Soon to come to the US, this software can catch deadly mistakes other EMRs would never notice.
What if your EMR detected medication errors in the same way that your credit card company detected fraud?
We spoke with Gidi Stein, MD, PhD, founder of MedAware, an Israel-based company fighting inpatient medication orders and prescription errors in a new way.
It all started one day when a nine-year-old boy with asthma fell off of his bike. He ended up dying from an intracranial hemorrhage. Just a week before, his PCP had mistakenly placed him on warfarin, instead of singulair by clicking the wrong entry on a prescription drop-down list.
It wasn’t bad judgement on the part of the physician, Dr. Stein argued, but rather a slip-up, a mix-up. “It’s like killing someone with a typo,” he explained.
Mistakes like the one potentiating the young boy’s intracranial hemorrhage motivated Dr. Stein to start MedAware, a system adding artificial intelligence to EMRs that goes beyond just the basic interactions between medications and allergies. The software analyzes the order much like a credit card company would detect fraud. It uses a mathematical model that describes which patients are likely or unlikely to be prescribed for that specific drug at that specific place and time, much like a bank would identify purchases not typical for a certain time, place or person as potential fraud.
Put in an order to do a pregnancy test on an 80-year-old man? Give mannitol to a 12-year-old with a URI? If they’re not allergic to it, most EMRs would let the order through. MedAware’s ability to learn a physician’s behavior and application of mathematical algorithms takes fact checking a step further, reducing false positives and increasing true positives.
According to their research, most EMRs flag 20-30% of prescriptions and have up to 90% false positives for suspected interactions or allergies for medications, leading to “alert fatigue,” and thus desensitization. It therefore makes sense that 90-95% of these alerts are ignored by physicians, Dr. Stein cited. So, not only are the messages from these rudimentary algorithms contributing to alert fatigue and are they often not helpful, but they don’t catch mistakes like warfarin being prescribed to a nine-year-old with a medical history of only asthma.
In contrast, MedAware only flags about 0.2 to 0.5% of all prescriptions, with about 75-80% being true positives, and only 35% are false negatives.
These deadly medication errors aren’t rare, either. Out of the four billion prescriptions written in the US every year, an estimated eight million contain deadly errors. Most of these errors weren’t typical in the old days, as handwritten prescriptions weren’t as susceptible to “typos,” Dr. Stein remarked. “When I wrote prescriptions, I’d write them on paper and physically hand it to the patient. Now, with electronic orders and prescriptions, it’s much easier to click on the wrong patient and give the wrong person the wrong medication.”
MedAware is currently live in Israel in their biggest (2,000 bed) hospital, where Dr. Stein practices, and is coming to the US in 2017.
“I spend about one week per month in my US office, in Stamford, Connecticut, and the rest at home in Israel, Dr. Stein explained. “We are now planning to enter the U.S. market through strategic partnerships. We just opened the office in the U.S. and are in discussions with leading EMRs for integration and implementation.”
There are several other companies out there working on reducing medication errors. What sets MedAware apart, though, is its artificial intelligence, Dr. Stein said. It learns physician behavior, and then identifies the outliers as the potential errors, which seems to be an industry first.
The software doesn’t tell physicians how to do their job
“Even if you’re a great poet and you write the best novels and poetry in the world, sometimes you may have a typo and you use a spellchecker to fix it; and it doesn’t say that your poems are not perfect. It just says you’re human,” Dr. Stein explained.
“We don’t tell them what to do. We just say: Hey, did you know that this patient has two platelets. Maybe you shouldn’t give him the aspirin? Just consider that. I’m not telling you what to do. So, we learn the behavior and then we know how to narrow it. We know how to project it. But we don’t give advice. We just give the warning: Hey, you’re an outlier.”
So, what will it cost providers?
The cost is commensurate with the practice size, from private practice all the way to the big hospitals. The cost is about 5% of the expected cost reduction, Dr. Stein said. About the industry standard.
What advice to do you have for physicians looking to pursue something entrepreneurial like you did?
“I think the challenge is really to decide that you stop doing what you did for the last 20 years and do something different, completely different; submerge yourself into the unknown. You know, take your ego and leave it on the beach and start from scratch and to try to make a difference. And it’s a long and tedious road. There’s a ninety percent chance you will fail – again and again and again. But it’s a fascinating ride and you have to enjoy it, with all the risks involved. I enjoy it. If you don’t enjoy it, you shouldn’t get in there.
“I think mainly understanding that you will fail again and again and again and that’s part of life. And it’s much more concentrated when you’re on a start-up roller coaster. But it’s fun. I like it.”
So what do other people think of all this?
Dr. David Bates, a leading national Patient Safety expert and opinion leader, Professor of Medicine at Harvard Medical School, and a Professor of Health Policy and Management at the Harvard School of Public Health said:
“It has been hard to find medication errors which come completely out of the blue – like a medication used only in pregnant women which is ordered for an elderly male – but this approach detects orders which appear to be anomalous in some way, and it represents a very exciting new way to pick these errors up before they get to the patient.”
Check out more at MedAware.com and their study in the Journal of the American Medical Informatics Association.