# WCRF/AICR Second Expert Report 2007 (Fig 4.3.2: red meat 'per time/week', cohort studies — the values are per time/day, see deviations)

All pooled estimates: DerSimonian–Laird random effects; weights are random-effects weights (%). Leave-one-out after each run.

## Per time/day – published per-study values (Fig 4.3.2, 9 rows / 7 studies; axis labelled per time/week)

Published: 1.43 (1.05–1.94). **Pooled: 1.43 (1.05–1.94)** (fixed 1.15 (1.00–1.32); I² 58 %; k = 9). **Reproduced.**

| Study | RR (95 % CI) | Weight % | Source / choice |
|---|---|---|---|
| willett1990 | 1.81 (1.17–2.80) | 17.3 | Fig 4.3.2 p121; 10 PDF excerpt (p121) confirms values and the axis |
| bostick1994 | 0.96 (0.80–1.14) | 24.4 | Fig 4.3.2 p121; 10 PDF excerpt (p121) confirms values and the axis |
| giovannucci1994 | 2.20 (1.24–3.91) | 13.7 | Fig 4.3.2 p121; 10 PDF excerpt (p121) confirms values and the axis |
| singh1998 | 4.51 (0.38–53.27) | 1.5 | Fig 4.3.2 p121; 10 PDF excerpt (p121) confirms values and the axis |
| chen1998_rapid | 2.57 (0.78–8.84) | 5.1 | Fig 4.3.2 p121; 10 PDF excerpt (p121) confirms values and the axis |
| chen1998_slow | 0.88 (0.45–1.75) | 11.5 | Fig 4.3.2 p121; 10 PDF excerpt (p121) confirms values and the axis |
| tiemersma2002_men | 3.44 (0.83–14.18) | 4.0 | Fig 4.3.2 p121; 10 PDF excerpt (p121) confirms values and the axis |
| tiemersma2002_women | 1.42 (0.22–9.00) | 2.5 | Fig 4.3.2 p121; 10 PDF excerpt (p121) confirms values and the axis |
| english2004 | 1.23 (0.88–1.73) | 20.1 | Fig 4.3.2 p121; 10 PDF excerpt (p121) confirms values and the axis |

## Per time/day – corrected: duplicate

**Pooled: 1.41 (1.04–1.90)** (fixed 1.15 (1.00–1.32); I² 58 %; k = 8). 

| Study | RR (95 % CI) | Weight % | Source / choice |
|---|---|---|---|
| willett1990 | 1.81 (1.17–2.80) | 17.8 | Fig 4.3.2 p121; 10 PDF excerpt (p121) confirms values and the axis |
| bostick1994 | 0.96 (0.80–1.14) | 26.1 | Fig 4.3.2 p121; 10 PDF excerpt (p121) confirms values and the axis |
| giovannucci1994 | 2.20 (1.24–3.91) | 14.0 | Fig 4.3.2 p121; 10 PDF excerpt (p121) confirms values and the axis |
| singh1998 | 4.51 (0.38–53.27) | 1.4 | Fig 4.3.2 p121; 10 PDF excerpt (p121) confirms values and the axis |
| chen1998_rapid | 1.14 (0.63–2.05) | 13.5 | duplicate: NAT2 strata combined (fixed) |
| tiemersma2002_men | 3.44 (0.83–14.18) | 3.9 | Fig 4.3.2 p121; 10 PDF excerpt (p121) confirms values and the axis |
| tiemersma2002_women | 1.42 (0.22–9.00) | 2.4 | Fig 4.3.2 p121; 10 PDF excerpt (p121) confirms values and the axis |
| english2004 | 1.23 (0.88–1.73) | 21.0 | Fig 4.3.2 p121; 10 PDF excerpt (p121) confirms values and the axis |

Leave-one-out: without willett1990: 1.31 (0.96–1.79); without bostick1994: 1.54 (1.21–1.97); without giovannucci1994: 1.28 (0.97–1.70); without singh1998: 1.39 (1.03–1.87); without chen1998_rapid: 1.48 (1.05–2.10); without tiemersma2002_men: 1.35 (1.01–1.81); without tiemersma2002_women: 1.42 (1.04–1.94); without english2004: 1.52 (1.02–2.26)

