3 ms·

Neither does the conclusion. Nor is there an easy to read chart that shows it.

by az226 3y ago

Aachen 3y ago
Yep. The conclusion just says "has increased", and that's unfortunately not followed by a "by X" or any other quantification. I skimmed the whole article by now and am reverting back to the 260 data points (10 counties × 13 cancers × 2 genders) shown on page 3 to do it myself. Currently I'm wondering how to consolidate into a single number (on mobile). Should I eyeball guesstimate the percentage for each country, look up their population, and do a weighted average or what would be a good way to do this? Ideally I'd also compute the lower and upper bounds with the 95% CI values... Hmm Edit: ah it refers to a supplementary table, not shown in the paper, for the original data. So I don't have to eyeball it. Downloading that. Let's see ... what did I expect. 4 pages filled from top to bottom with numbers in narrow table cells. Text is not copyable, maybe due to the mobile pdf reader idk. But I'm going to draw the line here Eyeballing that thing on page 3 of the article: - it's about +5% as a very rough estimate of an overall average. - The highest I see is typically in Korea, up to about 20%, but often (for different cancer types) also around 5% or even negative in one case. (Keep in mind that 20% more may be peanuts if the original chance was one in a million. Is thyroid cancer common?) - USA average is more like 2% I'd guess, with maxima of like 5%. - Sweden is negative as often as positive, maybe +1%? Large error bars, though. Dramatic? You be the judge, but the original authors never used that word
- maxerickson 3y ago
  Is there a meaningful way to boil it down to a single number? Like for instance, in wealthier countries, rates of esophageal cancer dropped. Maybe because of better drugs for treating reflux. How is that expected to relate to rates of kidney cancer in poorer countries?
- woleium 3y ago
  So we are getting 5% better at detecting cancer every 10 years? Sounds good to me!

Neither does the conclusion. Nor is there an easy to read chart that shows it.

by az226 3y ago

Neither does the conclusion. Nor is there an easy to read chart that shows it.

Yep. The conclusion just says "has increased", and that's unfortunately not followed by a "by X" or any other quantification. I skimmed the whole article by now and am reverting back to the 260 data points (10 counties × 13 cancers × 2 genders) shown on page 3 to do it myself. Currently I'm wondering how to consolidate into a single number (on mobile). Should I eyeball guesstimate the percentage for each country, look up their population, and do a weighted average or what would be a good way to do this? Ideally I'd also compute the lower and upper bounds with the 95% CI values... Hmm Edit: ah it refers to a supplementary table, not shown in the paper, for the original data. So I don't have to eyeball it. Downloading that. Let's see ... what did I expect. 4 pages filled from top to bottom with numbers in narrow table cells. Text is not copyable, maybe due to the mobile pdf reader idk. But I'm going to draw the line here Eyeballing that thing on page 3 of the article: - it's about +5% as a very rough estimate of an overall average. - The highest I see is typically in Korea, up to about 20%, but often (for different cancer types) also around 5% or even negative in one case. (Keep in mind that 20% more may be peanuts if the original chance was one in a million. Is thyroid cancer common?) - USA average is more like 2% I'd guess, with maxima of like 5%. - Sweden is negative as often as positive, maybe +1%? Large error bars, though. Dramatic? You be the judge, but the original authors never used that word

maxerickson 3y ago
Is there a meaningful way to boil it down to a single number? Like for instance, in wealthier countries, rates of esophageal cancer dropped. Maybe because of better drugs for treating reflux. How is that expected to relate to rates of kidney cancer in poorer countries?
woleium 3y ago
So we are getting 5% better at detecting cancer every 10 years? Sounds good to me!