159, 10 | Minimisation method with stratification factors reported | Centrally randomised | NA | All randomised patients included in the analyses | Yes, outcomes of interest are reported |
| CALGB 9020230 | Randomised block design with stratification factors | Central online registration and randomisations | Double-blind or placebo-controlled | All randomised patients are included in the efficacy analyses | Reports survival, but not failure-free survival as defined in the meta-analysis |
| RADAR21 | Minimisation with a random element and stratification factors | Central trials office computer based randomisation | Open label; the endpoints committee were unaware of patient identity or treatment assignment; treatment was not masked to the investigators, patients, or trial statistician | All randomised patients are included in the efficacy analyses | Reports survival, but not failure-free survival as defined in the meta-analysis |
| ZEUS29 | Minimisation method described by Pocock53 with stratification factors | Central randomisation by fax | Open label | 40 patients (3% of total randomised) excluded from analyses; seven patients were ineligible; 27 patients withdrew consent; six patients were lost to follow-up; exclusions are balanced by group | Reports survival, but not failure-free survival as defined in the meta-analysis |
| PR0412 | Minimisation method over five stratification factors | Central randomisation | Double blind; placebo-controlled; clinicians assessing cause of death were blinded to treatment allocation | In the primary analysis, no randomised patients were excluded from the analyses; in the analysis with long-term follow-up, 37 patients were excluded as they had not been flagged with the NHS Information Centre | Reports survival, but not failure-free survival as defined in the meta-analysis |
| PR0512 | Minimisation method over four stratification factors | Central randomisation | Double blind; placebo controlled | In the primary analysis, no randomised patients were excluded from the analyses; in the analysis with long-term follow-up, 33 patients were excluded as they had not been flagged with the NHS Information Centre | Reports survival, but not failure-free survival as defined in the meta-analysis |
| STAMPEDE8 | Used a method of minimisation over a number of clinically important stratification factors with an additional random element | Central telephone randomisation | Open label | All randomised patients included in the analyses | Yes, outcomes of interest are reported, including survival and failure-free survival |
NA=non-applicable. NHS=National Health Service.