One Number a Week: Brian Fengler on Getting Data Back to the Doctor
Brian Fengler had already published a protocol on pulmonary embolism when a nurse at St. Thomas in Nashville screamed for him one night. The patient was 36 weeks pregnant, with a massive clot between her heart and lungs. “What’s the risk to my baby?” she asked. “I don’t know,” he told her. He gave her the clot-busting drug. Eight hours later her vital signs were normal, and he still gets a photo of her son every birthday.
He was the expert, in air quotes, and still at a loss. The EHR his hospital was rolling out was, he told us, “a tool for documentation and coding.” EvidenceCare, which he co-founded and where he is chief medical officer, was his attempt to put the evidence where the decision happens.
The price at the moment of the order
Utilization data usually reaches doctors months after the care. EvidenceCare’s gauge shows green, yellow, or red while the order is going in. Is there a Hawthorne effect? “Absolutely,” Brian said. What he hears from doctors is “Why didn’t anybody ever tell me this was a three hundred dollar medication?” when an option costing a dollar was available.
One text a week
Health systems track thousands of quality metrics, Brian said, and individual doctors rarely see their own. EvidenceCare texts each physician one metric a week, unblinded against their colleagues. By Brian’s account, performance varies four- to eightfold between physicians, 80% engage, and 70% of the metrics they track improve with statistical significance. One doctor told them, “I just wanna be in the middle on every metric.”
Leon asked the Goodhart question: what stops people from gaming the number? Brian’s answer is a countermeasure for each metric. Push doctors to write discharge orders early, and a second measure watches for patients held over to the next day. Across their clients, he said, those countermeasures haven’t gone up. Most of what doctors are shown, he said, is “punitive,” badly attributed, and “weaponized against them.”
Who pays for quality
The company started with clinical pathways licensed from the emergency medicine societies. Physicians loved them; CFOs wouldn’t sign. Around 2019 a health system CEO cut a pitch short: “I’m losing two hundred million dollars a year off of bed status decisions.” He didn’t want the pathways. He wanted that problem solved, and it became the company’s admission tool.
Where the AI is
EvidenceCare built its products before the AI boom. “Healthcare doesn’t need more software,” Brian said. “Healthcare needs better solutions.” AI now pre-fills the admission checklist from data already in the record, because doctors want fewer clicks. AI and automation have also cut some data processing from three days to a few hours.
How do you know it was you?
A score people can see becomes a score people manage, so an improved number has to be explained twice: did the care get better, and was it the tool? Steve asked the second question. “I wish we had a magic bullet,” Brian said. “It is extremely difficult.” EvidenceCare compares actual results with an expected baseline that is recalculated as case mix and payer mix change, then compares doctors who use the tool with those who don’t. That narrows the question. It doesn’t close it, and he didn’t claim it does.
The speed of trust
Health systems can’t agree on AI itself. Some won’t talk to a vendor with any AI in its product; others won’t contract with one that lacks it. Steve also asked about data rights. Standard agreements don’t let a vendor build new products from client data, Brian said, and widening them takes executive support, diligence, and security reviews. Leon summed it up with a line from an old partner: “Data moves at the speed of trust,” and sometimes at the speed of lawyers.
Listen to the full conversation: https://practicalaiinhealthcare.com/episodes/#fengler
You Might Also Enjoy
- S1E33 with Edward H. “Ted” Shortliffe — The creator of MYCIN on fifty years of putting decision support in front of clinicians.
- S2E04 with Travis Zack, MD, PhD — Another physician executive on getting the evidence to the point of care.
- S1E43 with Isaac “Zak” Kohane — Who controls healthcare data, and what it takes to share it.