## Per time/day – corrected: inclusion

**Pooled: 1.14 (0.88–1.49)** (fixed 1.04 (0.90–1.21); I² 28 %; k = 7). 

| Study | RR (95 % CI) | Weight % | Source / choice |
|---|---|---|---|
| bostick1994 | 0.96 (0.80–1.14) | 47.1 | Fig 4.3.2 p121; 10 PDF excerpt (p121) confirms values and the axis |
| singh1998 | 4.51 (0.38–53.27) | 1.1 | Fig 4.3.2 p121; 10 PDF excerpt (p121) confirms values and the axis |
| chen1998_rapid | 2.57 (0.78–8.84) | 4.4 | Fig 4.3.2 p121; 10 PDF excerpt (p121) confirms values and the axis |
| chen1998_slow | 0.88 (0.45–1.75) | 12.0 | Fig 4.3.2 p121; 10 PDF excerpt (p121) confirms values and the axis |
| tiemersma2002_men | 3.44 (0.83–14.18) | 3.3 | Fig 4.3.2 p121; 10 PDF excerpt (p121) confirms values and the axis |
| tiemersma2002_women | 1.42 (0.22–9.00) | 2.0 | Fig 4.3.2 p121; 10 PDF excerpt (p121) confirms values and the axis |
| english2004 | 1.23 (0.88–1.73) | 30.2 | Fig 4.3.2 p121; 10 PDF excerpt (p121) confirms values and the axis |

## Per time/day – corrected: all classes

**Pooled: 1.10 (0.89–1.36)** (fixed 1.04 (0.90–1.21); I² 17 %; k = 6). 

| Study | RR (95 % CI) | Weight % | Source / choice |
|---|---|---|---|
| bostick1994 | 0.96 (0.80–1.14) | 56.1 | Fig 4.3.2 p121; 10 PDF excerpt (p121) confirms values and the axis |
| singh1998 | 4.51 (0.38–53.27) | 0.8 | Fig 4.3.2 p121; 10 PDF excerpt (p121) confirms values and the axis |
| chen1998_rapid | 1.14 (0.63–2.05) | 11.5 | duplicate: NAT2 strata combined (fixed) |
| tiemersma2002_men | 3.44 (0.83–14.18) | 2.2 | Fig 4.3.2 p121; 10 PDF excerpt (p121) confirms values and the axis |
| tiemersma2002_women | 1.42 (0.22–9.00) | 1.3 | Fig 4.3.2 p121; 10 PDF excerpt (p121) confirms values and the axis |
| english2004 | 1.23 (0.88–1.73) | 28.0 | Fig 4.3.2 p121; 10 PDF excerpt (p121) confirms values and the axis |

Leave-one-out: without bostick1994: 1.28 (0.97–1.70); without singh1998: 1.07 (0.88–1.31); without chen1998_rapid: 1.14 (0.85–1.53); without tiemersma2002_men: 1.03 (0.88–1.20); without tiemersma2002_women: 1.12 (0.87–1.45); without english2004: 1.10 (0.80–1.53)

## Per time/day – corrected + follow-ups

**Pooled: 1.02 (0.92–1.13)** (fixed 1.02 (0.92–1.12); I² 3 %; k = 7). 

Willett 1990 and Giovannucci 1994 replaced by the same cohorts' 2015 follow-up (2,731 cases) on the per-serving/day scale (1.2 × log RR per 100 g).

| Study | RR (95 % CI) | Weight % | Source / choice |
|---|---|---|---|
| bostick1994 | 0.96 (0.80–1.14) | 32.3 | Fig 4.3.2 p121; 10 PDF excerpt (p121) confirms values and the axis |
| singh1998 | 4.51 (0.38–53.27) | 0.2 | Fig 4.3.2 p121; 10 PDF excerpt (p121) confirms values and the axis |
| chen1998_rapid | 1.14 (0.63–2.05) | 3.1 | duplicate: NAT2 strata combined (fixed) |
| tiemersma2002_men | 3.44 (0.83–14.18) | 0.5 | Fig 4.3.2 p121; 10 PDF excerpt (p121) confirms values and the axis |
| tiemersma2002_women | 1.42 (0.22–9.00) | 0.3 | Fig 4.3.2 p121; 10 PDF excerpt (p121) confirms values and the axis |
| english2004 | 1.23 (0.88–1.73) | 9.5 | Fig 4.3.2 p121; 10 PDF excerpt (p121) confirms values and the axis |
| willett1990 | 1.00 (0.87–1.13) | 54.1 | follow-up: NHS+HPFS → Bernstein 2015 Suppl. S5, unprocessed per 100 g → per 120 g/day (one serving/day) |

