I am a physician and I am not satisfied with my electronic medical record system. This is unfortunately not a unique or interesting statement: the majority of physicians agree that EHRs (1) take time away from clinical care, (2) contribute to burnout, and (3) increase total hours worked. But as with any complicated relationship, there is some love behind that hate.
As conceived by the drafters of the 2009 HITECH act, EHRs are actually doing reasonably well. They provide a digital repository for health information, and connect (mostly) to other EHR systems, sometimes with some nudging. EHRs provide online portals for patients to access their information – a level of access impossible a decade ago.
With that in mind, I would suggest to my fellow physicians: let’s stop putting our hopes of offering satisfying, high quality medicine in the hands of our EHR vendors. They are not evil, they are just there to solve a different set of problems. The things we care about are different: (1) spending more time delivering clinical care, (2) coordinating a plan for our most complex patients, and (3) having insightful interactions with our colleagues. These are not the goals of the HITECH act or the Office of the National Coordinator (ONC) — and we were also never the customer.
If we want software to help us with these goals. We need to ask for it, demand it, and specify how we want it to work. So, how would we design a piece of software that gives us time with patients, coordinates care, and allows us to have insightful interactions? Here are some possibilities:
A tool that isn’t scared to show clinicians information: Clinicians learn to scan reams of information the day they start their clinical training on the hospital wards:
“What is the story behind this new patient?”
“I don’t know. Why don’t you read their 100 page discharge summary. I’ll check with you in five minutes.”
Clinicians are skilled at scanning information-dense displays, pages, notes, etc… and rapidly pattern matching against the thousands of patients they’ve seen before. Our modern design heuristics for consumer products attempt to simplify for the sake of elegance, thereby obfuscating crucial details behind drop downs or expanders. Clinicians do not need an elegant tool, they need a functional one that serves them information in well designed ways, at the speed their minds can process it.
A tool that realizes no information is better than mis-timed information: If we want physicians to stop being cowboys and instead be part of a pit crew, they need the right information at the right moment in time. Receiving a notification our patient was in the ER three days ago doesn’t help. It needs to be in real time. Will the real-time notification help? I think so, but people should study it to confirm. What we do know is getting the notification late only contributes to burnout from ‘click fatigue’.
A tool that invests in making a clinical recommendation only when it is certain it’s the right one: It’s possible that someday AI algorithms will obviate physicians from having to make complex decisions or exercise judgement. Until that day comes, physicians and algorithms will need to learn to work together. Systems that offer clinical insights to physicians currently strive to be mostly correct but when trust is fragile and the relationship between clinician and algorithm is tenuous, we need to aim higher. A single incorrect recommendation or inherent bias in the algorithm sets the relationship between man and machine back.
Patients continue to trust doctors more than hospitals, insurance companies, or pharma companies. This feat remains durable through a pandemic, the creation of a vaccine in record time, and the fact that hospitals became a central focus of the Covid-19 pandemic. No other actor in the healthcare system is better positioned to ask for what they need and demand the market to serve it. As physicians, let’s move the discussion beyond complaining about the EHRs we hate, and instead beginning to ask for the things we want to deliver great patient care.
