# WCRF/AICR 1997 – red meat and colorectal cancer: evidence table (cohort studies)

Written 2026-10-30. Data file: `wcrf1997_cohorts.csv` (9 rows: 7 cohorts + the 2 case-control studies the card names).

## Source and its limits

- **The report itself was not read.** Internet Archive item `foodnutritionp00amer` is lending-only. On 2026-10-30 the full text (`_djvu.txt`, `_hocr_searchtext.txt.gz`) returned 401, the search-inside endpoint (`fulltext/inside.php`) returned "Item not available", and WebFetch of the details page returned only metadata (678 pp.). **Table 4.10.10 could not be confirmed.**
- All report values below therefore come from the card `/home/claude/drive/Historikk 1965-2015/kort/25_WCRF1997.md` (IA "Search inside" snippets, read 2026-10-29), section 3 "Kunnskapsgrunnlaget" table and "Kildesitater". `kontroll/fase0_inventar.md` §1a (lines 23–35) copies the same values.
- Page numbers are the book reader's page numbers. The card says they probably match the printed pages but this was not checked.
- The card does not record the exposure label for each row of Table 4.10.10. Where the label is unknown, the CSV says "label not captured".

## The report's evidence and grading

- Section 4.10.2.6 (meat), Table 4.10.10 (cohorts) and 4.10.11 (case-control), pp. 243–246.
- "Seven cohort studies have examined meat intake" (p. 243). The report pools nothing: the cohort evidence is narrative.
- Grading: «red meat probably increases risk and processed meat possibly increases risk of colorectal cancer» (p. 246). Summary p. 217: «diets high in red meat, probably increase the risk».
- The report's own count (p. 243, per the card): of 16 risk estimates in Table 4.10.10, 8 were above 1.5 or significant, 8 were between 0.75 and 1.5, none below 0.75. These 16 include processed meat and adenomas.

## Table

| # | Study (report) | Cohort | Report's estimate / verdict (p.) | Database | Discrepancy |
|---|---|---|---|---|---|
| 1 | Phillips 1985 | Adventist mortality | "meat" 0.9 (0.6–1.5), unassociated (p. 243) | **not_in_db** (related: phillips1975, singh1998) | cannot check. Total meat, fatal cases |
| 2 | Willett 1990 | NHS | 1.77 (1.09–2.88); main dish 2.49 (1.24–5.03) | willett1990_028 1.77 (1.09–2.88); willett1990_005 2.49 (1.24–5.03), colon, women | none in numbers. Only age (+energy) adjusted; daily main-dish category n=16 |
| 3 | Thun 1992 | CPS-II (mortality) | "no difference"; M 1.21, W 1.05 (p. 243) | **not_in_db** (related: chao2005, CPS-II Nutrition subcohort) | cannot check. The report's 1.21 (men) is called "no difference" |
| 4 | Giovannucci 1994 | HPFS | 1.71 (1.15–2.55) | giovannucci1994_005 1.71 (1.15–2.55), colon, men | none in numbers. Age-adjusted only. "Red meat" includes processed items |
| 5 | Goldbohm 1994 | NLCS | "no increase" | goldbohm1994_015 total fresh meat 0.84 (0.51–1.37); beef/15 g 0.96 (0.87–1.06); minced beef+pork/15 g 0.91 (0.80–1.04), colon | none |
| 6 | Knekt 1994 | Finnish cohort (fried meat) | "no increase" | **not_in_db** (related: jarvinen2001, same survey) | cannot check |
| 7 | Bostick 1994 | IWHS | "no increase in risk with higher intakes of meat" (p. 243) | bostick1994_054 "red meat" 1.04 (0.62–1.76); bostick1994_008 main dish 1.21 (0.75–1.96), colon, women | minor: main dish point estimate 1.21, not significant |
| 8 | Manousos 1983 (c-c) | Athens | among 14 c-c studies with ≥1 significant raised OR (p. 244) | not_in_db (text on disk) | not checked |
| 9 | Miller 1983 (c-c) | Toronto | same (p. 244) | not_in_db (text on disk) | not checked |

## What the table shows

- **7 cohort studies.** 5 are in the database (Willett 1990, Giovannucci 1994, Goldbohm 1994, Bostick 1994, and since 30.10.2026 Thun 1992) and 2 are not (Phillips 1985, Knekt 1994).
- **Where we can check, every number the report prints matches the primary paper.** We found no transcription error; 3 of the 7 rows print numbers we can check (Willett, Giovannucci, Thun – men 1.21, women 1.05 match thun1992_005/_015).
- The two positive cohorts (NHS, HPFS) come from the same Harvard group, cover colon only, and are adjusted only for age (and energy). The report does not mention this; the card snippets do not show whether the full table does.
- Two of the null verdicts hide point estimates above 1: Thun 1992 for men (1.21, from the report itself) and Bostick 1994 main-dish red meat (1.21, from the paper). Neither is significant, so a "null" reading is defensible. The report's words are still more categorical than the numbers.
- **Case-control:** the report counts 26 (23 cancer, 3 polyp; p. 243). Only Manousos 1983 and Miller 1983 are named on the card. The other 24 cannot be identified without Table 4.10.11.