This, from the recent prominent study: "The first 14 years for the cohort resulted in cumulative 648,387 person-years and 140 incident pancreatic cancer cases. Individuals consuming ≥2 soft drinks/wk experienced a statistically significant increased risk of pancreatic cancer (hazard ratio, 1.87; 95% confidence interval, 1.10-3.15) compared with individuals who did not consume soft drinks after adjustment for potential confounders."
Does this mean that roughly 1/3 of the people who developed cancer did not have 2 or more soft drinks per week and about 2/3 did? (I’m calling 1.87 times the risk as pretty much twice the risk). Or, what, exactly, is degree of risk in this context?
The degree of risk is basically the probability that somebody who drinks 2 or more sodas per day will get pancreatic cancer over the probability that somebody who drinks less than 2 a day will devolop cancer. So if 1 out of 10,000 less than 2 a week drinkers get pancreatic cancer, it is estimated that 1.87 out of 10,000 more than 2 a week drinkers will get it.
The degree to which those that drink more than 2 a week will be represented amongst pancreatic cancer cases will depend on how prevalent they are in the general population. If only 1% of the population drinks more than 2 a week, then ~2% of all pancreatic cancer cases would be people who drank more than 2 a week. If 90% of the population drank more than 2 a week, than about 95% of pancreatic cancer cases would be people who drank more than 2 a week. Of course, all of that comes with error bars.
Then"risk" is the term used to reflect the actual findings, without saying that the sugar caused the disease? In the study, there were actually 1.87 times more cases of cancer in the folks who drank two or more, compared with those who did not drink any. And because they don’t want to say that the sugar caused the cancer, they say that in this study, the risk of getting cancer was 1.87 times as great. Is that right?
They don’t “know” that the sugar caused the cancer. All they have at this point is a correlation. There could be some other confounding factor that they haven’t accounted for. Hence, the correct thing to say is that people who drank two or more regular sodas a week had nearly twice the risk of developing pacreatic cancer.
To add a point that often helps to appreciate the purported risk reduction, it’s critical to distinguish between relative risk and absolute risk.
If a drug reduces the chance of you getting disease X by 50 percent, there’s a huge difference in implication and impact if disease X is 1 in 1,000 or 1 in 10.
A 50 percent risk reduction in a 1 in 1,000 disease, means you’ve now got a 1 in 2,000 chance of getting it, i.e. from 0.1 percent to 0.05 percent (an absolute risk reduction of 0.1 - 0.05 = 0.05 percent, i.e. not really much different).
A 50 percent risk reduction in a 1 in 10 disease means you’ve now got a 1 in 20 chance of getting it, i.e. from 10 percent to 5 percent (an absolute risk reduction of 10 - 5 = 5 percent, which is much more meaningful than the first example where you’re basically going from extremely unlikely to really extremely unlikely)
In retrospect, and more in keeping with the OP, I should have phrased my post differently. To wit:
If eating Wheaties for breakfast doubles the chance of you getting a certain cancer, there’s a huge difference in implication and impact if the lifetime incidence of that cancer is 1 in 10,000 or 1 in 100.
A doubling of risk for a 1 in 10,000 cancer, means you’ve now got a 1 in 5000 chance of getting it (an absolute risk increase of 0.01 percent, i.e. from 0.01 to 0.02 percent)
A doubling of risk for a 1 in 100 cancer means you’ve now got a 1 in 50 chance of getting it (an absolute risk increase of one percent, i.e. from one percent to a two percent chance)
If I increase my absolute risk by 0.01 percent, I may stick with Wheaties.
But, if I increase my absolute risk by one percent, I think I’d switch to Shreddies.
Same relative risk increase (a doubling), but quite a different absolute risk increase.