The required reproduction is Python 3.12 with env/requirements.txt, invoked by code/run. Optional independent statistical validation was run with R 4.5.1 and metafor 5.2.1 in a separate Docker container. code/validate.R reads the study mounted at /study and writes small validation CSV files to /out. It uses metafor::rma.uni(method="PM", test="adhoc") and rma.mh, independently of the Python implementation. A host needs R and metafor only to repeat this additional check; neither is required by the default reproduction or the core claim. The validation result and independent outputs are under results/; they are validation records, not additional clinical observations. The largest numerical difference is attributable to root-finding precision at an extremely wide two-trial stroke interval bound. Displayed values agree.