Leave-one-out: without bostick1994: 1.09 (0.91–1.29); without singh1998: 1.02 (0.92–1.12); without chen1998_rapid: 1.03 (0.90–1.18); without tiemersma2002_men: 1.01 (0.92–1.12); without tiemersma2002_women: 1.03 (0.91–1.18); without english2004: 1.00 (0.90–1.11); without willett1990: 1.10 (0.89–1.36)

## Per 100 g/day – Fig 4.3.3 (3 studies; Larsson's upper limit read off the figure as 1.94)

Published: 1.29 (1.05–1.59). **Pooled: 1.29 (1.04–1.60)** (fixed 1.29 (1.05–1.59); I² 6 %; k = 3). **Reproduced.**

The figure's summary 1.29 (1.05–1.59) is the fixed-effect pool; the text's 1.29 (1.04–1.60) is the random-effects pool. SLR 2010 Table 36's 1.29 (0.94–1.78) is not reproduced.

| Study | RR (95 % CI) | Weight % | Source / choice |
|---|---|---|---|
| pietinen1999 | 1.05 (0.74–1.48) | 36.1 | Fig 4.3.3 p121 as printed |
| norat2005 | 1.49 (0.91–2.43) | 18.6 | Fig 4.3.3 p121 as printed (EPIC calibrated estimate) |
| larsson2005 | 1.43 (1.05–1.94) | 45.3 | Fig 4.3.3 p121: printed '1.43 (1.05–0.95)'; upper limit 1.94 by calibrated pixel reading of the author-supplied PDF |

Leave-one-out: without pietinen1999: 1.45 (1.12–1.88); without norat2005: 1.24 (0.92–1.68); without larsson2005: 1.20 (0.86–1.67)

## Per 100 g/day – corrected: errors only (exposure; model choices as published)

**Pooled: 1.22 (0.96–1.56)** (fixed 1.22 (0.96–1.56); I² 0 %; k = 3). 

R105: Pietinen and Larsson replaced by the papers' red meat rows without processed meat; Norat's calibrated EPIC estimate kept as published. Fixed and random effects coincide (tau² = 0).

| Study | RR (95 % CI) | Weight % | Source / choice |
|---|---|---|---|
| pietinen1999 | 0.83 (0.43–1.61) | 13.7 | exposure: unprocessed red meat, GL (R105) |
| norat2005 | 1.49 (0.91–2.43) | 24.7 | Fig 4.3.3 p121 as printed (EPIC calibrated estimate) |
| larsson2005 | 1.23 (0.90–1.67) | 61.6 | exposure: unprocessed red meat as main dish, GL with 120 g/serving (R105) |

Leave-one-out: without pietinen1999: 1.30 (1.00–1.69); without norat2005: 1.13 (0.84–1.53); without larsson2005: 1.16 (0.66–2.04)

## Per 100 g/day – corrected: all classes

**Pooled: 1.19 (1.03–1.38)** (fixed 1.19 (1.03–1.38); I² 0 %; k = 3). 

R105 plus Norat 2005 observed instead of calibrated.

| Study | RR (95 % CI) | Weight % | Source / choice |
|---|---|---|---|
| pietinen1999 | 0.83 (0.43–1.61) | 4.8 | exposure: unprocessed red meat, GL (R105) |
| norat2005 | 1.21 (1.02–1.43) | 73.5 | model: observed estimate |
| larsson2005 | 1.23 (0.90–1.67) | 21.7 | exposure: unprocessed red meat as main dish, GL with 120 g/serving (R105) |

Leave-one-out: without pietinen1999: 1.21 (1.05–1.41); without norat2005: 1.13 (0.84–1.53); without larsson2005: 1.16 (0.91–1.47)

