June 22, 2005

Solving the Malpractice Problem

Kevin Drum's Malpractice Update points to the success the anesthesiologists had in reducing their death rates, and thus the cost of the insurance they had to carry to cover malpractice.

Indeed, this is a great success story and one that has lessons for the broader society.

Yes, the anesthesiologists have dramatically lowered their error rates -- but, this didn't happen because anesthesiologists today are a more careful bunch than they were in the past. They were able to acheive this type of dramatic improvement because of some excellent human factors studies and analysis into what caused the failures and mistakes which they applied into inventing processes and tools that prevent the failures and mistakes.

Dr. Atul Gawande, one of today's most interesting medical writers, has been studying how to prevent errors in medicine. In 1999, he wrote the definitative essay on this topic, When Doctors Make Mistakes [1], which documented his research into discovering why mistakes get made and what can be done to reduce medical errors. As he noted in that essay, the problem isn't that there are a bunch of bad doctors (although there are some), the real problem is that there are good doctors who make bad mistake despite their very best efforts.

So what was it that the anethesiologists did that made their outcomes so dramatically different? Here's a brief synopsis of what happened from a Boston Globe article discussing Dr. Gawande's studies:

The systems approach isn't new to medicine. In the 1970s, anesthesiologists, facing a malpractice insurance crisis, took a hint from the aviation and nuclear power industries and hired systems engineers to watch how they worked. The suggested changes were striking for their simplicity. Some involved designing oxygen tanks so someone couldn't accidentally shut off a patient's oxygen; others had all dials turning in the same on-off direction. Others prescribed using certain speech patterns - an element in what the airline industry calls "crew resource management" - to eliminate miscommunication. Those system tweaks caused the death rate in the United States from anesthesia to drop over the past 30 years to one-50th of the former number.

Are medical malpractice awards too high? Would capping the awards make malpractice less likely to happen? Most likely not. Dr. Gawande's investigations show that capping the awards would not help the problem of malpractice and would do nothing to solve the real problem. And so he advocates looking for solutions by understanding the root cause of the errors and by applying that insight to the problem solutions.

Gawande had been wondering if technology could help with the count [of sponges during surgery] and had been consulting with engineers from MIT. "For God's sake," he says, "you can't walk out of a bookstore without an alarm going off. How can a patient leave an OR with an instrument inside him?" Then a colleague pointed him to a young entrepreneur's business plan posted online. Brian Stewart of Santa Monica, California, was investing in companies making medical devices when he got into a discussion with his surgeon father. His father had been late for a meeting with him because of a missed-sponge count following an operation. Father and son were at a grocery store when Brian Stewart, who knew nothing of the complexities of the operating theater, ridiculed the nurses who had made his father late. "I said, `What can be so hard to count to 20? You just count to 20 and then do it again.' " Just then they reached the cashier, who slid their items past a bar-code scanner. "At that point, we just looked at each other and said, `Aah.'"

The younger Stewart arranged for a company to stamp bar codes onto surgical sponges, using waterproof polymers and a complex pattern that identifies each sponge individually. He had a fabricator produce a hand-held scanner that chirps every time it registers a sponge. The scanner has a nice big display that reads: "Total IN," "Total OUT," and "Total REMAINING."

Gawande began testing the system last spring, challenging nurses to see if they could trip up the scanner. They dipped the sponges in mustard and ketchup. Finally in October, Gawande and his colleagues suited up for surgery and draped a volunteer patient for the simulated operation. Each time the nurses handed Gawande a sponge, he'd plunk it in the bowl of cough syrup and chocolate sauce to simulate blood, stool, and pus.

It seemed to go well - nurses passing sponges in front of the scanner, a reassuring bleep sounding each time. Then Gawande palmed a sponge, and a couple stuck together in a single goopy mass. That's when the nurses came up three short. But instead of counting, recounting, and searching the patient, they could see from the scanner exactly which sponges were missing and when they were scanned in. With those clues, the nurses found the missing sponges within minutes, two of which were in the garbage.

The device still needs work. Nurses had to struggle with especially goopy sponges, stretching them this way and that under the scanner - a distracting and mildly frustrating task. "The point is not to prove that this is the greatest thing that's ever been made," Gawande told them, "but to get a sense that we've got a way to take some of the dog and pony show out of the counting."

It all needs adjusting. But Gawande hopes that thousands of such tweaks in all sorts of settings in hospitals across the country - provide engineering solutions to "idiotic" mistakes.

Dr. Gawande's refreshing and creative investigations will go much further in solving some of today's medical malpractices problems than any of the so-called initiatives to cap malpractice awards.

It seems to me that this is a stark example of the difference between having an ideological answer (cap malpractice awards) to a pragmatic answer (diagnose the root case of the errors and help set up systems that mitigate human error).
Indeed, this seems to be the difference between Bush Republican policy and Democratic policy.

[1] When Doctors Make Mistakes was originally published in the New Yorker, Feb 1, 1999. It has since been included in his book Complications: A Surgeon's Notes on an Imperfect Science

Posted by Mary at June 22, 2005 01:48 AM | Science | TrackBack(2) | Technorati links |

This just sounds ever-so-sensible, probably too sensible to be tried ;)

Posted by: natasha at June 22, 2005 10:53 AM

There's an even simpler thing to try -- RFID chips. Since they operate off of radio frequency scanning, they don't even need to have an optical scanner at work.

Posted by: beerwulf at June 22, 2005 11:33 AM

I'll second RFID, that would probably eliminate any problems due to goop and would eliminate the scanning step entirely, since it could be autoscanned as it moved in and out. Could also help in locating lost sponges.

