A few years ago, a nine-year-old Israeli boy dropped dead from his bike. Why? Because of preventable medical error.
- Turning poo into power:Startup can get you cooking with homemade gas
- How groundbreaking tree sensor can save our groundwater
- The Israeli startup that lets you eat meat - without eating the animal
The boy was killed by a brain hemorrhage after a doctor accidentally prescribed an anticoagulant instead of an asthma drug. “It’s like killing someone with a typo,” says MedAware CEO Gidi Stein.
Enter MedAware, a safety net with learning capacity, to catch errors. Its software can identify potential prescription mistakes and alert the doctor when the prescription is issued, rather than when the patient flops over.
Available alternative systems are rule-based, which, says Stein, is like looking under the lamp-post: you only find what you are specifically looking for, if it's there. Also, "they have so many false alarms that doctors get alert fatigue and start ignoring the warnings," he explains.
In short, they don't work well, and the problem is big. A Congress-commissioned report by the Institute of Medicine found that medication errors hurt or kill at least 1.5 million Americans each year.
“The most common medication errors include nurses administering the wrong medications or wrong dose in a intravenous drip, physicians prescribing drugs that could cause dangerous interactions with patients’ other medications and pharmacists dispensing 100-milligram tablets when 50-milligram tablets were prescribed," the report states.
Treating such error-related injuries sets America back $3.5 billion a year. MedAware however can flag potentially erroneous prescriptions in real-time, Stein says. How does it work?
Doctor writes nonsense, system blanches
Based on the electronic medical records it aggregates, MedAware can check whether a given prescription matches patterns among “similar physicians, treating similar patients, with similar scenarios,” Stein explains.
The doctor writes out a prescription on his computer system. The hospital computer system instantly sends it to the MedAware database, which validates the prescription, or issues an alert to the doctor that something is weird.
So: if 99% of doctors treat condition X with drug Y, and your doctor writes out a prescription for something else, or for an atypical dosage, the system would send an alert. The MedAlert software in essence crowdsources medical practice.
Having received an alert, the doctor can decide whether or not to dismiss it. “Then the software takes the response of the physician and feeds it back into the system, thus retaining a self-learning, self-adapting system,” Stein says.
And if the MedAware servers crash? The safety net vanishes until the system comes back, but that doesn't affect the hospital systems, Stein points out.
The product should be available in the U.S. in six to 12 months, he adds: MedAware is in the process of discussing collaboration with the Fortune 500 company Becton Dickenson.
Viagra for toddlers
The clients for the software are hospitals, HMOs, primary-care facilities, and so on, which buy the system for the doctors in the group. Stein has already sold a system to Israel's Tel Hashomer medical center and is in process of installing it in the Maccabi healthcare services organization.
Alerts for wrong prescriptions are just the start, Stein thinks. "We can take the model to new areas. Say I want to know if my mother is getting the right treatment in general. If I ask two doctors, I'll get three opinions. But it's critical to population health management."
The more data MedAware builds up, the bigger the picture it sees and better it can answer that question, and grade the quality of care that Stein's mother is getting. Or not getting. It can say that she's getting a drug that 99% of patients like her don't, or isn't getting one that 99% of patients do.
MedAware is still proving itself. The product is being evaluated in several medical facilities, including Harvard. Results should be available in a few months.
Meanwhile, the company transparently provides the outcome of testing in three medical facilities in Israel on its website.
The tests show how MedAware’s algorithm would function in real-time by analyzing all prescriptions that passed through the hospitals’ already-in-place systems. Information was analyzed for 23,092 inpatients, and 409,546 outpatients.
Upon being filtered through MedAware’s system, alerts were generated in 1 percent of outpatients and 3 percent of inpatients.
The software detected drug mix-ups, patient mix-ups, physician unawareness of clinical data, and outliers in monitored drugs. Of these four types of errors, physician unawareness of clinical data was the most common.
“We found patients without cancer being prescribed with chemotherapy, we found patients without diabetes prescribed with insulin, we found a two-year-old male child prescribed with Viagra,” Stein said. “I can go on, and on, and on forever.”
According to Stein, MedAware picked up on hundreds of life threatening mistakes.
When asked whether the product has already saved lives, he says, yes: “I know for sure.”