Not nitpicking your numbers at all (mainly cause im too lazy to go hunting down the original sources), but a big problem that science media gets completely wrong is how they report risk percentages. They conflate changes in absolute risk with relative risk constantly, and it really hurts messaging.
For example, a few years back, the WHO released a report on consumption of processed meat and how it relates to colorectal cancer risk. Even their own press release, which should be perfect, says “each 50 g portion eaten daily increases the risk by 18%”. That is really misleading if you dont know they are talking about a relative risk. The average person will interpret this as new risk %= baseline risk % + 18%.
The absolute lifetime risk of colorectal cancer is ~4%, so daily consumption of processed meat should bump it to ~4.7% (well, technically lower since the 4% includes processed meat consumers). Giving the before/after percentages helps communicate the risks way better. Even better is a risk curve showing how the risk changes as consumption increases (obviously that relies on the data being available).
Its also better to be able to contextualize so you can make well informed decisions across your life, e.g., it’s dumb to deprive yourself a joy that increases lifetime cancer risk by 0.5% while ignoring other facets of your life that increase cancer by a much larger margin.
I tried not to conflate absolute and relative risk. The numbers I was going with came from the link I posted, which was not from a science journalist, but from the US National Cancer Institute. Also, note that the comment you replied to was more about an off the cuff comparison of the risk between CT scans and drinking alcohol. It wasn’t meant to present scientific rigour.
Below is directly from the linked article, emphasis mine:
Using data from Australia, recalculated using US standard drinks, the recent Surgeon General’s Advisory reports that
among 100 women who have less than one drink per week, about 17 will develop an alcohol-related cancer
among 100 women who have one drink a day, 19 will develop an alcohol-related cancer
among 100 women who have two drinks a day, about 22 will develop an alcohol-related cancer
This means that women who have one drink a day have an absolute increase in the risk of an alcohol-related cancer of 2 per 100, and those who have two drinks a day an absolute increase of 5 per 100, compared with those who have less than one drink a week. For men, the number of alcohol-related cancers per 100 is 10 for those who have less than one drink a week, 11 for those who have one drink a day (an increase of 1 per 100), and 13 for those who have two drinks a day (an increase of 3 per 100).
Not nitpicking your numbers at all (mainly cause im too lazy to go hunting down the original sources), but a big problem that science media gets completely wrong is how they report risk percentages. They conflate changes in absolute risk with relative risk constantly, and it really hurts messaging.
For example, a few years back, the WHO released a report on consumption of processed meat and how it relates to colorectal cancer risk. Even their own press release, which should be perfect, says “each 50 g portion eaten daily increases the risk by 18%”. That is really misleading if you dont know they are talking about a relative risk. The average person will interpret this as new risk %= baseline risk % + 18%.
The absolute lifetime risk of colorectal cancer is ~4%, so daily consumption of processed meat should bump it to ~4.7% (well, technically lower since the 4% includes processed meat consumers). Giving the before/after percentages helps communicate the risks way better. Even better is a risk curve showing how the risk changes as consumption increases (obviously that relies on the data being available).
Its also better to be able to contextualize so you can make well informed decisions across your life, e.g., it’s dumb to deprive yourself a joy that increases lifetime cancer risk by 0.5% while ignoring other facets of your life that increase cancer by a much larger margin.
I tried not to conflate absolute and relative risk. The numbers I was going with came from the link I posted, which was not from a science journalist, but from the US National Cancer Institute. Also, note that the comment you replied to was more about an off the cuff comparison of the risk between CT scans and drinking alcohol. It wasn’t meant to present scientific rigour.
Below is directly from the linked article, emphasis mine:
Using data from Australia, recalculated using US standard drinks, the recent Surgeon General’s Advisory reports that
This means that women who have one drink a day have an absolute increase in the risk of an alcohol-related cancer of 2 per 100, and those who have two drinks a day an absolute increase of 5 per 100, compared with those who have less than one drink a week. For men, the number of alcohol-related cancers per 100 is 10 for those who have less than one drink a week, 11 for those who have one drink a day (an increase of 1 per 100), and 13 for those who have two drinks a day (an increase of 3 per 100).