Friday, December 5, 2008

What we have here is a failure to communicate

IT is the great white hope for healthcare, the healthcare administrator’s wet dream: we’ll be able to reduce the cost of this monster using IT and improve service at the same time. And like all fantasies of this type, what you get in the cold light of day just isn’t quite the same – that sensuous young woman turns out to be a withered old hag with a sour disposition (or, for my female readers – if I still have any: that buff young man turns out to be a crotchety old jerk with a hairy back).

One of the principles is easy to grasp. Anywhere between 50% and 80% of healthcare professionals’ time is spent tracking down information so they can provide proper healthcare. That’s right – that doctor who’s getting paid a million smackers a year: he spends most his time finding the right pieces of paper. That's not all - almost all of the preventable deaths that occur relate to missing information one way or another.

So, if you stick all that information on computers, and they can talk to each other, then the information will just be right there, exactly when and where the healthcare professional needs it. Magic! And we could get twice as much work for the same amount of money, and with less “totally negative health outcomes”. So you can see how seductive this idea is – up there with the supermodels. Also, see the Turkey I had for thanksgiving.

Actually, it would be magic if it worked that way, but the real magic is in the innocuous words “they can talk to each other”. In the healthcare IT industry, this is called “interoperability”, and it’s the Holy Grail. It bears startling resemblance to the Holy Grail too. Not only has no one ever seen it, we don’t even know what it actually is.

In order for computers to be able to talk to each other, they need to understand each other in a deep and meaningful way. At least that’s what the experts say.

I’m not so sure. I’m married, and I know that once you understand each other, you no longer need to talk anymore. Yeah, yeah, everyone laughs when I say that, and pities Mrs. Health IT Nerd. And I mean, I understand their pity, because I know me even better than they do, but they’ve missed the point: Mrs. Health IT Nerd and I are never going to understand each other (any of you that are married will know exactly what I mean). So our lives are full of interesting times, and we are forced to keep talking to each other.

So this is what makes interoperability so much fun: we’re never going to understand each other fully, but we have to get along anyway. I think this is one of the craziest things that happens in health IT, that the industry so seriously misunderstands what will enable interoperability, and what the results might be.

Classic interoperability theory says that in order for two computers to talk to each other, you need the following things:

  • A transmission channel between the two (usually, but not always, bidirectional)
  • A common set of terms (words) with meanings that both parties understand
  • A common set of information models (grammar/story plotlines) to allow the pieces of meaning in terms to be assembled into a coherent larger structure
  • An agreed process (who says what when, and what happens next)

This is called the “interoperability stack” (I presume “stack” like as in “Dad, I totally stacked your car”).

It’s the same requirements for humans to talk together, on any scale, from my two small kids arguing about who gets to be the doctor and who is the patient, to diplomats from two large countries resolving which side of the border their soldiers will get to acquire their need for emergency healthcare on.

Actually, that stack above is incomplete. There’s something else that most interoperability wonks don’t stress, but I can’t stress enough:

  • The two parties need to share an agreed context of operations

Like Mrs Health IT Nerd and I, no one knows how to even agree on what this “context of operations” thing is, how wide and deep it is.

Take a simple case: in Isaac Asimov’s Foundation, one of the characters says: “Violence is the last refuge of the incompetent”, by which he means, only the incompetent will use violence because it doesn’t solve anything.

Well, I have a friend (Yes, I *do* have one), and he’s a wingnut, so he says that this means that competent people would have resorted to violence long before it’s time for the last refuge. While that interpretation is the polar opposite of the one that was intended, the actual words and the grammar are understood the same way. It’s the different background values people use when evaluating the meaning of the phrase that make the difference here. (Which interpretation is correct? It’s not like it matters for this column, but I figure that what happened in Iraq - or any other war - shows the statement is wrong and stupid however you want to read it.)

This is why interoperability is so hard: there are so many layers to understanding. A whole industry exists to define interoperability based on standards that provide meaning for that stack, a whole alphabet soup of them, such as HL7, CEN, ISO, IHTSDO, ASTM, ANSI, WHO, W3C, OASIS, WS-I…. a never-ending profusion of standards bodies. You know what? These standards bodies, these definers of interoperability, they can’t even interoperate amongst themselves, so it’s the proven-blind leading the probably-blind.

These standards are all going to fail. Well, not so much fail (though it might be best if, umm, if we all don’t actually look too closely at them when we say that), as not quite deliver all the things people are demanding from them – just small things, like life, the universe, everything, and also world peace as well. These things won't happen, but there will be some outcomes: life will get better, healthcare will improve. But you know should know by now what happens when healthcare improves: costs go up; so even if these interoperability standards deliver everything anyone dreams of, the outcomes won’t be what they desired in terms of cost-cutting.

Even if the healthcare administrators and those who pay for healthcare (i.e. you!) scale back the expectations of what interoperability can achieve to something reasonable, these standards are not going to deliver, because they’re all based on the expectation that if you solve the technical problems, interoperability will just happen.

It’s people who insist on doing things differently, calling the same thing by different names or vice versa. It’s people, who, given the same patients, the same healthcare problems, and the same computer systems, find completely different ways to achieve roughly the same outcomes. And for all these people – both healthcare professionals, and healthcare informaticians (horrible word!): there’s my way of doing things, and all the wrong ways to do it. There’s even a step beyond that, people for whom there’s my way of doing something, and all the other ways that I am dedicated to destroying. These people are methodological terrorists, and they are attracted to standards. This is part of why the healthcare standards wars are such fun.

So the fundamental problem of interoperability, of getting the information to the right person at the right time, is the first and last steps – getting it out of the first person who has it, and into the other person who needs to understand it in the appropriate context, how it relates to all the other information they have. Compared to these two problems, everything else is just plumbing, though we can’t even get that right. Interoperability is about people, not technologies.

Perhaps the healthcare industry isn’t so stupid to spend below average amounts on IT after all.

However we’re clearly going to spend what we do have on chasing the chimera of getting computers to fully understand healthcare – that is, us. Well, that will never happen.

So I think that we need to start focusing on enabling interoperability without trying to understand each other. See, if we all focus together on trying to achieve something perfectly useless, there’s a reasonable chance that we might actually succeed, especially since we've already achieved one of the desired outcomes – we’ll never understand each other.


No comments: