Hacker News Reader

Comment by glofish | Hacker News Reader

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

View on HN

Yeah, this reminds me a lot of the "Impossibly Hungry Judges" problem[0].

[0] http://nautil.us/blog/impossibly-hungry-judges

> 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.

It would be a pita to control for time zones, or even culturally different lunch times and eating habits

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.

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.

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.

Has potential for the exact same selection bias as the judge study, you mean? See if we can discover some impossibly hungry programmers.

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.

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!

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).

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.

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.

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...

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.

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?

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

The amount of influence a surgeon has in her schedule varies from place to place and specialty to specialty. It ranges from little control, to being able to look at a calendar and pick the date. Also, most surgeons do not operate on all days. In most large hospitals, surgeons have set OR days (ex. Monday, Tuesday, and half of Thursday) every week and it is incredibly difficult to change that.

The problem is with multiple-day recovery periods after the surgery. The later in the week your surgery is scheduled, the more likely you'll overlap with the weekend when care is hypothesized to be lower quality.

Part of me also wonders if people who elect to have Monday surgeries are in a better socioeconomic place to take more time off to recover.

I've had a bunch of surgeries and I wasn't given much of a choice as to when they were scheduled. These were orthopedic surgeries, but nothing urgent (stuff like a labrum repair or whatever). I was simply told to be at the hospital/surgery center at X time and that was that.

Could just be that they are the most rested then. Or just how the surgeries are ordered. Probably a simple explanation.

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.

Reply on news.ycombinator.com