# K2 control – summary (30.10.2026)
Method: table numbers compared by script; definitions, models, flags and text estimates checked by reading. Database not edited.

| study | rows checked | numbers confirmed | critical / major / minor | missing | verdict |
|---|---|---|---|---|---|
| english2004 | 81 | 81 | 0 / 1 / 2 | 0 | Fit for use after the `dose_unit` fix |
| tiemersma2002 | 107 | 107 | 0 / 0 / 1 | 0 | Fit for use |
| jarvinen2001 | 42 | 42 | 0 / 0 / 3 | 0 | Fit for use; keep `red_undefined` / `unclear` |
| pietinen1999 | 40 | 40 | 0 / 0 / 0 | 16 rows | Fit for use; add the ratio rows |
| takachi2011 | 259 | 259 | 0 / 1 / 5 | 0 | Fit for use after filling the case fields in studies.csv |
| **Total** | **529** | **529** | **0 / 2 / 11** | **16** | |

Major: (1) english2004 continuous rows: dose_unit should be per_serving_week (paper: per one time per week). (2) takachi2011 studies.csv cases_colorectal/colon/rectum empty; paper: 1145/788/357.
Missing: pietinen1999 Table 5 ratio meat/(poultry+fish) (8 rows) and fried-meat frequency (8 rows).
Minor: adj_stratified_center harmonisation (Takachi yes vs Järvinen/Tiemersma no); Takachi adj_fish = dried/salted fish only; Takachi m1 adj_energy should be yes (residual method); Tiemersma uses exposure_includes_processed=partial; English ratio rows under exposure red_unprocessed; Järvinen poultry dose_unit empty (ever_never); asymmetry flags inconsistently applied.
Paper inconsistencies: Järvinen Q2 1.06 (0.67–2.01) misprint; Tiemersma abstract 0.5 (0.2–1.07) matches no table; English chicken bounds overlap; Takachi proximal+distal < colon, baseline years differ, corrected Q5 medians 120/105 vs 102/93.