## Deviations (synthesis vs primary paper)

| Outcome | Study | Class | Deviation | Page | Alternative used |
|---|---|---|---|---|---|
| colorectal/colon | english2004 | transcription | Fig 4.3.2 is labelled 'per time/week', but English 2004 prints 1.03 (0.98–1.08) per serving/week and the figure's 1.23 (0.88–1.73) equals 1.03^7 = 1.23 (0.87–1.71); SLR 2010 calls the same analysis 'per time/day' (p156, Table 36). The increment is per time/day; Willett's 1.81 and Giovannucci's 2.20 are likewise implausible as weekly slopes | SER Fig 4.3.2 p121; SLR 2010 p156–157; English 2004 Table 3 | none available (documented only) |
| colorectal/colon | chen1998_rapid | duplicate | Chen 1998 entered twice, split by NAT2 genotype (rapid / slow acetylators); one study | Fig 4.3.2 | 1.14 (0.63–2.05) – NAT2 strata combined (fixed) |
| colorectal/colon | chen1998_slow | duplicate | see rapid row | Fig 4.3.2 | removed |
| colorectal/colon | chen1998_rapid | inclusion | Physicians' Health Study nested case-control | Chen 1998 | none available (documented only) |
| colorectal/colon | willett1990 | inclusion | NHS 1980–86 (150 colon cases); the report cites Wei 2004 (NHS+HPFS to 2000, 1,139 colon + 339 rectal cases, ref 19) but used it only in the HvL plot. Wei's continuous estimate cannot substitute: its printed unit (per week) contradicts the paper's Methods (per day) and the row is 'disputed' (rettelogg R15) | SER refs 9, 19; SLR 2010 Table 37 | removed |
| colorectal/colon | giovannucci1994 | inclusion | HPFS 1986–92 (205 colon cases); superseded by Wei 2004 (same cohort), which the report cites | SER refs 10, 19 | removed |
| colorectal/colon | singh1998 | dose | 4.51 (0.38–53.27) per time/week from a never / <1 / ≥1 time per week scale; the top category is open (≥1/week) | Singh 1998 Table 3 | none available (documented only) |
| colorectal/colon | bostick1994 | transcription | 0.96 (0.80–1.14) per time/week has a CI four to five times narrower (log width 0.35) than a GL slope from Bostick's four categories, three above the reference (212 cases) can give (0.94; 0.43–2.04 per 100 g); cannot be checked without the 2005 SLR (called the 2006 SLR in the protocol) | Fig 4.3.2; Bostick 1994 Table 3 | none available (documented only) |
| colorectal (Fig 4.3.3, per 100 g/day) | pietinen1999 | exposure | Fig 4.3.3's 1.05 (0.74–1.48) is ATBC red meat including processed meat (Pietinen 1999 Table 5, multivariate): GL from those categories gives 1.05 (0.74–1.49). The paper's red meat row without processed meat (beef, pork, lamb; the row CUP 2010 and 2016 used) gives 0.83 (0.43–1.61) | SER Fig 4.3.3 p121; Pietinen 1999 Table 5; derived_per100g.csv | 0.83 (0.43–1.61) – unprocessed red meat, GL (R105) |
| colorectal (Fig 4.3.3, per 100 g/day) | larsson2005 | exposure | Fig 4.3.3's 1.43 (1.05–[0.95]) is SMC total red meat including processed meat (bacon, sausage, ham/cold cuts, blood pudding): GL from those categories gives 1.43 (1.05–1.95), which also confirms that the misprinted upper limit should be about 1.95. The paper's unprocessed 'red meat as main dish' row with CUP's 120 g/serving gives 1.23 (0.90–1.67) | SER Fig 4.3.3 p121; Larsson 2005 Table I; derived_per100g.csv | 1.23 (0.90–1.67) – unprocessed red meat as main dish, GL with 120 g/serving (R105) |
| colorectal (Fig 4.3.3, per 100 g/day) | norat2005 | model | Norat's 1.49 (0.91–2.43) is EPIC's measurement-error-calibrated estimate for fresh red meat; the observed estimate is 1.21 (1.02–1.43) (the value CUP 2010/2016 used). Shown as a model choice, not corrected | SER Fig 4.3.3; Norat 2005 Table 4 | 1.21 (1.02–1.43) – observed estimate |

## Notes

- R105 (2.11.2026): Fig 4.3.3's Pietinen and Larsson values are reproduced within rounding from the papers' total red meat rows (incl. processed; Pietinen's upper limit 1.49 vs printed 1.48): 1.0523 and 1.4326 (1.054–1.947). Not registered as deviations before K27a.
- Fig 4.3.3 (per 100 g/day, 3 studies): Larsson 2005 is printed '1.43 (1.05–0.95)' in the report itself (PDF excerpt supplied by the author). Reading the upper limit off the forest plot (≈1.94) reproduces the figure's summary as a fixed-effect pool and the text's as random effects; SLR 2010 Table 36's 1.29 (0.94–1.78) remains unexplained. Norat's 1.49 is EPIC's *calibrated* estimate. The HvL plot (Fig 4.3.1) has no summary; Fig 4.3.4 is a line graph without values. The 2006 SLR is not on disk; per-study weights and case numbers are not printed.