# IARC Monographs vol. 114 – red meat and colorectal cancer: evidence table (cohort studies)

Written 2026-10-30. Source: `/home/claude/drive/Historikk 1965-2015/Historikk/txt/36_IARC114.txt` (24,574 lines). Line numbers below refer to that file. Page numbers are the printed monograph pages. Page markers sit in the text: odd pages end with the page number, even pages start with it.

Files:
- `iarc114_cohorts.csv`: 37 rows. 23 rows are tabulated in Table 2.2.1 and 14 studies are discussed only in the text.
- `iarc114_table_values_check.csv`: every RR (CI) printed in Table 2.2.1 (178 values), each matched against `estimates.csv` for the same study_id on identical point and CI.

## Where things are

| Item | Page | Lines |
|---|---|---|
| 2.2.1 cohort studies, inclusion rules | p. 111 | 5651–5677 |
| (a) Red meat, text | pp. 111–117 | 5660–6000 |
| (c) Red and processed meat combined, text | pp. 123–124 | 6269–6345 |
| 2.2.3 Meta-analyses | pp. 142–144 | 7203–7336 |
| **Table 2.2.1** Cohort studies: red meat or red & processed meat combined | pp. 145–160 | 7341–8049 |
| 5.2.1 Summary, colorectum | pp. 492–493 | 23950–24010 |
| 6. Evaluation | p. 501 | 24321–24349 |

**Table 2.2.1 has no "comments" column.** Its columns are: reference/location/period/design; population and exposure assessment; organ site; exposure category; exposed cases; risk estimate (95% CI); covariates. In the CSV, the `iarc_wg_comment` column therefore holds the Working Group's bracketed notes from the text, with page and line.

## Table 2.2.1: summary of the 23 rows

| # | IARC row | Page | IARC: comparison → RR (95% CI) | DB | Status / discrepancy |
|---|---|---|---|---|---|
| 1 | Kato 1997, NYUWHS | 145 | Q4 v Q1 1.23 (0.68–2.22) | kato1997_004 | match |
| 2 | Chen 1998, PHS (nested c-c) | 145 | red/proc >1 v ≤0.5 serv/d 1.17 (0.68–2.02) | chen1998_003 | match. IARC internal: covariates column lists BMI, WG note says BMI was not controlled for |
| 3 | Singh & Fraser 1998, AHS | 145 | ≥1/wk v never 1.90 (1.16–3.11) | singh1998_012 (non-main); main singh1998_006 1.41 (0.90–2.21) | **major**: organ site printed "colon and rectum" (paper is colon only). RRs are from the restricted subset with white meat <1/wk (paper Table 4) but are printed with full-cohort case counts and covariates (Table 3) |
| 4 | Pietinen 1999, ATBC | 145 | Q4 v Q1 0.8 (0.5–1.2) | pietinen1999_008 | match. Q2 printed 0.6 (0.4–**1.1**), paper 0.6 (0.4–1.0) |
| 5 | Järvinen 2001, FMCHES | 146 | red/proc Q4 v Q1 1.50 (0.77–2.94) | jarvinen2001_016 | match (9/9 values) |
| 6 | Tiemersma 2002, Dutch Monitoring Project (nested c-c) | 147–148 | ≥5 v ≤3/wk 1.6 (0.9–2.9) | tiemersma2002_012 | match (21/21) |
| 7 | Flood 2003, BCDDP | 148 | red/proc Q5 v Q1 1.04 (0.77–1.41) | flood2003_020 (non-main); main _015 1.10 (0.83–1.45) | model choice (IARC uses the total-meat-adjusted model). Text calls the quintile medians cut-points |
| 8 | English 2004, MCCS | 148–149 | ≥6.5 v <3/wk 1.40 (1.00–1.90); rectum 2.30 (1.20–4.20) | english2004_004/_014 | match (14/14) |
| 9 | Chao 2005, CPS-II | 149–150 | red/proc colon Q5: M 1.30 (0.93–1.81), W 0.98 (0.68–1.40); rectum 1.71 (1.15–2.52) | chao2005_015/_030/_120 | RRs match (24/24). Unit printed **g/day** for medians that are **g/week**. Men Q5 cases 125 vs paper 124 |
| 10 | Larsson 2005a, SMC | 151 | beef+pork ≥4 v <2/wk 1.22 (0.98–1.53); distal 1.99 (1.26–3.14) | larsson2005_036/_044 | match |
| 11 | Norat 2005, EPIC | 152–153 | >80 v <10 g/d 1.17 (0.92–1.49); per 100 g 1.21 (1.02–1.43); calibrated 1.49 (0.91–2.43) | norat2005_010/_031/_032 | match (30/30). Text p. 114 (line 5840) prints pork CI "0.95–0.48", paper 0.95–1.48 |
| 12 | Berndt 2006, CLUE II | 154 | red/proc ≥86.3 v <44 g/d 1.32 (0.86–2.02) | **not_in_db** | IARC internal: 272 vs 250 cases; covariates differ between table and text |
| 13 | Oba 2006, Takayama | 154 | T3 v T1 colon M 1.03 (0.64–1.66), W 0.79 (0.49–1.28) | oba2006_018/_024 | match. **superseded_by wada2017** (M 1.44 (1.10–1.89), W 1.07) |
| 14 | Kabat 2007, CNBSS | 155 | red/proc ≥40.3 v <14.25 g/d 1.12 (0.86–1.46); rectum 1.95 (1.21–3.16) | kabat2007_030/_025 | match (12/12) |
| 15 | Butler 2008b, SCHS | 155 | Q4 v Q1 1.01 (0.82–1.26) | butler2008_001 | match. **DB issue**: no butler2008 row has is_main_model=yes |
| 16 | Sørensen 2008, DCH (case-cohort) | 156 | per 25 g/d 1.03 (0.97–1.09) | **not_in_db** | "exposed cases 105" (379 cases) |
| 17 | Andersen 2009, DCH (case-cohort) | 156 | per 25 g/d 1.02 (0.94–1.12) | **not_in_db** | the same cohort is tabulated twice, against IARC's own one-publication-per-cohort rule (line 5658) |
| 18 | Lee 2009, SWHS | 156–157 | ≥67 v <24 g/d 0.8 (0.6–1.1) | lee2009_020 | match |
| 19 | Cross 2010, NIH-AARP | 157–158 | red/proc Q5 v Q1 1.24 (1.09–1.42); per 100 g 1.23 (1.10–1.36) | cross2010_005/_049 | match (20/20). **superseded_by zouiouich2025** |
| 20 | Ollberding 2012, MEC | 158 | unprocessed Q5 v Q1 0.98 (0.87–1.10) | ollberding2012_070 | match |
| 21 | Figueiredo 2014, GECCO/CCFR | 159 | per quartile 1.15 (no CI) | **not_in_db** | pooled case-control; out of db scope |
| 22 | Ananthakrishnan 2015, GECCO/CCFR | 159 | red/proc Q4 v Q1 1.29 (1.15–1.44) | **not_in_db** | pooled case-control. IARC's text says the association came only from the retrospective studies |
| 23 | Bernstein 2015, NHS+HPFS | 160 | per serving/d, cumulative: 0.99 (0.87–1.13); distal 0.75 (0.68–0.82) | bernstein2015_024/_096 (non-main) | numbers match. Rectum rows labelled "**Processed** red meat", but the values (1.05; 1.14) are unprocessed (bernstein2015_127/_132). Distal CI is the paper's own suspect value (rettelogg R24) |

