Table 2. Summary of evidence on adjudication in stroke trials.
| Study | Trial(s) included | Summary of findings |
|---|---|---|
| Adjudication of stroke or a composite outcome including stroke | ||
| Ninomiya et al.16 | PROGRESS |
Stroke SI HR = 0.74, 95% CI = (0.64–0.85). CA HR = 0.72, 95% CI = (0.62–0.83). |
| Godolphin et al.2 | CABACS, ENGAGE AF, ESPRIT, HAEST, ICSS, J-STARS, NASCET, PROGRESS, SOCRATES, SPS3, TARDIS |
Stroke and composite including stroke Pooled RTE comparing CA and SI = 1.02, 95% CI = (0.95–1.10). |
| Easton et al.20 | SOCRATES |
Stroke SI HR = 0.85, 95% CI = (0.75–0.97). CA HR = 0.86, 95% CI = (0.75–0.97). Composite including stroke SI HR = 0.88, 95% CI = (0.78–1.00). CA HR = 0.89, 95% CI = (0.78–1.01). |
| Farrant et al.21 | POINT |
Composite including stroke SI HR = 0.76, 95% CI = (0.60–0.95). CA HR = 0.75, 95% CI = (0.59–0.95). Ischemic stroke SI HR = 0.74, 95% CI = (0.58–0.93). CA HR = 0.72, 95% CI = (0.56–0.92). Major hemorrhage SI HR = 2.58, 95% CI = (1.19–5.58). CA HR = 2.32, 95% CI = (1.10–4.87). |
| Adjudication of functional outcome | ||
| McArthur et al.17 | CARS | Agreement between SI and CA at Day 30: weighted K = 0.84. Agreement between SI and CA at Day 90: weighted K = 0.80. |
| López-Cancio et al.18 | REVASCAT | SI cOR = 0.50, 95% CI = (0.30–0.83). CA cOR = 0.57, 95% CI = (0.35–0.95). |
| Godolphin et al.2 | REVASCAT | RTE comparing CA and SI = 0.87, 95% CI = (0.43–1.79) |
| Van der Ende et al.24 | MR CLEAN | SI cOR = 0.63, 95% CI = (0.45–0.86). CA cOR = 0.60, 95% CI = (0.45–0.83). |
| Adjudication of safety outcomes | ||
| Ninomiya et al.16 | PROGRESS |
Cause of death (CV vs non-CV vs cancer) Agreement between SI and CA = 88%, unweighted K = 0.79. |
| Godolphin et al.19 | ENOS |
SAEs SI reported patients with SAEs = 1031 (treatment = 522, control = 509). CA reported patients with SAEs = 1022 (treatment = 520, control = 502). ROR for any SAE comparing CA and SI = 0.96, 95% CI = (0.70–1.32). Likely causality of SAEs Agreement between SI and CA = 54%, weighted K = 0.31. |
| Easton et al.20 | SOCRATES |
Cause of death (CV vs non-CV) Agreement between SI and CA = 92%, unweighted K = 0.83. Bleeding (major vs no major) Agreement between SI and CA = 88%, unweighted K = 0.74. |
| Adjudication of baseline covariates | ||
| Godolphin et al.3 | ENOS |
Interaction p-values for subgroup analysis by stroke type, observed subgroup effect Agreement between SI and CA perfect (K = 1.00): p = 0.39. Agreement between SI and CA good (K = 0.78): p = 0.40. Agreement between SI and CA poor (K = 0.32): p = 0.55. Interaction p-values for subgroup analysis by stroke type, simulated subgroup effect Agreement between SI and CA perfect (K = 1.00): p = 0.01. Agreement between SI and CA good (K = 0.78): p = 0.03. Agreement between SI and CA poor (K = 0.32): p = 0.16. |
| Method of adjudication | ||
| No stroke-specific evidence identified | ||
| Blinding status of site assessors | ||
| Godolphin et al.22 | HAEST, ICSS, NASCET, REVASCAT, TARDIS |
Blinding is not possible or compromised Small amount of systematic error needed before trial results change—example, for a binary outcome: between 2.1% and 6% of participants need to be misclassified differentially. Study is adequately blinded Large amount of random error needed before trial results change— example, for a trial with binary outcome, 5000 patients, treatment effect (relative risk) = 0.82 and 20% event rate = 64.9% of events need to be misclassified non-differentially. |
| Cost of adjudication | ||
| Godolphin et al.23 | CABACS, ESPRIT, FASTEST, HAEST, J-STARS, NASCET, PROGRESS, TARDIS, VITATOPS |
Total cost Range: £2733.18–£135,627.40. cost-benefit of adjudication Mean cost per corrected outcome: £2295.10 (SD =£1482.42). |
SI: site investigator; CA: central adjudicator; HR: hazard ratio; cOR: common odds ratio; ROR: ratio of odds ratios; RTE: ratio of treatment effect; CV: cardiovascular; SAEs: serious adverse events; SD: standard deviation.