Nearly all U.S. health care systems and many physician practices have by now migrated from paper charts to electronic health records (EHRs). But though this shift could have been a transformative change, current EHRs are largely digital remakes of traditional systems, just as many early motion pictures were merely plays captured on celluloid. In time, movies began using on-location settings and special effects to make the two-dimensional screen deeper than the three-dimensional stage.
As compared with other digital transformations that have redefined the way we consume information, the effect of EHRs on clinicians’ engagement seems limited and effortful. Physicians in the hospital can keep up with feeds on the Philadelphia Eagles, Taylor Swift, and the price of Bitcoin without consulting a newspaper. Yet they must still go to the chart to check on their patients. What would it be like to instead subscribe to Ms. Jones in room 328?
For one thing, receipt of important information on patients’ conditions would depend less on physicians’ remembering to search the chart. For stewardship of antibiotics and antiepileptics for inpatients, for example, Penn Medicine had established automatic medication expiration, but the system required that residents remember when renewals were due. Necessary medications were not reordered in 10% of cases because a physician didn’t check the chart in time or didn’t notice the need for renewal. So we developed a Web application to pull real-time information from our health system’s multiple digital sources and allow it to be reassembled into customizable dashboards, mobile displays, and push notifications. The result was a platform that can tailor streams of data for particular clinical scenarios — and measure the impact.
In an early pilot, residents who opted in were subscribed to push notices about their patients’ medication expirations. Residents did not need to visit the chart to learn of an expiration, and the percentage of doses of antibiotics and antiepileptics that were missed was cut by a third. Digitizing the chart made clinical data more legible and accessible remotely, but the more transformative change was eliminating the need to be in the chart to know that a task had been overlooked in the first place.
Similarly, inpatient teams subscribed to text reminders for their patients who needed total parenteral nutrition reordered before the 3 p.m. administrative deadline. Subscribing to this “last call” kept residents on top of this small but important task, for which some had previously created phone alarms. It was one more checklist item that providers were relieved to have off their minds. So many health care processes are built around passive engagement with a medical record, relying on the hope that physicians will get to the chart on time, see what they need to do, and do it.
A second change is that subscription services can shorten the lag time between when information becomes available and when it’s used. Our old approach evaluated mechanically ventilated patients to see whether they could breathe without assistance when ICU rounds were over, and patients’ readiness was evaluated when it was convenient for providers. Digital information has both enabled patients to be automatically evaluated according to specified readiness criteria and allowed clinicians to receive prompts to act when the criteria are met. Freeing these evaluations from dependence on the whims and routines of the day reduced delays so that patients spent, on average, a half day less on a ventilator.
Not all information needs to be learned as soon as it’s available, and not all clinicians need to learn all information.1 A third change is that subscription services allow filtering to emphasize what’s important and relevant, just as we may select newspaper feeds of only the content of interest (headlines, yes; automobiles, no), mode of delivery (let me pull posts from columnists when I have time, but push the sports news), and alert triggers (I don’t care about cat food unless it costs less than 52 cents a can). It would be rare for a clinician to elect to set every patient feed individually, but we can imagine default settings that follow rules for different circumstances and roles.
For example, our renal consult service used to rely on manual updates of impending discharges from inpatient teams but was often left out of the loop. When the covering nephrology fellows began subscribing to the patients receiving inpatient dialysis, however, they learned immediately about discharges and received summaries of care details to relay to the outpatient dialysis center; this system relieved them of the burdens and errors of daily chart scans. A program for monitoring patients discharged on intravenous (IV) antibiotics used to rely on a similar process to identify patients who needed long-term follow-up. When the infectious disease team began subscribing to all patients slated for discharge on IV antibiotics, handoffs happened within hours rather than days after discharge, and in some cases plans for IV treatment were redirected to oral treatment. For these teams, the ability to define first which patients they wanted to follow and then what they wanted to know about them replaced more laborious and error-prone practices.
Alerts themselves are not new, and systems must be carefully designed to reduce the risk of alert fatigue, which can become self-defeating.2 The successes of these examples may reflect their focus on salient data delivered to clinicians when the information is most actionable. What is new is the ability to separate the documentation of information from its communication, changing what used to be a communal trough of patient information into channels responsive to various people’s needs to know particular information at specific times.
Indeed, beyond one-off intervention alerts, a fourth change is that subscription services permit management of panels of patients. The awareness provided by pushed alerts in inpatient settings is also relevant for patients in any setting. One of our programs monitored the 30 highest patient users of care in one of our hospitals, using a dashboard to follow their needs. Key details such as the best ways to communicate with the patient and engage with the family and next steps for setting up social services were kept on the dashboard. The covering cross-disciplinary team was automatically alerted to these patients’ arrival in the emergency department and were pointed to the previously prepared action plan, which guided them in real time. After 1 year of use of this system, 30-day readmissions and total hospital days had decreased by 67% and 56%, respectively. Subscription services help erode the distinction in follow-up care between inpatients and outpatients, focusing on what patients need rather than on where they are.
Success in creating such a subscription service requires a blend of programming skills and clinical sensibilities, foundational platforms and application programming interfaces providing access to real-time data, and a leadership commitment to providing the time and license to test clinicians’ imagination of how things might be different. In our experience, many organizations lack some of those ingredients: they may fear introducing theoretical points of failure or needing to adapt security measures to new channels, but often they simply see EHRs as digital remakes of what used to be. It’s no wonder physicians find contemporary EHRs burdensome rather than liberating.3
Treating clinical data as static files to be retrieved misses opportunities to relieve physicians of outdated, unnecessary burdens. It doesn’t take that much “design thinking” to anticipate searches or to target information so it’s useful and not just passively available, but most electronic systems and most health care organizations haven’t made this shift. Whereas movies can now enrich stories using technology unavailable on the stage, we still await the main act of EHRs’ digital evolution.
From the Center for Health Care Innovation, University of Pennsylvania (K.C., Y.G., D.A.A.); and the Cpl. Michael J. Crescenz VA Medical Center (D.A.A.) — both in Philadelphia.
1. Asch DA, Terwiesch C, Volpp KG. How to reduce primary care doctors’ workloads while improving care. Harvard Business Review. November 13, 2017 (https://hbr.org/2017/11/how-to-reduce-primary-care-doctors-workloads-while-improving-care).
2. Kesselheim AS, Cresswell K, Phansalkar S, Bates DW, Sheikh A. Clinical decision support systems could be modified to reduce ‘alert fatigue’ while still minimizing the risk of litigation. Health Aff (Millwood) 2011;30:2310-2317. CrossRef | Medline
3. Verghese A, Shah NH, Harrington RA. What this computer needs is a physician: humanism and artificial intelligence. JAMA 2018;319:19-20. CrossRef | Medline
This Perspective article originally appeared in The New England Journal of Medicine.