TABLE 2.
Results from simulation model primary analysis and case study a
| Primary simulation | Case study | |
|---|---|---|
| Hypothetical base case | RWD versus RWD simulation | |
| INPUT | ||
| Treatment Group A (reference) | Drug A | Chemotherapy |
| True median PFS (95% CI), months | 4.0 (3.5, 4.5) | 6.0 (5.3, 6.8) b |
| Imaging frequency, median weeks between scans | 12.0 | 10.6 |
| Treatment Group B | Drug B | PD‐L1inh. |
| True median PFS (95% CI), months | 5.0 (4.4, 5.7) | 12.0 (10.5, 13.6) b |
| Imaging frequency, median weeks between scans | 9.2 | 10.8 |
| True difference in median PFS, months | 1.0 | 6.0 |
| True HR, (95% CI) | 0.80 (0.71, 0.91) | 0.50 (0.44–0.57) b |
| RESULT | ||
| Observed difference in median PFS, months | 0.64 | 6.6 |
| Observed HR, (95% CI) | 0.86 (0.76, 0.97) | 0.51 (0.44–0.57) |
| Bias in HR, mean relative % (95% CI) | −7.0% (−20.6, 5.7) | −1.3% (−14.0, 11.7) |
| Conclusions differ, c % of 1000 simulations | 30% | 0 |
Abbreviations: CI, confidence interval; EGFR, epidermal growth factor receptor; HR, hazard ratio; PD‐(L)1, programmed death (ligand) 1; RWD, real‐world data; TKI, tyrosine kinase inhibitor.
a
EHR‐derived data analysis stratified by treatment class. Note: Each case study and the main analysis simulate 500 patients in each treatment group and 1000 comparative‐effectiveness trials.
b
Based on the Keynote‐024 study. 20
c
In the trials simulations where conclusions differed, the 95% CI of the observed HR crossed 1.0 and the null hypothesis could not be rejected.