Daniel Lemire's blog

, 3 min read

Science and Technology links (June 22nd 2019)

2 thoughts on “Science and Technology links (June 22nd 2019)”

  1. Greg Linden says:

    The sunscreen result is odd but interesting. Here’s another article that argues, convincingly, I thought, that there’s so many confounding variables that the meta-analysis is very hard to do well.


    They also note that the only randomized controlled study did find a strong positive effect.

    It does seem to me that this meta-analysis would be hard to do well. Self-reported use of sunscreen, that sunscreen is used more heavily by people at risk of skin cancer, and that sunscreen is used inconsistently in areas with heavy sun exposure all seem problematic to doing a good analysis in anything but a randomized controlled trial.

    1. Nice link.

      Aside from trying to determine a causal effect, there are some things that we do know.

      1. Studies on people who work all day outside without sunscreen, in the sun, shows that they do have more skin cancers… However, the risk factor is always less than 2.

      2. This means that “cancer causing” effect of the sun is relatively small (*)… and because it is relatively small, it means that you are likely to have trouble measuring the effect.

      3. If you do not see why a risk factor less than 2 is ‘small’, consider that your risk of cancer doubles every 8 years or so, just due to aging. And, more importantly, skin cancer is not what kills most people. You want to look at all-cause mortality rates. And that’s where it becomes interesting because sun exposure seems to have all sorts of benefits. That is, you should be eager to trade a 40% increase risk of skin cancer if it comes with a 10% reduction in heart diseases (and that seems to be the kind of numbers we are talking about) because you will lower you all-cause mortality rate.

      4. Full disclosure: I always wear suncreen and will continue to do so.

      *- It is possible that people who work all day long in the sun do not have maximal risk… e.g., office workers who occasionally go in the sun without protection could actually be worse off… but will they be 2x more likely to have cancer than people working outside all the time? Basically, the most reasonable