13 ms·

Almost certainly the selection bias plays a major role - surgeries done early in a week are different than those scheduled later in the week.

mabbo 7y ago
Yeah, this reminds me a lot of the "Impossibly Hungry Judges" problem[0]. [0] http://nautil.us/blog/impossibly-hungry-judges http://nautil.us/blog/impossibly-hungry-judges
- gnat 7y ago
  This is a great article. The original paper was in my head as a truth, and I've revised that since reading this post. Thanks!
- samatman 7y ago
  > If hunger had an effect on our mental resources of this magnitude, our society would fall into minor chaos every day at 11:45 a.m. ...this has potential! Scrape GitHub, find closed issues marked 'bug', do a blame on the offending line or lines, find the timestamp and plot.
  - fwip 7y ago
    Has potential for the exact same selection bias as the judge study, you mean? See if we can discover some impossibly hungry programmers.
  - tmpz22 7y ago
    It would be a pita to control for time zones, or even culturally different lunch times and eating habits
    - JoshTriplett 7y ago
      Git commits do embed a timezone in the commit timestamp. And while lunchtime will vary by person, there may still be a visible effect around common lunchtimes over a large number of people.
    - samatman 7y ago
      Time is an illusion; lunchtime, doubly so. The hypothesis I would personally register is that bugs increase after 5:30 PM, and that this effect drops off in late evening hours. Basically, that compelling one's employees to stay late is an anti-pattern, where one loses more than one gains.
    - chris1993 7y ago
      I read this as 'pita' as in 'pita bread'. A clever reference to lunch time if that was the intention? Culturally different too as we don't often eat pita for lunch here, but I imagine it's common elsewhere.
- Fjolsvith 7y ago
  Having been through the court system twice, from experience I've seen dockets arranged so that more severe cases are saved for last, and cases with a quick disposition (like probation or fine) are taken care of first. Edit: This allows the law officials to remove offenders faster from the courtroom.
kerkeslager 7y ago
That's certainly a possible confounding factor that should be investigated, but let's not jump to conclusions. Jumping to the conclusion that Monday surgeries are simpler could hide an important phenomena if that's not the reason Monday surgeries have lower 30 day mortality rates.
beowulfey 7y ago
They do take this into account in the text. Odds ratios imply less risky procedures were normally done on weekends anyway, and you see the effect on all days other than Monday (to a lesser extent).
- glofish 7y ago
  Yes, they do mention this - though it can be challenging to account for all factors. You can't quite correct for variables that are not collected/known.
mgamache 7y ago
I am willing to accept bias, but Why would Monday surgeries be statistically different from Wednesday? Clearly weekend surgeries might be different, but the effect is still present Monday-Thursday. I don't work in healthcare so there might be a selection bias that I am blind to, and I am curious if you have a bias in mind. Also they did track specific procedures and for all but one (with a low n) the procedures tracked the results.
- Frost1x 7y ago
  If I were a surgeon, after a few days off of leisure to rewind, I certainly wouldn't start with my most risky, potentially problematic cases. Instead, I would try to pick operations that were the most straightforward. These would be almost like "warm ups" for more complex surgeries I would do later in the week. I'm also not a surgeon...
  - mgamache 7y ago
    Sure, but that implies that surgeons control the operating room schedules. I am not sure that's true (it might be), but it's a contested resource in most hospitals.
    - Frost1x 7y ago
      Outside of emergency surgeries, I imagine anything scheduled that isn't of immediate risk surgeons have a reasonable amount of control over, given of course the magnitude and distribution of surgeries that need performed. I can't imagine they don't look at their schedules regularly and talk to whomever does scheduling to make their lives more bearable. When they're on ER rotation, surgeons of course really have no possible way to manage this. My question would be: what's the ratio of emergency to planned/scheduled operations?
      - glofish 7y ago
        ER cases might also not serve as a proper control - the types of injuries and accidents over the weekend will likely be different from those during the week days

mabbo 7y ago

Yeah, this reminds me a lot of the "Impossibly Hungry Judges" problem[0]. [0] http://nautil.us/blog/impossibly-hungry-judges http://nautil.us/blog/impossibly-hungry-judges

gnat 7y ago
This is a great article. The original paper was in my head as a truth, and I've revised that since reading this post. Thanks!
samatman 7y ago
> If hunger had an effect on our mental resources of this magnitude, our society would fall into minor chaos every day at 11:45 a.m. ...this has potential! Scrape GitHub, find closed issues marked 'bug', do a blame on the offending line or lines, find the timestamp and plot.
- fwip 7y ago
  Has potential for the exact same selection bias as the judge study, you mean? See if we can discover some impossibly hungry programmers.
- tmpz22 7y ago
  It would be a pita to control for time zones, or even culturally different lunch times and eating habits
  - JoshTriplett 7y ago
    Git commits do embed a timezone in the commit timestamp. And while lunchtime will vary by person, there may still be a visible effect around common lunchtimes over a large number of people.
  - samatman 7y ago
    Time is an illusion; lunchtime, doubly so. The hypothesis I would personally register is that bugs increase after 5:30 PM, and that this effect drops off in late evening hours. Basically, that compelling one's employees to stay late is an anti-pattern, where one loses more than one gains.
  - chris1993 7y ago
    I read this as 'pita' as in 'pita bread'. A clever reference to lunch time if that was the intention? Culturally different too as we don't often eat pita for lunch here, but I imagine it's common elsewhere.
Fjolsvith 7y ago
Having been through the court system twice, from experience I've seen dockets arranged so that more severe cases are saved for last, and cases with a quick disposition (like probation or fine) are taken care of first. Edit: This allows the law officials to remove offenders faster from the courtroom.

kerkeslager 7y ago

That's certainly a possible confounding factor that should be investigated, but let's not jump to conclusions. Jumping to the conclusion that Monday surgeries are simpler could hide an important phenomena if that's not the reason Monday surgeries have lower 30 day mortality rates.

beowulfey 7y ago

They do take this into account in the text. Odds ratios imply less risky procedures were normally done on weekends anyway, and you see the effect on all days other than Monday (to a lesser extent).

glofish 7y ago
Yes, they do mention this - though it can be challenging to account for all factors. You can't quite correct for variables that are not collected/known.

mgamache 7y ago

I am willing to accept bias, but Why would Monday surgeries be statistically different from Wednesday? Clearly weekend surgeries might be different, but the effect is still present Monday-Thursday. I don't work in healthcare so there might be a selection bias that I am blind to, and I am curious if you have a bias in mind. Also they did track specific procedures and for all but one (with a low n) the procedures tracked the results.

Frost1x 7y ago
If I were a surgeon, after a few days off of leisure to rewind, I certainly wouldn't start with my most risky, potentially problematic cases. Instead, I would try to pick operations that were the most straightforward. These would be almost like "warm ups" for more complex surgeries I would do later in the week. I'm also not a surgeon...
- mgamache 7y ago
  Sure, but that implies that surgeons control the operating room schedules. I am not sure that's true (it might be), but it's a contested resource in most hospitals.
  - Frost1x 7y ago
    Outside of emergency surgeries, I imagine anything scheduled that isn't of immediate risk surgeons have a reasonable amount of control over, given of course the magnitude and distribution of surgeries that need performed. I can't imagine they don't look at their schedules regularly and talk to whomever does scheduling to make their lives more bearable. When they're on ER rotation, surgeons of course really have no possible way to manage this. My question would be: what's the ratio of emergency to planned/scheduled operations?
    - glofish 7y ago
      ER cases might also not serve as a proper control - the types of injuries and accidents over the weekend will likely be different from those during the week days