Electronic health records (EHRs) offered the promise of better healthcare and lower costs. They were seen as the antidote to dangerously illegible handwriting, misplaced charts, and tests repeated because results performed elsewhere couldn’t be found. Legislation and financial incentives in the early 2010s encouraged the conversion from paper to digital charts, and nearly all U.S. healthcare systems and many physician practices have now made the shift. But rather than bringing a transformative change, EHRs became a source of physician discontent and burnout. Most doctors hate EHRs, and it isn’t clear that they make patients any better off either.
That experience contrasts with what we’ve seen in other industries. When financial services went digital—largely replacing human tellers, paper statements, and checkbook balancing—both bankers and customers embraced the change. Digital transformations in the travel, entertainment, and retail settings were similarly welcomed. But not in healthcare.
The problem is that current EHRs merely digitized what was already in paper form. The few adaptations made to the new digital environment prioritized billing and administrative tasks over clinical ones. In the opposite of typical design processes that emphasize the user experience, doctors became the technicians working for the built system. The result was that both documentation and data retrieval became burdensome tasks.
Here’s the point that seems to have been missed: Once the clinical information that used to be in paper charts is digital, why go to the chart at all? The important information you need can come to you. At Penn Medicine, we’ve reimagined EHRs as dynamic streams instead of static charts so that the same doctors who bypass going to the newspaper and instead subscribe to feeds about their favorite sports teams can bypass going to the chart and instead subscribe to Ms. Jones in Room 328.
By designing alongside clinicians, we have learned some important lessons:
Relying on busy clinicians to make it to the chart in time to act invites mistakes
In one case, institutional policy required clinicians to renew orders for seizure medications and antibiotics at certain time intervals, requiring them to actively remember when and for whom renewals were due. Most of the time this occurred without issue, but in 10% of cases, medication renewals fell through the cracks and patients missed doses. We piloted a system that allowed clinicians to subscribe to notices about their patients’ medication expirations, and the number of missed doses was cut by one third. Likewise, intravenous nutrition doses—critical for patients who can’t eat normally—required clinicians to preorder them before a 3 p.m. administrative deadline each day. This was an extra task that could be missed by busy doctors. While rare, that could mean patients losing a day’s nutrition or, more likely, frantic last-minute arrangements. Allowing inpatient teams to subscribe to text reminders–a “last call” for intravenous nutrition–relieved doctors of one more checklist item on their minds.
There’s just too much data to sift through, and too many competing tasks
In the intensive care unit, patients on mechanical ventilators (breathing machines) were examined each day to determine if they could be weaned from the ventilator and breathe without assistance. But these evaluations happened only once to twice a day, when it was convenient for providers, in part because the assessment involved gathering information from multiple parts of the patient record. Taking advantage of the digital information already in the electronic records allowed us to automate those evaluations. Combining that automation with a subscription service alerting doctors and respiratory therapists meant that patients were weaned from the ventilator an average of a half day earlier, significantly speeding their recovery and decreasing ICU costs.
Not everyone needs to know everything
Traditional charts are like huge communal troughs of information. Doctors in their personal lives subscribe to information that’s important to them (maybe “yes” to headline news, “later” for sports, and “no thanks” for the automotive section). Conventional record systems don’t allow such filtering. Our nephrology consultation service followed patients in the hospital with kidney disease, but often became aware of discharges only after they happened. When the covering nephrologists began subscribing to patients receiving inpatient dialysis, they learned immediately about discharges—without having to wade through full records—and received summaries of care they could use for a safer handoff to the outpatient dialysis center.
Inpatients and outpatients are all patients—no matter where they are
Clinicians see patients (and therefore think about them) either in an outpatient office or in a hospital bed. But they are responsible for patients even when they’re not right in front of them. We extended the subscription concept to develop a dashboard permitting a kind of SWAT team of clinicians to track the highest patient users of care–so called “superutilizers.” They collected details about how to communicate with the patient, engage with the family, and coordinate appropriate social services, and when these patients hit the emergency room, the team was alerted by the dashboard. That kind of information in the moment of intervention helped them take a comprehensive, not episodic, approach to manage this high-risk population across visits. Repeat hospital admissions and total hospital days were reduced by 67% and 56%, respectively.
None of these opportunities was possible until patient records became electronic, but these opportunities have been missed in the first efforts of EHR creation. Early movies began as plays captured on film, but eventually movies became something richer, taking advantage of special effects and on location settings that weren’t available on stage. Current EHRs haven’t made that transition.
There is so much more to do. The cases we describe are examples of improving the retrieval of information, making it available and useful to busy clinicians. But so many doctors’ complaints also involve the hassles of entering information: the multiple clicks and clumsy interfaces. The input for EHRs has expanded rather than reduced work. And EHRs’ remote accessibility—originally hailed as a major benefit—also brought that work into physicians’ home lives. Indeed, electronic logs reveal activity spikes between 9 p.m. and midnight when physicians spend what used to be “pajama time” charting instead of enjoying their families. It’s no wonder physician burnout has become a national epidemic.
What will it take for EHRs to realize more potential from their digital medium? It will take the kind of co-creation among clinicians and developers we expect from contemporary design in other industries. Our team consists of hybrid doctor-developers and relies on close partnership with clinicians of various specialties who come to us with their needs and tell us when we’ve failed them. That’s the only way we’ll learn what works. It’s not how current EHRs were developed, but now we have a chance to redesign them.
Katherine Choi, MD; David Do, MD; Yevgeniy Gitelman, MD;and David A. Asch, MD, are from the Center for Health Care Innovation, at the University of Pennsylvania. A related article appeared May 24, 2018 in The New England Journal of Medicine.