Table 2.
Comparison of treatment effects of trials using surrogate outcomes with trials using final patient relevant outcomes: primary and sensitivity analyses
| Method of analysis | Risk ratio* (95% CI) | Ratio of odds ratios or relative risk ratio (95% CI) | |||
|---|---|---|---|---|---|
| Surrogate outcomes | Patient relevant outcomes | Unadjusted | Adjusted† | ||
| Primary analysis: | |||||
| Binary outcomes (51 surrogate v 83 patient relevant) | 0.51 (0.42 to 0.60) | 0.76 (0.70 to 0.82) | 1.47 (1.07 to 2.01) | 1.46 (1.05 to 2.04) | |
| Sensitivity analyses: | |||||
| Inclusion of risk ratios as reported by authors (57 v 86) | 0.56 (0.48 to 0.65) | 0.80 (0.75 to 0.86) | 1.38 (1.12 to 1.71) | 1.36 (1.08 to 1.70) | |
| Inclusion of continuous outcomes (84 v 101) | 0.46 (0.39 to 0.54) | 0.68 (0.62 to 0.74) | 1.44 (0.83 to 2.49) | 1.48 (0.83 to 2.62) | |
| Binary outcomes, matched pairs (43 v 43) | 0.48 (0.39 to 0.59) | 0.68 (0.61 to 0.77) | 1.38 (1.01 to 1.88) | — | |
*Pooled using DerSimonian and Laird random effects meta-analyses.
†Adjusted for trial level characteristics of clinical area of intervention, patient population, type of intervention, sponsor, journal, mean sample size, and mean follow-up time.