Table 3

Risk of bias

Study (author and year)Adequate sequence generationAllocation concealmentMaskingIncomplete outcome data addressedFree of selective reportingFree of other bias (eg, similarity at baseline, power assessment)Funder
Geneva 20106–8LowLowPartial: patients and assessors of efficacy variablesLow: ITT analysis, 94% FU at 6 monthsLowPower: 81% power to detect difference in primary outcome with n=495 for each trial
Similarity at baseline: yes
Allergan Inc
Score 200923–36LowUnclearPartial (physicians and patients masked to dose but not triamcinolone vs observation)Low: ITT analysis, 83–90% FU at 12 monthsLowPower: 80% power to detect difference in primary outcome with n=486 (but only 271 randomised)
Similarity at baseline: yes
National Eye Institute grants, Allergan
Copernicus 201212 13LowUnclearLow: double-blindLow: ITT analysis, 89.9% assessed at primary end pointLowPower: 90% power to detect difference in primary outcome with n=165
Similarity at baseline: yes
Bayer HealthCare, Regeneron Pharmaceuticals
Galileo 201211UnclearUnclearLow: double-blindLow: ITT analysis, 86% assessed at primary end pointLowPower: 90% power to detect difference in primary outcome with n=150
Similarity at baseline: yes
Bayer HealthCare, Regeneron Pharmaceuticals
Cruise 20109 10LowUnclearLow: patients and evaluating examiners, injecting physicians masked to doseLow: ITT analysis, 88.5–97.7% completed 6 monthsLowPower: not reported
Similarity at baseline: yes
Genentech Inc.
Epstein 201242–44UnclearLowLow: patients, outcome assessorsLow: ITT analysis; missing data for 2 patients (primary endpoint)LowPower: 80% power to detect difference in primary outcome with n=24 per group
Similarity at baseline: yes
Unclear; authors are consultants for Allergan, Novartis, Alcon, Bayer
  • FU, follow-; ITT, intention to treat.