Table 4.
Cox proportional hazard | |||
---|---|---|---|
Hazard ratio (95% CI) | p‐value | p‐value of comparison RWD to RCT | |
RCT approach—trial data | |||
Survival model 1 | 0.78 (0.68;0.89) | <0.01 | NA |
Multilevel survival model 1 | 0.77 (0.43;1.10) | 0.13 | Comparator |
Survival model 2 | 0.93 (0.76;1.16) | 0.55 | NA |
Multilevel survival model 2 | 0.93 (0.72;1.15) | 0.55 | NA |
RWD approach—observational data unadjusted | |||
Naive survival model | 1.65 (1.13;2.42) | 0.01 | 0.12 |
RWD approach—observational data adjusted based on propensity scores | |||
PS matching—caliper 0 | |||
Multivariate survival model | 0.95 (0.50;1.80) | 0.88 | 0.49 |
PS matching—caliper 0.2*SD logit propensity | |||
Multivariate survival model | 1.00 (0.58;1.70) | 0.98 | 0.41 |
PS inverse weighting | |||
Multivariate survival model | 0.88 (0.24;3.21) | 0.99 | 0.71 |
PS stratification | |||
Multivariate survival model | 1.05 (0.04;2.06) | 0.99 | 0.47 |
Survival model 1 refers to the analysis in which a treatment effect was estimated for a fluoropyrimidine regimen compared to control. IMPACT, QUASAR and Schippinger et al. were included in this analysis. Survival model 2 refers to the analysis in which a treatment effect was estimated for fluoropyrimidine in combination with oxaliplatin compared to fluoropyrimidine monotherapy. MOSAIC and NSABP C07 were included in this analysis. Bold results are considered as main results.
Abbreviations: NA, not applicable; PS, propensity score; RCT, randomized clinical trial; RWD, real‐world data; SD, standard deviation.