# K9b – jin2023 (opus-S3 rows 001–055) vs supplementary file

Source: /home/claude/drive/supplements/jin2023_supplementary.txt (S2 all CRC, S3 colon/rectal, S4 by sex). The txt was cross-checked against the source PDF (pdftotext) for every meat item: identical.
Method: all 55 rows compared by script (point, CI, outcome, sex, source table); FDR P and P-heterogeneity in notes spot-read.

## Discrepancies
| estimate_id | field | in database | in paper | severity |
|---|---|---|---|---|
| 004, 009, 014, 019, 024, 029, 034, 039, 044, 049, 054 (women) | n_cases / n_in_category | empty | not printed per sex, but derivable from Table S1: 1,466 − 831 = 635 cases; 118,210 − 52,664 = 65,546 women | minor |
| 016–020 | exposure | organ | WebQ item "Other meat, offal" mixes offal with other meats; already flagged in notes | minor |

## Per-SD rows
- All rows per 1-SD of daily intake (footnote: dietary factors entered as standardized continuous variables), dose_unit per_sd correct. The SD in g/day is not printed anywhere in the supplement, so the rows cannot be rescaled. Not usable for per-gram pooling; is_main_model=yes for whole items is formally correct, but these rows must not enter g-based pools.
- Subtype rows (breaded chicken, oily fish, white fish/tuna, breaded fish, shellfish) is_main_model=no. Reasonable.

## Checked and confirmed
- All 55 HR/CI values match S2/S3/S4 (beef, lamb, pork, other meat/offal, processed meat, poultry, breaded chicken, oily fish, white fish and tinned tuna, breaded fish, shellfish × CRC, colon, rectum, women, men).
- S4 column order (Female first, then Male) handled correctly.
- Men n 831/52,664 (Table S1) correct.
- adjustments jin2023_m1 matches the S2–S4 footnote (age, sex, TDI, education, ethnicity, family history, aspirin, screening, diabetes, BMI, PA, pack-years, energy; no alcohol).

All 55 rows confirmed numerically.

## Missing estimates
- Optional: "Sushi" (S2 1.03, S3/S4 also printed) is a fish-containing item that was not coded. A judgement call; not required.
- No red-meat aggregate exists in the paper (correct that none was coded).

## Paper-internal inconsistencies
- Colon 842 + rectum 359 = 1,201 of 1,466 CRC (265 unexplained); already noted.