Table 2.
Number and percentage of meta-analyses overall and by cancer type that meet the individual and the overall criteria used for the grading of the evidence on diet and cancer risk.
| Total (n = 860) | Head and necka (n = 38) | Esophageal (n = 48) | Stomach (n = 122) | Colorectal (n = 221) | Liver (n = 20) | Gallbladder (n = 2) | Lung (n = 144) | Skin (n = 18) | Breast (n = 163) | Kidney (n = 41) | Urinary bladder (n = 43) | |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Criterion | ||||||||||||
| P value <10−6, n (%) | 25 (2.9) | 5 (13.2) | 2 (4.2) | 0 (0) | 12 (5.4) | 2 (10) | 0 (0) | 0 (0) | 1 (5.6) | 3 (1.8) | 0 (0) | 0 (0) |
| P value <10−3, n (%) | 75 (8.7) | 11 (28.9) | 6 (12.5) | 2 (1.6) | 24 (10.9) | 6 (30) | 0 (0) | 11 (7.6) | 4 (2.2) | 11 (6.7) | 0 (0) | 0 (0) |
| P value <0.05, n (%) | 247 (28.7) | 28 (73.7) | 17 (35.4) | 19 (15.6) | 69 (31.2) | 10 (50) | 0 (0) | 54 (37.5) | 6 (3.3) | 35 (21.5) | 7 (17.1) | 2 (4.7) |
| I2 > 50%, n (%) | 227 (26.4) | 15 (39.5) | 14 (29.2) | 25 (20.5) | 44 (20) | 7 (35) | 0 (0) | 45 (31.3) | 6 (3.3) | 46 (28.2) | 12 (29.3) | 13 (30.2) |
| I2 ≤ 25%, n (%) | 450 (52.3) | 18 (47.4) | 27 (56.3) | 77 (63.1) | 115 (52.0) | 8 (40) | 2 (100) | 73 (50.7) | 8 (4.4) | 79 (48.5) | 23 (56.1) | 20 (46.5) |
| Prediction interval excluding the null, n (%) | 46 (5.3) | 1 (7.9) | 1 (2.1) | 4 (3.3) | 22 (10) | 2 (10) | 0 (0) | 5 (3.5) | 0 (0) | 10 (6.1) | 0 (0) | 1 (2.3) |
| Evidence of small study biasb, n (%) | 69 (8.0) | 5 (13.2) | 7 (14.9) | 7 (5.4) | 13 (5.9) | 2 (10) | 0 (0) | 19 (13.2) | 2 (1.1) | 6 (3.7) | 5 (12.2) | 3 (7.0) |
| Evidence of excess significance biasc, n (%) | 121 (14.1) | 15 (39.5) | 6 (12.5) | 12 (9.8) | 23 (10.4) | 6 (30) | 0 (0) | 32 (22.2) | 4 (2.2) | 20 (12.3) | 3 (7.3) | 0 (0) |
| Overall grading | ||||||||||||
| Not significant, n (%) | 613 (71.3) | 10 (26.3) | 31 (64.6) | 103 (84.4) | 152 (68.8) | 10 (50.0) | 2 (100) | 90 (62.5) | 12 (66.7) | 128 (78.5) | 34 (82.9) | 41 (95.4) |
| Weak, n (%) | 182 (21.2) | 23 (60.5) | 12 (25.0) | 17 (13.9) | 46 (20.8) | 5 (25.0) | 0 (0) | 44 (30.6) | 2 (11.1) | 24 (14.7) | 7 (17.1) | 2 (4.7) |
| Suggestive, n (%) | 42 (4.9) | 1 (2.6) | 4 (8.3) | 2 (1.6) | 11 (5.0) | 3 (15.0) | 0 (0) | 10 (6.9) | 3 (16.7) | 8 (4.9) | 0 (0) | 0 (0) |
| Highly suggestive, n (%) | 13 (1.5) | 4 (10.5) | 1 (2.1) | 0 (0) | 4 (1.8) | 2 (10.0) | 0 (0) | 0 (0) | 1 (5.6) | 1 (0.6) | 0 (0) | 0 (0) |
| Strong, n (%) | 10 (1.2) | 0 (0) | 0 (0) | 0 (0) | 8 (3.6) | 0 (0) | 0 (0) | 0 (0) | 0 (0) | 2 (1.2) | 0 (0) | 0 (0) |
aThis report presents results for cancers of mouth, pharynx, larynx, and upper aerodigestive tract.
bSmall study bias is based on the P value from the Egger’s regression asymmetry test (two-sided P value ≤0.1) and the random-effects summary estimate was larger compared to the point estimate of the largest study in a meta-analysis.
cExcess significance bias is based on the P value (two-sided P value ≤0.1) of the excess significance test using the largest study (smallest standard error) in a meta-analysis as the plausible effect size.