Hacker News Reader

Comment by usaar333 | Hacker News Reader

When my wife received an epidural, I noticed a warning label on the connector that said "Epidural, do not connect IV".

The only thing stopping such a horrible error is this label. IV and Epidural connectors are fully compatible even though they should never be used interchangeably.

I can only imagine how many other simple medical failsafes are missing.

View on HN

Indeed, while the anesthesiologist was ultimately at fault, this story made me wonder about the design parameters around the mistake. Are the vials the same color? Do they have the same size and shape? Was this anesthesiologist near the end of a 16-hour shift and had to differentiate two very similar objects?

Your comment about the physically compatible, deadly-if-switched connectors was an eye opener: It's almost like if the USB standards group settled on a two-prong design that was physically compatible with a North American 110V AC outlet, and every USB cable shipped with a label saying "DO NOT PLUG INTO WALL OUTLETS".

It's almost like if fueling stations pumped gasoline and diesel with the same shape nozzles.

This is a very interesting little point, that consumer electronics (especially ports) have little physical safeguards to prevent errors—e.g., you can plug USB2 into USB3 for backward compatibility, but not vice versa, for the B plug; ungrounded extension cords try to prevent you from plugging in grounded plugs. Medical equipment generally doesn't do this. I think the problem is that there is a tradeoff between safeguards and allowing improvisation in a time of shortage or emergency; for example, I've seen blood draw lines repurposed as external irrigators (not as catheters) and that's totally safe as long as you dispose of the needle in a sharps bin.

That's a good point.

I guess one problem is that the failures of the `can't improvise' case are really easy to see and imagine. Whereas the other failure of mixing up connectors could always be blamed on human error.

Someone should figure out which scenario causes more damage. With enough stocking, we should be able to avoid the need for improvisation?

As a technical solution, medical connectors could come with bits that prevent mistakes (and clearly different colours and perhaps even outside textures to use sense of touch, too); but with the safety bits simple to break off on both sides, so that you can still plug them together.

So you can still plug together arbitrary things, but you get a short moment where you have to consciously use some force.

As a programmer with a slightly paranoid persona, I naturally spent countless hours thinking about 'non representable states' (my attachment for FP and formal systems wouldn't surprise anybody). And the same for real life situations, be it driving[2], or even more healthcare. Things that shouldn't happen shouldn't be possible unless clearly thought out and willingly carried out actions. I wonder how 'CS wishful thinking' this is, if this is a cultural problem. After all medicine is doing "ok"[1] without it, would it bring significant and stable benefits, at what cost ... Also we (whatever the country) should support this kind of effort way more, and pay and employ healthcare workers enough so they can work optimally. And governments should make it as easy and transparent as possible. The rest is ridiculously secondary.

[1] not diminishing what happened to people at all, but most of the time we're satisfied by medical approximations as much as we go on with our own mistakes.

[2] I'm all for self driving car studies increase, now that I've seen how people are taught and drive.

> non representable states

Far off-topic from medicine. This is something that has only possibly occurred to me over the past few weeks: this is a healthy perspective of software. We work in computer science and so far as science is defined today, it is about quantifying the incorrect. We spend far too much time talking about what a system does, instead of what it doesn't.

I've found that documenting constraints (before they occur) results in a clearer and more obvious path to valuable results.

Your tone indicates that you might see this as a personality flaw - I've recently discovered that it's most certainly the exact opposite.

To tie this back into medicine, it makes complete sense that epidural equipment should have the constraint of being incompatible with IV. This would drive up costs but life is priceless so far as I am concerned. To a far lesser degree, nobody should have to live with the guilt of making a mistake when the mistake could have been so easily avoided.

I used to have ideas about negative design where you'd plan the failing interface before fitting the correct working one. So often errors are just there because people spent the time and fun on the 'productive' parts of a component.

But even though computers are often about pre-established rules, protocols and structures. There's also a part dealing with unknowns, probabilities (Ward Cunningham even made some parsing theory based on grammarless systems, very original).

Sadly I never ran into that in my studies.

Good point about the 'value' of life. It should dominate the decisions. That said, I also believe that when a system is dumbed down to a point where nothing bad can happen, people are even more sloppy. Solution lying in the middle, as people say.

ps: about my tone, it's not a flaw to be paranoid, to a certain extent. Not saying we should rejoy about PHP/Wordpress 25% of the web though. Again, balance.

> Sadly I never ran into that in my studies.

The common `complaint' about eg Haskell's type system is that it only tells you off if something is amiss, but doesn't actually do anything for you when your program is fine. So, that's just the opposite.

See eg the key-note by Conor McBride at LambdaJam: "What are Types for, or are they only Against?"

https://a.confui.com/public/conferences/56b46f17db9ac1529400...

My understanding is that most medical equipment is now designed to be incompatible for accidental hook-ups to the wrong thing. If the epi and IV aren't incompatible, there's somebody already working on making things that way..

I agree with you.

Poka-yoke, a technique from the Lean world, is useful here.

Reply on news.ycombinator.com