Hacker News new | ask | show | jobs
by richard_todd 2219 days ago
Presumably some women would have had an exam in February but developed enough cancer to be visible in March, so delaying a couple of months had a huge positive impact on their outcome. How can we measure the net result of all these variables?
1 comments
IgorPartola 2219 days ago
The way you measure it is how the GP defined it: percent women who develop early stage breast cancer per month * number of months closed per year.
link
jjnoakes 2218 days ago
That only makes sense if a woman gets screened monthly, which they don't.
link
IgorPartola 2217 days ago
Think about it this way: you run a clinic that detects 100 cases of early stage breast cancer a month. So you detect a total of 1200 per year, right? Now imagine you close for three months. What happens? Well, 300 of these cases do not get detected during that period. Could they get detected later after you reopen? Sure but by its nature it means that they may not be early stage anymore. Also, some percentage of the women would not reschedule an appointment right away since when the clinic reopens it will be overdue by three months worth of appointments so they might wait much longer to be seen vs the regular schedule, exacerbating the problem.

You are correct that on an individual level it is a game of chance: if you are going to develop breast cancer it’s a bad thing but if by chance you develop it in the right window of time right before your annual exam, your outcome is likely to be better. But from the point of view of screening a large population stopping testing for a period of time is bad.

Think about it in terms of COVID: what would happen if all testing was shut down for a month? No, not everyone who gets COVID would get it in that month but the people who do will absolutely not get tested, right?

link
jjnoakes 2215 days ago
No argument here, but you've changed from "percent women who develop early stage breast cancer per month * number of months closed per year" (which I responded to) to "percent early stage breast cancer cases caught per month * number of months closed per year", which sounds more accurate.
link