Quote:
Originally Posted by spiderplant
That would be true if they'd only sampled 1000 people, but actually the 7 in 1000 figure is downscaled from a sample size of hundreds of thousands. So scaling back up is statistically valid.
|
The studies they reviewed had such a wide range of results, that the results were deemed a low certainty of being valid.
Original report
Quote:
|
cause mortality, cardiovascular mortality, stroke, myocardial infarction, and type 2 diabetes (range, 1 fewer to 12 fewer events per 1000 persons with a decrease of 3 servings/wk), with no statistically significant difference in 1 additional outcome (cardiovascular disease) (16). For cohort studies addressing adverse cancer outcomes (31 cohorts with 3.5 million participants providing data for our dose–response analysis), we also found low- to very low-certainty evidence that a decreased intake of processed meat was associated with a very small absolute risk reduction in overall lifetime cancer mortality; prostate cancer mortality; and the incidence of esophageal, colorectal, and breast cancer (range, 1 fewer to 8 fewer events per 1000 persons with a decrease of 3 servings/wk), with no statistically significant differences in incidence or mortality for 12 additional cancer outcomes
|
A finding of 1 in 1,000 and another of 12 in 1,000, demonstrate the original studies can't be relied upon.