Posted by: Boronx at June 22, 2005 08:14 PM

And, we should videotape ALLLLL surgical procedures for later study in case something goes wrong, or right.

Of course, the surgeons will just LOVE this idea. Maybe they can discourage videotaping by charging, say, $10,000 for the tape, just as the courts do for trial transcripts. Say -- oh just off the top of my head -- someone's brother got horribly mangled in an outpatient surgical procedure. Say, the op. was supposed to be a simple 20 min. look-see, then off to a ball game. But, something went horribly wrong, and the ambulance wasn't called for 2 hrs. Brother winds up permanently paralyzed and terribly damaged psychologically. Doctor wins in court, right? There's no videotape of the operation to show what went horribly wrong there, so we'd like to see a transcript of the trial, just to see what went so horribly wrong THERE, but it'll cost us $10k, right? They made an audio tape of the trial, but we can't get that, apparently. It's gotta be a paper transcript, and that costs $10k.

These people are hiding.

Posted by: ferd at June 22, 2005 09:57 PM

I favor genetically engineered sponges that scream on cue.

A Berkeley CS professor I know is working on a couple of white papers related to process efficiencies in medicine. Ironically, they're applications of technologies that have been widely implimented in manufacturing and online vending, but for which medicinal science has been slow on the uptake. He seems to think there is a lot of low-hanging fruit in the whole field.

Posted by: Saam Barrager at June 22, 2005 10:03 PM

Mary describes this pragmatic approach as being to

diagnose the root ca[u]se of the errors and help set up systems that mitigate human error

But I don't think this approach cares about "root causes" at all. They're just tracking sponges -- because missing sponges are the form of the error. The root cause might be carelessness, exhaustion, inadequate training, malice, blindness, or any of a hundred other things. This approach doesn't care, and that's a source of its strength.

Posted by: Vance Maverick at June 22, 2005 11:02 PM

Don't use RFID!!! The Man will know everything about you and your sponges!

Posted by: Preston at June 23, 2005 09:06 AM

Nice work, and more confirmation that there are much better approaches to the malpractice problem than caps. At the other end of the process, I'm intrigued by some of the things I've heard regarding the success of APOLOGIES in reducing potential lawsuits after a mistake. Some health outlets are experimenting with what you might call "no-fault" mistake programs, where being forthright about mistakes to patients replaces covering them up and hoping nobody notices. I would need to further research the topic to be specific, but I believe they have found that, strange as it may seem, merely being up front about the mistake, and being sincerely apologetic, is enough to satisfy a large chunk of potential litigants. Imagine that--honesty being the best policy!

Posted by: torridjoe at June 23, 2005 09:13 AM

Counting sponges is nice, it helps eliminate evidence for a law suit. This will lower the case load. But it don't help the problem.

The problem is that we don't have a structured system for diag. Most times the doc takes a guess and tries a few pills, looks at some blood tests and finds a 'diagnosis' that is a good fit for the symptoms.

That's not a methodology!!

This focus on incorrect procedures is too focused on the OR. Yes that's were lots of people die and where a lot of cases come from. But the BIG problem, the one that sucks down most health care dollars is getting the diagnosis wrong. It happens so much that we now plan for it. The doctor makes a guess, prescribes some pills or treatment and asks you to come back in a few days to see if his guess is right. Most times the body heals it's self and the doctor thinks he did it.

We need a structured methodology in diagnosis with every possible question and test built into a large computerized knowledge base. There are already some very good systems that have been developed for individual dieses.

Now we need one that works for 99% of cases. Heck if we can map the human gean we can draw out the methodology for diagnosing what ails people.

So long as the system has a way to double check its assumptions via qualitive tests it should work well. It can then chuck the cases that it can't diagnose over to the 'specialists'.

We have seen knowlage bases in other industries that are more powerful and more complete than any one persons understanding of the subject. We can do that for medicine.

It should be possible crate a very large database with the right questions and correct tests to diagnose most problems correctly.

Correct diagnosis via applying the 'appropriate' tests would dramatically cut medical costs. Everyone has had a doctor say "Do .... and come back ....." What this means is let me guess and if not come back and I will take another guess.

Everyone has a friend who can tell a tale of woe about trapsing from one specialist to the next, each one making incorrect guess, until finally, often months later and $$$$ later, one found the right answer.

All the wrong and right answers need to be tracked so that the knowlage base engineers can correct the decision model. Right now there is NO feedback system that will tell the first doctors in the chain what they did wrong in diagnosis and correct the producers used.

The problem is this, a lot of doctors think they are infallible, or think they get it right 90% of the time. This is just not true. Without a structured methodology and a closed feedback loop their delusions will continue. We can do better than this.

Posted by: James M at June 23, 2005 11:31 AM

RFID doesn't sound like a good idea to me in surgical sponges. It would require transmitters be put in sponges.

What about sponges that breakdown into stuff that can be processed by the body? Then even an improper count wouldn't be cause for concern.

Posted by: Zach at June 23, 2005 12:39 PM

I saw something the other day about attaching a little BB to each sponge so all of the sponges could be put in a x-ray scanner that counts the BB's attached to each sponge. They were saying that it's better than bar coding because you don't have to handle the sponges at all, and that it's better than RFID tags because RFID tags are hard to read once they're wet and bloody. It sounds like an interesting idea to me. Has anybody heard about it?

Posted by: George at June 26, 2005 10:24 AM