Value-level check (`iarc114_table_values_check.csv`): of the 178 RR/CI triples in Table 2.2.1, 166 have an identical row in the database. 11 belong to studies not in the database. 1 has no match: Pietinen Q2.

## What IARC counted

- Text: "Fourteen cohort studies and two cohort consortia provided informative data" on red meat (p. 111, line 5660). The 14 are not listed. Reconstructed from the order of section (a), they are: NYUWHS, Dutch Monitoring Project, Takayama, AHS, ATBC, IWHS, DCH, SWHS, MCCS, SMC, SCHS, EPIC, NHS/HPFS and MEC. The two consortia are UKDCC (Spencer 2010) and GECCO/CCFR (Kantor 2014 / Figueiredo 2014 / Ananthakrishnan 2015).
- Table 2.2.1 does **not** contain IWHS (Bostick 1994) or UKDCC (Spencer 2010), although the text counts both. It does contain 7 studies of red and processed meat combined (Chen, Järvinen, Flood, Chao, Berndt, Kabat, Cross), which IARC itself calls "less informative" (line 6271–6276).
- 5.2.1 (p. 493, lines 23967–23974): "Positive associations … were observed in seven studies, including … EPIC …, SMC, and MCCS." The seven are not named. One reading in which the highest-vs-lowest point estimate is above 1 gives exactly seven: Kato (1.23), Tiemersma (1.6), Singh (1.90), Bostick (1.21), English (1.4), Larsson (1.22) and Norat (1.17). In that reading only Singh's RR is nominally significant, and only in the restricted subset. Norat is significant per 100 g, and English's CI touches 1.0. **This is a judgment call** (see NOTES).

## Meta-analyses IARC relied on (section 2.2.3, pp. 142–144, lines 7203–7336)

IARC ran no meta-analysis of its own. Numbers as IARC prints them, checked against the source where it is on disk:

| Meta-analysis | IARC's numbers (red meat unless stated) | Check |
|---|---|---|
| Chan 2011 (PLoS ONE 6:e20456; PMID 21674008) | HvL CRC 1.10 (1.00–1.21), I² 22%, 12 studies; colon 1.18 (1.04–1.35), 10; rectum 1.14 (0.83–1.56), 7. Per 100 g/d: CRC 1.17 (1.05–1.31), **8 studies**; colon 1.17 (1.02–1.33), 10; rectum 1.18 (0.98–1.42), 7. Processed HvL CRC 1.17 (1.09–1.25), 13; per 50 g 1.18 (1.10–1.28), 9 | all match Chan Table (`cup/chan.txt` lines 187–218). **But** 5.2.1 (p. 493) and Bouvard 2015 say "a meta-analysis including data from **10** cohort studies" for 1.17 per 100 g. Chan has 8 for CRC (10 is the colon n) |
| Alexander 2015 (J Am Coll Nutr 34:521–43; PMID 25941850) | 27 cohorts; red+processed HvL 1.11 (1.03–1.19), I² 33.6%; unprocessed red HvL 1.05 (0.98–1.12), I² 8.4%, 17 studies; colon 1.11 (1.04–1.18), 16; rectum 1.17 (0.99–1.39), 13; unprocessed colon 1.06 (0.97–1.16), 11; rectum 1.03 (0.88–1.21), 10; dose: **1.02 (1.00–1.14) per 1 serving/day**, 10 studies; 1.05 (0.97–1.13) per 70 g/d, 13 | Alexander Table (`Kilder/tidligere_kritikk/alexander2015_jacn.txt` lines 722–724) prints **1.02 (1.00–1.04) per serving per *week***. IARC has the wrong upper CI and the wrong unit. IARC also leaves out I² = 88.6% for the 70 g analysis |
| Alexander 2010 (processed) | HvL 1.16 (1.10–1.23), 20 studies; per 30 g 1.10 (1.05–1.15) | not checked |
| Pham 2014 (6 Japanese cohorts) | red meat HvL colon 1.20 (1.00–1.44), 4; rectum 0.95 (0.71–1.28), 3; with 13 c-c: CRC 1.16 (1.001–1.34), colon 1.21 (1.03–1.43) | not on disk |
| Carr 2016 (meat subtypes) | beef HvL CRC 1.11 (1.01–1.22), colon 1.24 (1.07–1.44), rectum 0.95 (0.78–1.16); lamb 1.24 (1.08–1.44); pork 1.07 (0.90–1.27) | not on disk |
| Qiao & Feng 2013 (haem iron) | HvL 1.14 (1.04–1.24); per 1 mg/d 1.11 (1.03–1.18) | not on disk |
| Earlier, not quantified | Sandhu 2001, Norat 2002, Larsson 2006, Huxley 2009 ("moderate but significant increase", line 7208–7212) | – |

WG note (lines 7290–7310): rescaling servings to grams (120 g red meat) "may have increased the measurement error … It is possible that the rescaling … may have attenuated the observed association."

## IARC's conclusions on red meat (verbatim)

- **6.1 Cancer in humans** (p. 501, lines 24324–24327): «There is *limited evidence* in humans for the carcinogenicity of consumption of red meat. Positive associations have been observed between consumption of red meat and cancers of the colorectum, pancreas, and prostate.»
- **6.3 Overall evaluation** (p. 501, lines 24346–24348): «Consumption of red meat is probably carcinogenic to humans (Group 2A).»
- **6.4 Rationale** (p. 501, lines 24323–24333, right column): «… the substantial epidemiological data showing a positive association between consumption of red meat and cancer of the colorectum, and the strong mechanistic evidence. Taken as a whole, this evidence led the Working Group to classify red meat as probably carcinogenic to humans (Group 2A).»
- **5.2.1** (p. 493, lines 23999–24002, right column): «chance, bias, or confounding could not be ruled out for consumption of red meat, as no association was observed in several of the larger studies.»
- **Bouvard 2015** (Lancet Oncol 16:1599–600; `/home/claude/drive/historisk/bouvard2015.txt` lines 83–145): 14 cohort studies, positive in half. Of 15 informative case-control studies, 7 were positive. Chan 2011 is cited as "ten cohort studies", 17% per 100 g (1.05–1.31). «Chance, bias, and confounding could not be ruled out with the same degree of confidence for the data on red meat consumption, since no clear association was seen in several of the high quality studies and residual confounding from other diet and lifestyle risk is difficult to exclude» (lines 133–141). «The Working Group concluded that there is limited evidence in human beings for the carcinogenicity of the consumption of red meat» (lines 142–145).
- Note: the 2A rating rests on limited human evidence plus "strong mechanistic evidence". Animal evidence is "inadequate" (6.2).