Hacker News new | past | comments | ask | show | jobs | submit login
"No Waiting Room" and Queuing Theory (apenwarr.ca)
28 points by bensummers on Jan 16, 2010 | hide | past | favorite | 7 comments



My girlfriend is working tonight, in casualty in the UK. Emergency Room in America. Not surprisingly, she had today off, but got called in to fill in. She called me as she got there, telling me about the seven ambulances sitting outside, knowing its going to be an ugly queue inside.

Some days she gets there and finds a five hour wait, which they work hard to get down to a sensible level. Traffic is variable, often seemingly random, and not just number of people. Different kinds of problems need different kinds of solutions.

Some days she tells me about the number of people who died on her. Those are the hard nights. It's a hard job. It's not a job that lends itself to predictability.


This is an interesting riff off my blog post.

What struck me about the particular hospital I visited in Munich was not that they were better at efficient triage or pipelining incoming cases, but that it had no queueing facility. There was no waiting room at all (unless you count a few empty chairs in the hallway).

I suspect the real reason we have such bad ER wait times in the US (3 hours and 42 minutes mean waiting time according to a Press Ganey study from 2006) is that for many Americans the ER is the go-to care provider. It's the provider of last resort for people who don't have insurance, or who simply don't have a primary care doctor. And even for people who do have insurance, there's a whole culture around going to the ER that's ingrained in many people.

As a Canadian looking from the outside in, the author of this post seems to think that the US health care system is a brutally capitalistic free market, and that it may leave some poor people without care but that supply will meet demand in some predictable way.

However, many of the people who show up at the ER are deeply unprofitable for the hospital to care for, and yet the hospital is legally required to treat them. (Though the hospital probably makes up for these losses with insured people in serious distress, making the ER profitable overall.)

The US health care system is not a free market, it's a tangled mess where patients, doctors and hospitals have no ability to control costs and are often motivated in the wrong directions. So it probably won't work to try to explain its behaviors from first principles.


I think you hinted at the real problem. Even if you have good insurance you may not be able to see your physician for a couple days. Some people choose to visit an emergency department when they want some kind of immediate relief from a non-urgent problem. If you show up while having a heart attack you aren't going to sit in the waiting room.


A lot of places have urgent care facilities, which is designed for stuff that isn't really an emergency but which you want treated that day.


He's wrong. Long lines mean a large number of people are being turned away. When your website is swapping and it takes 12 seconds to render a page, that doesn't mean "My website is serving all comers, there's just a bit of latency", it means lots of people are not being served.


The bit of missing data I am curious about would be the number of people who are turned away, or leave the hospital because of the long lines. i.e. if they had stayed the lines would have gotten longer and longer implying that there simply were not enough doctors not that they needed to be clever.


You can model waiting room arrivals same way you can model server load - using the Poisson Distribution

http://en.wikipedia.org/wiki/Poisson_distribution

As long as you can calculate the expected number of people that arrive you can adjust your operations to handle the worst case as you define it to be (ie, do we want to handle 70% within an hour? 90% within 2 hours?, etc.)




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: