TABLE 3.
NutriGrade assessment of confidence in estimate effect of studies evaluated the association between various food groups and risk of breast cancer
| Outcomes | Risk of bias1 | Precision2 | Indirectness | Heterogeneity3 | Publication bias4 | Effect size5 | Dose response | Funding bias | Total score | Confidence of evidence6 |
|---|---|---|---|---|---|---|---|---|---|---|
| Total meat | 2 | 1 | 1 | 0.5 | 0 | 0 | 1 | 1 | 6.5 | Moderate |
| Red meat | 2 | 1 | 1 | 0.5 | 0 | 0 | 1 | 1 | 6.5 | Moderate |
| Processed meat | 1 | 1 | 1 | 0.5 | 1 | 0 | 1 | 1 | 6.5 | Moderate |
| Poultry | 1 | 1 | 1 | 0.5 | 1 | 0 | 0 | 1 | 5.5 | Low |
| Fish | 2 | 1 | 1 | 0.5 | 0 | 0 | 0 | 1 | 5.5 | Low |
| Egg | 2 | 1 | 1 | 0.5 | 1 | 0 | 0 | 1 | 6.5 | Moderate |
| Dairy | 2 | 1 | 1 | 0.5 | 1 | 0 | 0 | 1 | 6.5 | Moderate |
| Milk | 2 | 1 | 1 | 0.5 | 1 | 0 | 1 | 1 | 7.5 | Moderate |
| Yogurt | 2 | 0 | 1 | 0.5 | 0.5 | 0 | 0 | 1 | 7 | Moderate |
| Cheese | 2 | 1 | 1 | 0.5 | 1 | 0 | 1 | 1 | 6.5 | Moderate |
| Fruit | 2 | 1 | 1 | 0.5 | 1 | 0 | 1 | 1 | 6.5 | Moderate |
| Vegetable | 2 | 1 | 1 | 0.5 | 1 | 0 | 1 | 1 | 7.5 | Moderate |
| Fruit + vegetable | 2 | 1 | 1 | 0.5 | 0.5 | 0 | 1 | 1 | 7 | Moderate |
| Juice | 2 | 1 | 1 | 0.5 | 0.5 | 0 | 0 | 1 | 6 | Moderate |
| Cereals | 1 | 1 | 1 | 0.5 | 1 | 0 | 0 | 1 | 5.5 | Low |
| Soy | 1 | 1 | 1 | 0.5 | 0.5 | 0 | 1 | 1 | 6 | Moderate |
| Nut | 2 | 1 | 1 | 0.5 | 0.5 | 0 | 0 | 1 | 6 | Moderate |
| Legume | 1 | 1 | 1 | 0.5 | 0 | 0 | 0 | 1 | 4.5 | Low |
NutriGrade, Nutrition Grading of Recommendations Assessment, Development and Evaluation; RR, risk ratio.
Risk of bias was based on the Newcastle-Ottawa Scale, where ≥7 = 2 points; 4–6.9 = 1 point; and 0–3.9 = 0 points.
Precision is 1 point if the number of events ≥500 and the 95% CI excludes the null value; precision is 0 points if the number of events <500 or number of events ≥500, but 95% CI includes the null value (i.e., CI includes RR of 1.0) and 95% CI fails to exclude an important benefit (RR of 0.8) or harm (RR of 1.2).
When I2 was <40% or I2 was ≥40% but the source of heterogeneity was found by subgroup analysis 1 point was assigned; otherwise, 0 points were assigned.
Based on the funnel plots, Egger or Begg's test. For the outcomes with small number of studies (n < 10), the risk of publication bias was not formally assessed.
If the RR or HR <0.80–0.50 and >1.20–2.00, respectively, 1 point is assigned and the corresponding test is statistically significant; if the RR or HR <0.50 and >2.00, respectively, 0 points are assigned and the corresponding test is statistically significant.
6Moderate quality indicates that we are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low quality indicates that our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect.