{
  "fame_subgroup_model_discrepancy": "Appendix S9 HR for stable primary composite is 1.39 (0.97-2.00); main Figure 4 non-ACS is 1.37 (0.96-1.97). Use S9 consistently with its counts, show Figure 4 as a source sensitivity; do not silently treat the estimates as identical. Main overall percentages are KM; S9 footnote explicitly calls percentages crude.",
  "hrev_composite": "Use only separately reported marginal outcomes in quantitative synthesis. The reported MACCE count equals the sum of its component counts in both arms; the individual overlap is unavailable. Do not derive a new composite from the marginals or assert that equal sums alone prove an error.",
  "mortality_missing": "FAME 3 gives overall discontinuation counts, not stable-specific missingness. Crude fractions use randomized denominators and are not KM fixed-horizon risks. HREVS uses complete-case denominators and loses seven patients in the relevant arms.",
  "best_abstract_only": "Main BEST extended full text/supplement could not be retrieved; main abstract and a full primary diabetes report were read. Stable-only outcomes not located does not assert that no such dataset exists.",
  "prior_meta_overlap": "Dillen is the FAME 3 true-bifurcation subgroup, not a separate RCT. Di Gioia is observational. Tsai et al. 2026 already identified these design errors; do not claim to discover them. Quantitative reanalysis is a reproduction/extension of that published critique. Prior synthesis Fig 2 Dillen denominators are 350/303 (653), unlike Table 1 full-cohort 757/743.",
  "scope": "Contemporary DES refers to stent platform, not every component of 2026 PCI practice. These trials do not establish effects in left-main disease or surgery-ineligible patients, and compare two revascularization strategies, not medical therapy alone."
}
