Fig. 11: Inferential specification curve plot of observed overall effects (logHRs) across reasonable meta-analytic specifications from prior meta-analyses (red line) against the 95% CI band of 1000 simulated true-zero effects (gray area) using the currently available corpus of primary studies.

LogHR <0 indicates a survival benefit of the intervention group; logHR >0 indicates a survival benefit of the control group. The black dotted line indicates a zero effect (logHR = 0). The red line shows the actually observed overall effects across specifications of prior meta-analyses (sorted by size) plotted against the 95% CI band of 1000 simulated overall effects for which a zero effect was true (gray area). When the red line falls outside the gray area, it suggests evidence for a robust and generalizable true overall effect across reasonable meta-analytic specifications, k = 32 studies.